v1.1.8

* fix raw.arrow missing rows

---------

Co-authored-by: SWivid <swivid@qq.com>

.github/workflows/sync-hf.yaml

@@ -1,18 +0,0 @@
-name: Sync to HF Space
-
-on:
-  push:
-    branches:
-      - main
-
-jobs:
-  trigger_curl:
-    runs-on: ubuntu-latest
-
-    steps:
-    - name: Send cURL POST request
-      run: |
-        curl -X POST https://mrfakename-sync-f5.hf.space/gradio_api/call/refresh \
-        -s \
-        -H "Content-Type: application/json" \
-        -d "{\"data\": [\"${{ secrets.REFRESH_PASSWORD }}\"]}"

.pre-commit-config.yaml

@@ -3,11 +3,14 @@ repos:
     # Ruff version.
     rev: v0.11.2
     hooks:
-      # Run the linter.
       - id: ruff
+        name: ruff linter
         args: [--fix]
-      # Run the formatter.
       - id: ruff-format
+        name: ruff formatter
+      - id: ruff
+        name: ruff sorter
+        args: [--select, I, --fix]
   - repo: https://github.com/pre-commit/pre-commit-hooks
     rev: v5.0.0
     hooks:

README.md

@@ -91,7 +91,7 @@ conda activate f5-tts
 > ```bash
 > git clone https://github.com/SWivid/F5-TTS.git
 > cd F5-TTS
-> # git submodule update --init --recursive  # (optional, if need > bigvgan)
+> # git submodule update --init --recursive  # (optional, if use bigvgan as vocoder)
 > pip install -e .
 > ```
 
@@ -107,9 +107,27 @@ docker container run --rm -it --gpus=all --mount 'type=volume,source=f5-tts,targ
 docker container run --rm -it --gpus=all --mount 'type=volume,source=f5-tts,target=/root/.cache/huggingface/hub/' -p 7860:7860 ghcr.io/swivid/f5-tts:main f5-tts_infer-gradio --host 0.0.0.0
 ```
 
+### Runtime
+
+Deployment solution with Triton and TensorRT-LLM.
+
+#### Benchmark Results
+Decoding on a single L20 GPU, using 26 different prompt_audio & target_text pairs, 16 NFE.
+
+| Model               | Concurrency    | Avg Latency | RTF    | Mode            |
+|---------------------|----------------|-------------|--------|-----------------|
+| F5-TTS Base (Vocos) | 2              | 253 ms      | 0.0394 | Client-Server   |
+| F5-TTS Base (Vocos) | 1 (Batch_size) | -           | 0.0402 | Offline TRT-LLM |
+| F5-TTS Base (Vocos) | 1 (Batch_size) | -           | 0.1467 | Offline Pytorch |
+
+See [detailed instructions](src/f5_tts/runtime/triton_trtllm/README.md) for more information.
+
 
 ## Inference
 
+- In order to achieve desired performance, take a moment to read [detailed guidance](src/f5_tts/infer).
+- By properly searching the keywords of problem encountered, [issues](https://github.com/SWivid/F5-TTS/issues?q=is%3Aissue) are very helpful.
+
 ### 1. Gradio App
 
 Currently supported features:
@@ -176,11 +194,6 @@ f5-tts_infer-cli -c custom.toml
 f5-tts_infer-cli -c src/f5_tts/infer/examples/multi/story.toml
 ```
 
-### 3. More instructions
-
-- In order to have better generation results, take a moment to read [detailed guidance](src/f5_tts/infer).
-- The [Issues](https://github.com/SWivid/F5-TTS/issues?q=is%3Aissue) are very useful, please try to find the solution by properly searching the keywords of problem encountered. If no answer found, then feel free to open an issue.
-
 
 ## Training
 
@@ -231,6 +244,7 @@ Note: Some model components have linting exceptions for E722 to accommodate tens
 - [mrfakename](https://x.com/realmrfakename) huggingface space demo ~
 - [f5-tts-mlx](https://github.com/lucasnewman/f5-tts-mlx/tree/main) Implementation with MLX framework by [Lucas Newman](https://github.com/lucasnewman)
 - [F5-TTS-ONNX](https://github.com/DakeQQ/F5-TTS-ONNX) ONNX Runtime version by [DakeQQ](https://github.com/DakeQQ)
+- [Yuekai Zhang](https://github.com/yuekaizhang) Triton and TensorRT-LLM support ~
 
 ## Citation
 If our work and codebase is useful for you, please cite as:

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "f5-tts"
-version = "1.0.8"
+version = "1.1.8"
 description = "F5-TTS: A Fairytaler that Fakes Fluent and Faithful Speech with Flow Matching"
 readme = "README.md"
 license = {text = "MIT License"}
@@ -20,12 +20,13 @@ dependencies = [
     "click",
     "datasets",
     "ema_pytorch>=0.5.2",
-    "gradio>=3.45.2",
+    "gradio>=5.0.0",
     "hydra-core>=1.3.0",
     "jieba",
     "librosa",
     "matplotlib",
     "numpy<=1.26.4",
+    "pydantic<=2.10.6",
     "pydub",
     "pypinyin",
     "safetensors",
@@ -37,6 +38,7 @@ dependencies = [
     "tqdm>=4.65.0",
     "transformers",
     "transformers_stream_generator",
+    "unidecode",
     "vocos",
     "wandb",
     "x_transformers>=1.31.14",

ruff.toml

@@ -6,5 +6,5 @@ target-version = "py310"
 dummy-variable-rgx = "^_.*$"
 
 [lint.isort]
-force-single-line = true
+force-single-line = false
 lines-after-imports = 2

src/f5_tts/api.py

@@ -9,13 +9,13 @@ from hydra.utils import get_class
 from omegaconf import OmegaConf
 
 from f5_tts.infer.utils_infer import (
+    infer_process,
     load_model,
     load_vocoder,
-    transcribe,
     preprocess_ref_audio_text,
-    infer_process,
     remove_silence_for_generated_wav,
     save_spectrogram,
+    transcribe,
 )
 from f5_tts.model.utils import seed_everything
 

src/f5_tts/configs/F5TTS_Base.yaml

@@ -31,6 +31,8 @@ model:
     text_mask_padding: False
     conv_layers: 4
     pe_attn_head: 1
+    attn_backend: torch  # torch | flash_attn
+    attn_mask_enabled: False
     checkpoint_activations: False  # recompute activations and save memory for extra compute
   mel_spec:
     target_sample_rate: 24000

src/f5_tts/configs/F5TTS_Small.yaml

@@ -31,6 +31,8 @@ model:
     text_mask_padding: False
     conv_layers: 4
     pe_attn_head: 1
+    attn_backend: torch  # torch | flash_attn
+    attn_mask_enabled: False
     checkpoint_activations: False  # recompute activations and save memory for extra compute
   mel_spec:
     target_sample_rate: 24000

src/f5_tts/configs/F5TTS_v1_Base.yaml

@@ -32,6 +32,8 @@ model:
     qk_norm: null  # null | rms_norm
     conv_layers: 4
     pe_attn_head: null
+    attn_backend: torch  # torch | flash_attn
+    attn_mask_enabled: False
     checkpoint_activations: False  # recompute activations and save memory for extra compute
   mel_spec:
     target_sample_rate: 24000

src/f5_tts/eval/ecapa_tdnn.py

@@ -4,6 +4,7 @@
 # part of the code is borrowed from https://github.com/lawlict/ECAPA-TDNN
 
 import os
+
 import torch
 import torch.nn as nn
 import torch.nn.functional as F

src/f5_tts/eval/eval_infer_batch.py

@@ -1,6 +1,7 @@
 import os
 import sys
 
+
 sys.path.append(os.getcwd())
 
 import argparse
@@ -23,6 +24,7 @@ from f5_tts.infer.utils_infer import load_checkpoint, load_vocoder
 from f5_tts.model import CFM
 from f5_tts.model.utils import get_tokenizer
 
+
 accelerator = Accelerator()
 device = f"cuda:{accelerator.process_index}"
 
@@ -146,10 +148,15 @@ def main():
         vocab_char_map=vocab_char_map,
     ).to(device)
 
-    ckpt_path = rel_path + f"/ckpts/{exp_name}/model_{ckpt_step}.pt"
-    if not os.path.exists(ckpt_path):
+    ckpt_prefix = rel_path + f"/ckpts/{exp_name}/model_{ckpt_step}"
+    if os.path.exists(ckpt_prefix + ".pt"):
+        ckpt_path = ckpt_prefix + ".pt"
+    elif os.path.exists(ckpt_prefix + ".safetensors"):
+        ckpt_path = ckpt_prefix + ".safetensors"
+    else:
         print("Loading from self-organized training checkpoints rather than released pretrained.")
         ckpt_path = rel_path + f"/{model_cfg.ckpts.save_dir}/model_{ckpt_step}.pt"
+
     dtype = torch.float32 if mel_spec_type == "bigvgan" else None
     model = load_checkpoint(model, ckpt_path, device, dtype=dtype, use_ema=use_ema)
 

src/f5_tts/eval/eval_librispeech_test_clean.py

@@ -5,17 +5,16 @@ import json
 import os
 import sys
 
+
 sys.path.append(os.getcwd())
 
 import multiprocessing as mp
 from importlib.resources import files
 
 import numpy as np
-from f5_tts.eval.utils_eval import (
-    get_librispeech_test,
-    run_asr_wer,
-    run_sim,
-)
+
+from f5_tts.eval.utils_eval import get_librispeech_test, run_asr_wer, run_sim
+
 
 rel_path = str(files("f5_tts").joinpath("../../"))
 

src/f5_tts/eval/eval_seedtts_testset.py

@@ -5,17 +5,16 @@ import json
 import os
 import sys
 
+
 sys.path.append(os.getcwd())
 
 import multiprocessing as mp
 from importlib.resources import files
 
 import numpy as np
-from f5_tts.eval.utils_eval import (
-    get_seed_tts_test,
-    run_asr_wer,
-    run_sim,
-)
+
+from f5_tts.eval.utils_eval import get_seed_tts_test, run_asr_wer, run_sim
+
 
 rel_path = str(files("f5_tts").joinpath("../../"))
 

src/f5_tts/eval/utils_eval.py

@@ -126,8 +126,13 @@ def get_inference_prompt(
         else:
             text_list = text
 
+        # to mel spectrogram
+        ref_mel = mel_spectrogram(ref_audio)
+        ref_mel = ref_mel.squeeze(0)
+
         # Duration, mel frame length
-        ref_mel_len = ref_audio.shape[-1] // hop_length
+        ref_mel_len = ref_mel.shape[-1]
+
         if use_truth_duration:
             gt_audio, gt_sr = torchaudio.load(gt_wav)
             if gt_sr != target_sample_rate:
@@ -142,10 +147,6 @@ def get_inference_prompt(
             gen_text_len = len(gt_text.encode("utf-8"))
             total_mel_len = ref_mel_len + int(ref_mel_len / ref_text_len * gen_text_len / speed)
 
-        # to mel spectrogram
-        ref_mel = mel_spectrogram(ref_audio)
-        ref_mel = ref_mel.squeeze(0)
-
         # deal with batch
         assert infer_batch_size > 0, "infer_batch_size should be greater than 0."
         assert min_tokens <= total_mel_len <= max_tokens, (

src/f5_tts/infer/README.md

@@ -13,7 +13,7 @@ To avoid possible inference failures, make sure you have seen through the follow
 - Add some spaces (blank: " ") or punctuations (e.g. "," ".") <ins>to explicitly introduce some pauses</ins>.
 - If English punctuation marks the end of a sentence, make sure there is a space " " after it. Otherwise not regarded as when chunk.
 - <ins>Preprocess numbers</ins> to Chinese letters if you want to have them read in Chinese, otherwise in English.
-- If the generation output is blank (pure silence), <ins>check for ffmpeg installation</ins>.
+- If the generation output is blank (pure silence), <ins>check for FFmpeg installation</ins>.
 - Try <ins>turn off `use_ema` if using an early-stage</ins> finetuned checkpoint (which goes just few updates).
 
 
@@ -24,7 +24,7 @@ Currently supported features:
 - Basic TTS with Chunk Inference
 - Multi-Style / Multi-Speaker Generation
 - Voice Chat powered by Qwen2.5-3B-Instruct
-- [Custom inference with more language support](src/f5_tts/infer/SHARED.md)
+- [Custom inference with more language support](SHARED.md)
 
 The cli command `f5-tts_infer-gradio` equals to `python src/f5_tts/infer/infer_gradio.py`, which launches a Gradio APP (web interface) for inference.
 
@@ -129,6 +129,28 @@ ref_text = ""
 ```
 You should mark the voice with `[main]` `[town]` `[country]` whenever you want to change voice, refer to `src/f5_tts/infer/examples/multi/story.txt`.
 
+## API Usage
+
+```python
+from importlib.resources import files
+from f5_tts.api import F5TTS
+
+f5tts = F5TTS()
+wav, sr, spec = f5tts.infer(
+    ref_file=str(files("f5_tts").joinpath("infer/examples/basic/basic_ref_en.wav")),
+    ref_text="some call me nature, others call me mother nature.",
+    gen_text="""I don't really care what you call me. I've been a silent spectator, watching species evolve, empires rise and fall. But always remember, I am mighty and enduring. Respect me and I'll nurture you; ignore me and you shall face the consequences.""",
+    file_wave=str(files("f5_tts").joinpath("../../tests/api_out.wav")),
+    file_spec=str(files("f5_tts").joinpath("../../tests/api_out.png")),
+    seed=None,
+)
+```
+Check [api.py](../api.py) for more details.
+
+## TensorRT-LLM Deployment
+
+See [detailed instructions](../runtime/triton_trtllm/README.md) for more information.
+
 ## Socket Real-time Service
 
 Real-time voice output with chunk stream:

src/f5_tts/infer/SHARED.md

@@ -22,6 +22,8 @@
     - [F5-TTS Base @ fi @ AsmoKoskinen](#f5-tts-base--fi--asmokoskinen)
 - [French](#french)
     - [F5-TTS Base @ fr @ RASPIAUDIO](#f5-tts-base--fr--raspiaudio)
+- [German](#german)
+    - [F5-TTS Base @ de @ hvoss-techfak](#f5-tts-base--de--hvoss-techfak)
 - [Hindi](#hindi)
     - [F5-TTS Small @ hi @ SPRINGLab](#f5-tts-small--hi--springlab)
 - [Italian](#italian)
@@ -97,6 +99,22 @@ Config: {"dim": 1024, "depth": 22, "heads": 16, "ff_mult": 2, "text_dim": 512, "
 - [Discussion about this training can be found here](https://github.com/SWivid/F5-TTS/issues/434).
 
 
+## German
+
+#### F5-TTS Base @ de @ hvoss-techfak
+|Model|🤗Hugging Face|Data (Hours)|Model License|
+|:---:|:------------:|:-----------:|:-------------:|
+|F5-TTS Base|[ckpt & vocab](https://huggingface.co/hvoss-techfak/F5-TTS-German)|[Mozilla Common Voice 19.0](https://commonvoice.mozilla.org/en/datasets) & 800 hours Crowdsourced |cc-by-nc-4.0|
+
+```bash
+Model: hf://hvoss-techfak/F5-TTS-German/model_f5tts_german.pt
+Vocab: hf://hvoss-techfak/F5-TTS-German/vocab.txt
+Config: {"dim": 1024, "depth": 22, "heads": 16, "ff_mult": 2, "text_dim": 512, "text_mask_padding": False, "conv_layers": 4, "pe_attn_head": 1}
+```
+
+- Finetuned by [@hvoss-techfak](https://github.com/hvoss-techfak)
+
+
 ## Hindi
 
 #### F5-TTS Small @ hi @ SPRINGLab
@@ -137,11 +155,11 @@ Config: {"dim": 1024, "depth": 22, "heads": 16, "ff_mult": 2, "text_dim": 512, "
 #### F5-TTS Base @ ja @ Jmica
 |Model|🤗Hugging Face|Data (Hours)|Model License|
 |:---:|:------------:|:-----------:|:-------------:|
-|F5-TTS Base|[ckpt & vocab](https://huggingface.co/Jmica/F5TTS/tree/main/JA_25498980)|[Emilia 1.7k JA](https://huggingface.co/datasets/amphion/Emilia-Dataset/tree/fc71e07) & [Galgame Dataset 5.4k](https://huggingface.co/datasets/OOPPEENN/Galgame_Dataset)|cc-by-nc-4.0|
+|F5-TTS Base|[ckpt & vocab](https://huggingface.co/Jmica/F5TTS/tree/main/JA_21999120)|[Emilia 1.7k JA](https://huggingface.co/datasets/amphion/Emilia-Dataset/tree/fc71e07) & [Galgame Dataset 5.4k](https://huggingface.co/datasets/OOPPEENN/Galgame_Dataset)|cc-by-nc-4.0|
 
 ```bash
-Model: hf://Jmica/F5TTS/JA_25498980/model_25498980.pt
-Vocab: hf://Jmica/F5TTS/JA_25498980/vocab_updated.txt
+Model: hf://Jmica/F5TTS/JA_21999120/model_21999120.pt
+Vocab: hf://Jmica/F5TTS/JA_21999120/vocab_japanese.txt
 Config: {"dim": 1024, "depth": 22, "heads": 16, "ff_mult": 2, "text_dim": 512, "text_mask_padding": False, "conv_layers": 4, "pe_attn_head": 1}
 ```
 

src/f5_tts/infer/examples/multi/story.toml

@@ -13,8 +13,8 @@ output_file = "infer_cli_story.wav"
 [voices.town]
 ref_audio = "infer/examples/multi/town.flac"
 ref_text = ""
+speed = 0.8  # will ignore global speed
 
 [voices.country]
 ref_audio = "infer/examples/multi/country.flac"
 ref_text = ""
-

src/f5_tts/infer/examples/multi/story.txt

@@ -1 +1 @@
-A Town Mouse and a Country Mouse were acquaintances, and the Country Mouse one day invited his friend to come and see him at his home in the fields. The Town Mouse came, and they sat down to a dinner of barleycorns and roots, the latter of which had a distinctly earthy flavour. The fare was not much to the taste of the guest, and presently he broke out with [town] “My poor dear friend, you live here no better than the ants. Now, you should just see how I fare! My larder is a regular horn of plenty. You must come and stay with me, and I promise you you shall live on the fat of the land.” [main] So when he returned to town he took the Country Mouse with him, and showed him into a larder containing flour and oatmeal and figs and honey and dates. The Country Mouse had never seen anything like it, and sat down to enjoy the luxuries his friend provided: but before they had well begun, the door of the larder opened and someone came in. The two Mice scampered off and hid themselves in a narrow and exceedingly uncomfortable hole. Presently, when all was quiet, they ventured out again; but someone else came in, and off they scuttled again. This was too much for the visitor. [country] “Goodbye,” [main] said he, [country] “I’m off. You live in the lap of luxury, I can see, but you are surrounded by dangers; whereas at home I can enjoy my simple dinner of roots and corn in peace.”
+A Town Mouse and a Country Mouse were acquaintances, and the Country Mouse one day invited his friend to come and see him at his home in the fields. The Town Mouse came, and they sat down to a dinner of barleycorns and roots, the latter of which had a distinctly earthy flavour. The fare was not much to the taste of the guest, and presently he broke out with [town] "My poor dear friend, you live here no better than the ants! Now, you should just see how I fare! My larder is a regular horn of plenty. You must come and stay with me, and I promise you you shall live on the fat of the land." [main] So when he returned to town he took the Country Mouse with him, and showed him into a larder containing flour and oatmeal and figs and honey and dates. The Country Mouse had never seen anything like it, and sat down to enjoy the luxuries his friend provided: but before they had well begun, the door of the larder opened and someone came in. The two Mice scampered off and hid themselves in a narrow and exceedingly uncomfortable hole. Presently, when all was quiet, they ventured out again; but someone else came in, and off they scuttled again. This was too much for the visitor. [country] "Goodbye," [main] said he, [country] "I'm off. You live in the lap of luxury, I can see, but you are surrounded by dangers; whereas at home I can enjoy my simple dinner of roots and corn in peace."

src/f5_tts/infer/infer_cli.py

@@ -12,22 +12,23 @@ import tomli
 from cached_path import cached_path
 from hydra.utils import get_class
 from omegaconf import OmegaConf
+from unidecode import unidecode
 
 from f5_tts.infer.utils_infer import (
-    mel_spec_type,
-    target_rms,
-    cross_fade_duration,
-    nfe_step,
     cfg_strength,
-    sway_sampling_coef,
-    speed,
-    fix_duration,
+    cross_fade_duration,
     device,
+    fix_duration,
     infer_process,
     load_model,
     load_vocoder,
+    mel_spec_type,
+    nfe_step,
     preprocess_ref_audio_text,
     remove_silence_for_generated_wav,
+    speed,
+    sway_sampling_coef,
+    target_rms,
 )
 
 
@@ -112,6 +113,11 @@ parser.add_argument(
     action="store_true",
     help="To save each audio chunks during inference",
 )
+parser.add_argument(
+    "--no_legacy_text",
+    action="store_false",
+    help="Not to use lossy ASCII transliterations of unicode text in saved file names.",
+)
 parser.add_argument(
     "--remove_silence",
     action="store_true",
@@ -197,6 +203,12 @@ output_file = args.output_file or config.get(
 )
 
 save_chunk = args.save_chunk or config.get("save_chunk", False)
+use_legacy_text = args.no_legacy_text or config.get("no_legacy_text", False)  # no_legacy_text is a store_false arg
+if save_chunk and use_legacy_text:
+    print(
+        "\nWarning to --save_chunk: lossy ASCII transliterations of unicode text for legacy (.wav) file names, --no_legacy_text to disable.\n"
+    )
+
 remove_silence = args.remove_silence or config.get("remove_silence", False)
 load_vocoder_from_local = args.load_vocoder_from_local or config.get("load_vocoder_from_local", False)
 
@@ -321,9 +333,10 @@ def main():
         text = re.sub(reg2, "", text)
         ref_audio_ = voices[voice]["ref_audio"]
         ref_text_ = voices[voice]["ref_text"]
+        local_speed = voices[voice].get("speed", speed)
         gen_text_ = text.strip()
         print(f"Voice: {voice}")
-        audio_segment, final_sample_rate, spectragram = infer_process(
+        audio_segment, final_sample_rate, spectrogram = infer_process(
             ref_audio_,
             ref_text_,
             gen_text_,
@@ -335,7 +348,7 @@ def main():
             nfe_step=nfe_step,
             cfg_strength=cfg_strength,
             sway_sampling_coef=sway_sampling_coef,
-            speed=speed,
+            speed=local_speed,
             fix_duration=fix_duration,
             device=device,
         )
@@ -344,6 +357,8 @@ def main():
         if save_chunk:
             if len(gen_text_) > 200:
                 gen_text_ = gen_text_[:200] + " ... "
+            if use_legacy_text:
+                gen_text_ = unidecode(gen_text_)
             sf.write(
                 os.path.join(output_chunk_dir, f"{len(generated_audio_segments) - 1}_{gen_text_}.wav"),
                 audio_segment,

src/f5_tts/infer/infer_gradio.py

@@ -1,20 +1,25 @@
 # ruff: noqa: E402
 # Above allows ruff to ignore E402: module level import not at top of file
 
+import gc
 import json
+import os
 import re
 import tempfile
 from collections import OrderedDict
+from functools import lru_cache
 from importlib.resources import files
 
 import click
 import gradio as gr
 import numpy as np
 import soundfile as sf
+import torch
 import torchaudio
 from cached_path import cached_path
 from transformers import AutoModelForCausalLM, AutoTokenizer
 
+
 try:
     import spaces
 
@@ -30,15 +35,16 @@ def gpu_decorator(func):
         return func
 
 
-from f5_tts.model import DiT, UNetT
 from f5_tts.infer.utils_infer import (
-    load_vocoder,
-    load_model,
-    preprocess_ref_audio_text,
     infer_process,
+    load_model,
+    load_vocoder,
+    preprocess_ref_audio_text,
     remove_silence_for_generated_wav,
     save_spectrogram,
+    tempfile_kwargs,
 )
+from f5_tts.model import DiT, UNetT
 
 
 DEFAULT_TTS_MODEL = "F5-TTS_v1"
@@ -76,6 +82,8 @@ def load_custom(ckpt_path: str, vocab_path="", model_cfg=None):
         vocab_path = str(cached_path(vocab_path))
     if model_cfg is None:
         model_cfg = json.loads(DEFAULT_TTS_MODEL_CFG[2])
+    elif isinstance(model_cfg, str):
+        model_cfg = json.loads(model_cfg)
     return load_model(DiT, model_cfg, ckpt_path, vocab_file=vocab_path)
 
 
@@ -88,7 +96,7 @@ chat_tokenizer_state = None
 
 
 @gpu_decorator
-def generate_response(messages, model, tokenizer):
+def chat_model_inference(messages, model, tokenizer):
     """Generate response using Qwen"""
     text = tokenizer.apply_chat_template(
         messages,
@@ -110,6 +118,17 @@ def generate_response(messages, model, tokenizer):
     return tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
 
 
+@gpu_decorator
+def load_text_from_file(file):
+    if file:
+        with open(file, "r", encoding="utf-8") as f:
+            text = f.read().strip()
+    else:
+        text = ""
+    return gr.update(value=text)
+
+
+@lru_cache(maxsize=1000)  # NOTE. need to ensure params of infer() hashable
 @gpu_decorator
 def infer(
     ref_audio_orig,
@@ -117,6 +136,7 @@ def infer(
     gen_text,
     model,
     remove_silence,
+    seed,
     cross_fade_duration=0.15,
     nfe_step=32,
     speed=1,
@@ -126,8 +146,15 @@ def infer(
         gr.Warning("Please provide reference audio.")
         return gr.update(), gr.update(), ref_text
 
+    # Set inference seed
+    if seed < 0 or seed > 2**31 - 1:
+        gr.Warning("Seed must in range 0 ~ 2147483647. Using random seed instead.")
+        seed = np.random.randint(0, 2**31 - 1)
+    torch.manual_seed(seed)
+    used_seed = seed
+
     if not gen_text.strip():
-        gr.Warning("Please enter text to generate.")
+        gr.Warning("Please enter text to generate or upload a text file.")
         return gr.update(), gr.update(), ref_text
 
     ref_audio, ref_text = preprocess_ref_audio_text(ref_audio_orig, ref_text, show_info=show_info)
@@ -140,7 +167,7 @@ def infer(
             show_info("Loading E2-TTS model...")
             E2TTS_ema_model = load_e2tts()
         ema_model = E2TTS_ema_model
-    elif isinstance(model, list) and model[0] == "Custom":
+    elif isinstance(model, tuple) and model[0] == "Custom":
         assert not USING_SPACES, "Only official checkpoints allowed in Spaces."
         global custom_ema_model, pre_custom_path
         if pre_custom_path != model[1]:
@@ -164,44 +191,59 @@ def infer(
 
     # Remove silence
     if remove_silence:
-        with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as f:
-            sf.write(f.name, final_wave, final_sample_rate)
+        with tempfile.NamedTemporaryFile(suffix=".wav", **tempfile_kwargs) as f:
+            temp_path = f.name
+        try:
+            sf.write(temp_path, final_wave, final_sample_rate)
             remove_silence_for_generated_wav(f.name)
             final_wave, _ = torchaudio.load(f.name)
+        finally:
+            os.unlink(temp_path)
         final_wave = final_wave.squeeze().cpu().numpy()
 
     # Save the spectrogram
-    with tempfile.NamedTemporaryFile(suffix=".png", delete=False) as tmp_spectrogram:
+    with tempfile.NamedTemporaryFile(suffix=".png", **tempfile_kwargs) as tmp_spectrogram:
         spectrogram_path = tmp_spectrogram.name
-        save_spectrogram(combined_spectrogram, spectrogram_path)
+    save_spectrogram(combined_spectrogram, spectrogram_path)
 
-    return (final_sample_rate, final_wave), spectrogram_path, ref_text
+    return (final_sample_rate, final_wave), spectrogram_path, ref_text, used_seed
 
 
-with gr.Blocks() as app_credits:
-    gr.Markdown("""
-# Credits
-
-* [mrfakename](https://github.com/fakerybakery) for the original [online demo](https://huggingface.co/spaces/mrfakename/E2-F5-TTS)
-* [RootingInLoad](https://github.com/RootingInLoad) for initial chunk generation and podcast app exploration
-* [jpgallegoar](https://github.com/jpgallegoar) for multiple speech-type generation & voice chat
-""")
 with gr.Blocks() as app_tts:
     gr.Markdown("# Batched TTS")
     ref_audio_input = gr.Audio(label="Reference Audio", type="filepath")
-    gen_text_input = gr.Textbox(label="Text to Generate", lines=10)
+    with gr.Row():
+        gen_text_input = gr.Textbox(
+            label="Text to Generate",
+            lines=10,
+            max_lines=40,
+            scale=4,
+        )
+        gen_text_file = gr.File(label="Load Text to Generate from File (.txt)", file_types=[".txt"], scale=1)
     generate_btn = gr.Button("Synthesize", variant="primary")
     with gr.Accordion("Advanced Settings", open=False):
-        ref_text_input = gr.Textbox(
-            label="Reference Text",
-            info="Leave blank to automatically transcribe the reference audio. If you enter text it will override automatic transcription.",
-            lines=2,
-        )
-        remove_silence = gr.Checkbox(
-            label="Remove Silences",
-            info="The model tends to produce silences, especially on longer audio. We can manually remove silences if needed. Note that this is an experimental feature and may produce strange results. This will also increase generation time.",
-            value=False,
-        )
+        with gr.Row():
+            ref_text_input = gr.Textbox(
+                label="Reference Text",
+                info="Leave blank to automatically transcribe the reference audio. If you enter text or upload a file, it will override automatic transcription.",
+                lines=2,
+                scale=4,
+            )
+            ref_text_file = gr.File(label="Load Reference Text from File (.txt)", file_types=[".txt"], scale=1)
+        with gr.Row():
+            randomize_seed = gr.Checkbox(
+                label="Randomize Seed",
+                info="Check to use a random seed for each generation. Uncheck to use the seed specified.",
+                value=True,
+                scale=3,
+            )
+            seed_input = gr.Number(show_label=False, value=0, precision=0, scale=1)
+            with gr.Column(scale=4):
+                remove_silence = gr.Checkbox(
+                    label="Remove Silences",
+                    info="If undesired long silence(s) produced, turn on to automatically detect and crop.",
+                    value=False,
+                )
         speed_slider = gr.Slider(
             label="Speed",
             minimum=0.3,
@@ -236,21 +278,45 @@ with gr.Blocks() as app_tts:
         ref_text_input,
         gen_text_input,
         remove_silence,
+        randomize_seed,
+        seed_input,
         cross_fade_duration_slider,
         nfe_slider,
         speed_slider,
     ):
-        audio_out, spectrogram_path, ref_text_out = infer(
+        if randomize_seed:
+            seed_input = np.random.randint(0, 2**31 - 1)
+
+        audio_out, spectrogram_path, ref_text_out, used_seed = infer(
             ref_audio_input,
             ref_text_input,
             gen_text_input,
             tts_model_choice,
             remove_silence,
+            seed=seed_input,
             cross_fade_duration=cross_fade_duration_slider,
             nfe_step=nfe_slider,
             speed=speed_slider,
         )
-        return audio_out, spectrogram_path, ref_text_out
+        return audio_out, spectrogram_path, ref_text_out, used_seed
+
+    gen_text_file.upload(
+        load_text_from_file,
+        inputs=[gen_text_file],
+        outputs=[gen_text_input],
+    )
+
+    ref_text_file.upload(
+        load_text_from_file,
+        inputs=[ref_text_file],
+        outputs=[ref_text_input],
+    )
+
+    ref_audio_input.clear(
+        lambda: [None, None],
+        None,
+        [ref_text_input, ref_text_file],
+    )
 
     generate_btn.click(
         basic_tts,
@@ -259,35 +325,46 @@ with gr.Blocks() as app_tts:
             ref_text_input,
             gen_text_input,
             remove_silence,
+            randomize_seed,
+            seed_input,
             cross_fade_duration_slider,
             nfe_slider,
             speed_slider,
         ],
-        outputs=[audio_output, spectrogram_output, ref_text_input],
+        outputs=[audio_output, spectrogram_output, ref_text_input, seed_input],
     )
 
 
 def parse_speechtypes_text(gen_text):
-    # Pattern to find {speechtype}
-    pattern = r"\{(.*?)\}"
+    # Pattern to find {str} or {"name": str, "seed": int, "speed": float}
+    pattern = r"(\{.*?\})"
 
     # Split the text by the pattern
     tokens = re.split(pattern, gen_text)
 
     segments = []
 
-    current_style = "Regular"
+    current_type_dict = {
+        "name": "Regular",
+        "seed": -1,
+        "speed": 1.0,
+    }
 
     for i in range(len(tokens)):
         if i % 2 == 0:
             # This is text
             text = tokens[i].strip()
             if text:
-                segments.append({"style": current_style, "text": text})
+                current_type_dict["text"] = text
+                segments.append(current_type_dict)
         else:
-            # This is style
-            style = tokens[i].strip()
-            current_style = style
+            # This is type
+            type_str = tokens[i].strip()
+            try:  # if type dict
+                current_type_dict = json.loads(type_str)
+            except json.decoder.JSONDecodeError:
+                type_str = type_str[1:-1]  # remove brace {}
+                current_type_dict = {"name": type_str, "seed": -1, "speed": 1.0}
 
     return segments
 
@@ -298,44 +375,55 @@ with gr.Blocks() as app_multistyle:
         """
     # Multiple Speech-Type Generation
 
-    This section allows you to generate multiple speech types or multiple people's voices. Enter your text in the format shown below, and the system will generate speech using the appropriate type. If unspecified, the model will use the regular speech type. The current speech type will be used until the next speech type is specified.
+    This section allows you to generate multiple speech types or multiple people's voices. Enter your text in the format shown below, or upload a .txt file with the same format. The system will generate speech using the appropriate type. If unspecified, the model will use the regular speech type. The current speech type will be used until the next speech type is specified.
     """
     )
 
     with gr.Row():
         gr.Markdown(
             """
-            **Example Input:**                                                                      
-            {Regular} Hello, I'd like to order a sandwich please.                                                         
-            {Surprised} What do you mean you're out of bread?                                                                      
-            {Sad} I really wanted a sandwich though...                                                              
-            {Angry} You know what, darn you and your little shop!                                                                       
-            {Whisper} I'll just go back home and cry now.                                                                           
-            {Shouting} Why me?!                                                                         
+            **Example Input:** <br>
+            {Regular} Hello, I'd like to order a sandwich please. <br>
+            {Surprised} What do you mean you're out of bread? <br>
+            {Sad} I really wanted a sandwich though... <br>
+            {Angry} You know what, darn you and your little shop! <br>
+            {Whisper} I'll just go back home and cry now. <br>
+            {Shouting} Why me?!
             """
         )
 
         gr.Markdown(
             """
-            **Example Input 2:**                                                                                
-            {Speaker1_Happy} Hello, I'd like to order a sandwich please.                                                            
-            {Speaker2_Regular} Sorry, we're out of bread.                                                                                
-            {Speaker1_Sad} I really wanted a sandwich though...                                                                             
-            {Speaker2_Whisper} I'll give you the last one I was hiding.                                                                     
+            **Example Input 2:** <br>
+            {"name": "Speaker1_Happy", "seed": -1, "speed": 1} Hello, I'd like to order a sandwich please. <br>
+            {"name": "Speaker2_Regular", "seed": -1, "speed": 1} Sorry, we're out of bread. <br>
+            {"name": "Speaker1_Sad", "seed": -1, "speed": 1} I really wanted a sandwich though... <br>
+            {"name": "Speaker2_Whisper", "seed": -1, "speed": 1} I'll give you the last one I was hiding.
             """
         )
 
     gr.Markdown(
-        "Upload different audio clips for each speech type. The first speech type is mandatory. You can add additional speech types by clicking the 'Add Speech Type' button."
+        'Upload different audio clips for each speech type. The first speech type is mandatory. You can add additional speech types by clicking the "Add Speech Type" button.'
     )
 
     # Regular speech type (mandatory)
-    with gr.Row() as regular_row:
-        with gr.Column():
+    with gr.Row(variant="compact") as regular_row:
+        with gr.Column(scale=1, min_width=160):
             regular_name = gr.Textbox(value="Regular", label="Speech Type Name")
             regular_insert = gr.Button("Insert Label", variant="secondary")
-        regular_audio = gr.Audio(label="Regular Reference Audio", type="filepath")
-        regular_ref_text = gr.Textbox(label="Reference Text (Regular)", lines=2)
+        with gr.Column(scale=3):
+            regular_audio = gr.Audio(label="Regular Reference Audio", type="filepath")
+        with gr.Column(scale=3):
+            regular_ref_text = gr.Textbox(label="Reference Text (Regular)", lines=4)
+            with gr.Row():
+                regular_seed_slider = gr.Slider(
+                    show_label=False, minimum=-1, maximum=999, value=-1, step=1, info="Seed, -1 for random"
+                )
+                regular_speed_slider = gr.Slider(
+                    show_label=False, minimum=0.3, maximum=2.0, value=1.0, step=0.1, info="Adjust the speed"
+                )
+        with gr.Column(scale=1, min_width=160):
+            regular_ref_text_file = gr.File(label="Load Reference Text from File (.txt)", file_types=[".txt"])
 
     # Regular speech type (max 100)
     max_speech_types = 100
@@ -343,25 +431,55 @@ with gr.Blocks() as app_multistyle:
     speech_type_names = [regular_name]
     speech_type_audios = [regular_audio]
     speech_type_ref_texts = [regular_ref_text]
+    speech_type_ref_text_files = [regular_ref_text_file]
+    speech_type_seeds = [regular_seed_slider]
+    speech_type_speeds = [regular_speed_slider]
     speech_type_delete_btns = [None]
     speech_type_insert_btns = [regular_insert]
 
     # Additional speech types (99 more)
     for i in range(max_speech_types - 1):
-        with gr.Row(visible=False) as row:
-            with gr.Column():
+        with gr.Row(variant="compact", visible=False) as row:
+            with gr.Column(scale=1, min_width=160):
                 name_input = gr.Textbox(label="Speech Type Name")
-                delete_btn = gr.Button("Delete Type", variant="secondary")
                 insert_btn = gr.Button("Insert Label", variant="secondary")
-            audio_input = gr.Audio(label="Reference Audio", type="filepath")
-            ref_text_input = gr.Textbox(label="Reference Text", lines=2)
+                delete_btn = gr.Button("Delete Type", variant="stop")
+            with gr.Column(scale=3):
+                audio_input = gr.Audio(label="Reference Audio", type="filepath")
+            with gr.Column(scale=3):
+                ref_text_input = gr.Textbox(label="Reference Text", lines=4)
+                with gr.Row():
+                    seed_input = gr.Slider(
+                        show_label=False, minimum=-1, maximum=999, value=-1, step=1, info="Seed. -1 for random"
+                    )
+                    speed_input = gr.Slider(
+                        show_label=False, minimum=0.3, maximum=2.0, value=1.0, step=0.1, info="Adjust the speed"
+                    )
+            with gr.Column(scale=1, min_width=160):
+                ref_text_file_input = gr.File(label="Load Reference Text from File (.txt)", file_types=[".txt"])
         speech_type_rows.append(row)
         speech_type_names.append(name_input)
         speech_type_audios.append(audio_input)
         speech_type_ref_texts.append(ref_text_input)
+        speech_type_ref_text_files.append(ref_text_file_input)
+        speech_type_seeds.append(seed_input)
+        speech_type_speeds.append(speed_input)
         speech_type_delete_btns.append(delete_btn)
         speech_type_insert_btns.append(insert_btn)
 
+    # Global logic for all speech types
+    for i in range(max_speech_types):
+        speech_type_audios[i].clear(
+            lambda: [None, None],
+            None,
+            [speech_type_ref_texts[i], speech_type_ref_text_files[i]],
+        )
+        speech_type_ref_text_files[i].upload(
+            load_text_from_file,
+            inputs=[speech_type_ref_text_files[i]],
+            outputs=[speech_type_ref_texts[i]],
+        )
+
     # Button to add speech type
     add_speech_type_btn = gr.Button("Add Speech Type")
 
@@ -383,27 +501,44 @@ with gr.Blocks() as app_multistyle:
 
     # Function to delete a speech type
     def delete_speech_type_fn():
-        return gr.update(visible=False), None, None, None
+        return gr.update(visible=False), None, None, None, None
 
-    # Update delete button clicks
+    # Update delete button clicks and ref text file changes
     for i in range(1, len(speech_type_delete_btns)):
         speech_type_delete_btns[i].click(
             delete_speech_type_fn,
-            outputs=[speech_type_rows[i], speech_type_names[i], speech_type_audios[i], speech_type_ref_texts[i]],
+            outputs=[
+                speech_type_rows[i],
+                speech_type_names[i],
+                speech_type_audios[i],
+                speech_type_ref_texts[i],
+                speech_type_ref_text_files[i],
+            ],
         )
 
     # Text input for the prompt
-    gen_text_input_multistyle = gr.Textbox(
-        label="Text to Generate",
-        lines=10,
-        placeholder="Enter the script with speaker names (or emotion types) at the start of each block, e.g.:\n\n{Regular} Hello, I'd like to order a sandwich please.\n{Surprised} What do you mean you're out of bread?\n{Sad} I really wanted a sandwich though...\n{Angry} You know what, darn you and your little shop!\n{Whisper} I'll just go back home and cry now.\n{Shouting} Why me?!",
-    )
+    with gr.Row():
+        gen_text_input_multistyle = gr.Textbox(
+            label="Text to Generate",
+            lines=10,
+            max_lines=40,
+            scale=4,
+            placeholder="Enter the script with speaker names (or emotion types) at the start of each block, e.g.:\n\n{Regular} Hello, I'd like to order a sandwich please.\n{Surprised} What do you mean you're out of bread?\n{Sad} I really wanted a sandwich though...\n{Angry} You know what, darn you and your little shop!\n{Whisper} I'll just go back home and cry now.\n{Shouting} Why me?!",
+        )
+        gen_text_file_multistyle = gr.File(label="Load Text to Generate from File (.txt)", file_types=[".txt"], scale=1)
 
     def make_insert_speech_type_fn(index):
-        def insert_speech_type_fn(current_text, speech_type_name):
+        def insert_speech_type_fn(current_text, speech_type_name, speech_type_seed, speech_type_speed):
             current_text = current_text or ""
-            speech_type_name = speech_type_name or "None"
-            updated_text = current_text + f"{{{speech_type_name}}} "
+            if not speech_type_name:
+                gr.Warning("Please enter speech type name before insert.")
+                return current_text
+            speech_type_dict = {
+                "name": speech_type_name,
+                "seed": speech_type_seed,
+                "speed": speech_type_speed,
+            }
+            updated_text = current_text + json.dumps(speech_type_dict) + " "
             return updated_text
 
         return insert_speech_type_fn
@@ -412,15 +547,24 @@ with gr.Blocks() as app_multistyle:
         insert_fn = make_insert_speech_type_fn(i)
         insert_btn.click(
             insert_fn,
-            inputs=[gen_text_input_multistyle, speech_type_names[i]],
+            inputs=[gen_text_input_multistyle, speech_type_names[i], speech_type_seeds[i], speech_type_speeds[i]],
             outputs=gen_text_input_multistyle,
         )
 
-    with gr.Accordion("Advanced Settings", open=False):
-        remove_silence_multistyle = gr.Checkbox(
-            label="Remove Silences",
-            value=True,
-        )
+    with gr.Accordion("Advanced Settings", open=True):
+        with gr.Row():
+            with gr.Column():
+                show_cherrypick_multistyle = gr.Checkbox(
+                    label="Show Cherry-pick Interface",
+                    info="Turn on to show interface, picking seeds from previous generations.",
+                    value=False,
+                )
+            with gr.Column():
+                remove_silence_multistyle = gr.Checkbox(
+                    label="Remove Silences",
+                    info="Turn on to automatically detect and crop long silences.",
+                    value=True,
+                )
 
     # Generate button
     generate_multistyle_btn = gr.Button("Generate Multi-Style Speech", variant="primary")
@@ -428,6 +572,30 @@ with gr.Blocks() as app_multistyle:
     # Output audio
     audio_output_multistyle = gr.Audio(label="Synthesized Audio")
 
+    # Used seed gallery
+    cherrypick_interface_multistyle = gr.Textbox(
+        label="Cherry-pick Interface",
+        lines=10,
+        max_lines=40,
+        show_copy_button=True,
+        interactive=False,
+        visible=False,
+    )
+
+    # Logic control to show/hide the cherrypick interface
+    show_cherrypick_multistyle.change(
+        lambda is_visible: gr.update(visible=is_visible),
+        show_cherrypick_multistyle,
+        cherrypick_interface_multistyle,
+    )
+
+    # Function to load text to generate from file
+    gen_text_file_multistyle.upload(
+        load_text_from_file,
+        inputs=[gen_text_file_multistyle],
+        outputs=[gen_text_input_multistyle],
+    )
+
     @gpu_decorator
     def generate_multistyle_speech(
         gen_text,
@@ -455,41 +623,60 @@ with gr.Blocks() as app_multistyle:
 
         # For each segment, generate speech
         generated_audio_segments = []
-        current_style = "Regular"
+        current_type_name = "Regular"
+        inference_meta_data = ""
 
         for segment in segments:
-            style = segment["style"]
+            name = segment["name"]
+            seed_input = segment["seed"]
+            speed = segment["speed"]
             text = segment["text"]
 
-            if style in speech_types:
-                current_style = style
+            if name in speech_types:
+                current_type_name = name
             else:
-                gr.Warning(f"Type {style} is not available, will use Regular as default.")
-                current_style = "Regular"
+                gr.Warning(f"Type {name} is not available, will use Regular as default.")
+                current_type_name = "Regular"
 
             try:
-                ref_audio = speech_types[current_style]["audio"]
+                ref_audio = speech_types[current_type_name]["audio"]
             except KeyError:
-                gr.Warning(f"Please provide reference audio for type {current_style}.")
-                return [None] + [speech_types[style]["ref_text"] for style in speech_types]
-            ref_text = speech_types[current_style].get("ref_text", "")
+                gr.Warning(f"Please provide reference audio for type {current_type_name}.")
+                return [None] + [speech_types[name]["ref_text"] for name in speech_types] + [None]
+            ref_text = speech_types[current_type_name].get("ref_text", "")
 
-            # Generate speech for this segment
-            audio_out, _, ref_text_out = infer(
-                ref_audio, ref_text, text, tts_model_choice, remove_silence, 0, show_info=print
-            )  # show_info=print no pull to top when generating
+            if seed_input == -1:
+                seed_input = np.random.randint(0, 2**31 - 1)
+
+            # Generate or retrieve speech for this segment
+            audio_out, _, ref_text_out, used_seed = infer(
+                ref_audio,
+                ref_text,
+                text,
+                tts_model_choice,
+                remove_silence,
+                seed=seed_input,
+                cross_fade_duration=0,
+                speed=speed,
+                show_info=print,  # no pull to top when generating
+            )
             sr, audio_data = audio_out
 
             generated_audio_segments.append(audio_data)
-            speech_types[current_style]["ref_text"] = ref_text_out
+            speech_types[current_type_name]["ref_text"] = ref_text_out
+            inference_meta_data += json.dumps(dict(name=name, seed=used_seed, speed=speed)) + f" {text}\n"
 
         # Concatenate all audio segments
         if generated_audio_segments:
             final_audio_data = np.concatenate(generated_audio_segments)
-            return [(sr, final_audio_data)] + [speech_types[style]["ref_text"] for style in speech_types]
+            return (
+                [(sr, final_audio_data)]
+                + [speech_types[name]["ref_text"] for name in speech_types]
+                + [inference_meta_data]
+            )
         else:
             gr.Warning("No audio generated.")
-            return [None] + [speech_types[style]["ref_text"] for style in speech_types]
+            return [None] + [speech_types[name]["ref_text"] for name in speech_types] + [None]
 
     generate_multistyle_btn.click(
         generate_multistyle_speech,
@@ -502,7 +689,7 @@ with gr.Blocks() as app_multistyle:
         + [
             remove_silence_multistyle,
         ],
-        outputs=[audio_output_multistyle] + speech_type_ref_texts,
+        outputs=[audio_output_multistyle] + speech_type_ref_texts + [cherrypick_interface_multistyle],
     )
 
     # Validation function to disable Generate button if speech types are missing
@@ -519,7 +706,7 @@ with gr.Blocks() as app_multistyle:
 
         # Parse the gen_text to get the speech types used
         segments = parse_speechtypes_text(gen_text)
-        speech_types_in_text = set(segment["style"] for segment in segments)
+        speech_types_in_text = set(segment["name"] for segment in segments)
 
         # Check if all speech types in text are available
         missing_speech_types = speech_types_in_text - speech_types_available
@@ -542,43 +729,58 @@ with gr.Blocks() as app_chat:
     gr.Markdown(
         """
 # Voice Chat
-Have a conversation with an AI using your reference voice! 
-1. Upload a reference audio clip and optionally its transcript.
+Have a conversation with an AI using your reference voice!
+1. Upload a reference audio clip and optionally its transcript (via text or .txt file).
 2. Load the chat model.
-3. Record your message through your microphone.
+3. Record your message through your microphone or type it.
 4. The AI will respond using the reference voice.
 """
     )
 
-    if not USING_SPACES:
-        load_chat_model_btn = gr.Button("Load Chat Model", variant="primary")
+    chat_model_name_list = [
+        "Qwen/Qwen2.5-3B-Instruct",
+        "microsoft/Phi-4-mini-instruct",
+    ]
 
-        chat_interface_container = gr.Column(visible=False)
+    @gpu_decorator
+    def load_chat_model(chat_model_name):
+        show_info = gr.Info
+        global chat_model_state, chat_tokenizer_state
+        if chat_model_state is not None:
+            chat_model_state = None
+            chat_tokenizer_state = None
+            gc.collect()
+            torch.cuda.empty_cache()
 
-        @gpu_decorator
-        def load_chat_model():
-            global chat_model_state, chat_tokenizer_state
-            if chat_model_state is None:
-                show_info = gr.Info
-                show_info("Loading chat model...")
-                model_name = "Qwen/Qwen2.5-3B-Instruct"
-                chat_model_state = AutoModelForCausalLM.from_pretrained(
-                    model_name, torch_dtype="auto", device_map="auto"
-                )
-                chat_tokenizer_state = AutoTokenizer.from_pretrained(model_name)
-                show_info("Chat model loaded.")
+        show_info(f"Loading chat model: {chat_model_name}")
+        chat_model_state = AutoModelForCausalLM.from_pretrained(chat_model_name, torch_dtype="auto", device_map="auto")
+        chat_tokenizer_state = AutoTokenizer.from_pretrained(chat_model_name)
+        show_info(f"Chat model {chat_model_name} loaded successfully!")
 
-            return gr.update(visible=False), gr.update(visible=True)
+        return gr.update(visible=False), gr.update(visible=True)
 
-        load_chat_model_btn.click(load_chat_model, outputs=[load_chat_model_btn, chat_interface_container])
+    if USING_SPACES:
+        load_chat_model(chat_model_name_list[0])
 
-    else:
-        chat_interface_container = gr.Column()
+    chat_model_name_input = gr.Dropdown(
+        choices=chat_model_name_list,
+        value=chat_model_name_list[0],
+        label="Chat Model Name",
+        info="Enter the name of a HuggingFace chat model",
+        allow_custom_value=not USING_SPACES,
+    )
+    load_chat_model_btn = gr.Button("Load Chat Model", variant="primary", visible=not USING_SPACES)
+    chat_interface_container = gr.Column(visible=USING_SPACES)
 
-        if chat_model_state is None:
-            model_name = "Qwen/Qwen2.5-3B-Instruct"
-            chat_model_state = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype="auto", device_map="auto")
-            chat_tokenizer_state = AutoTokenizer.from_pretrained(model_name)
+    chat_model_name_input.change(
+        lambda: gr.update(visible=True),
+        None,
+        load_chat_model_btn,
+        show_progress="hidden",
+    )
+    load_chat_model_btn.click(
+        load_chat_model, inputs=[chat_model_name_input], outputs=[load_chat_model_btn, chat_interface_container]
+    )
 
     with chat_interface_container:
         with gr.Row():
@@ -586,22 +788,35 @@ Have a conversation with an AI using your reference voice!
                 ref_audio_chat = gr.Audio(label="Reference Audio", type="filepath")
             with gr.Column():
                 with gr.Accordion("Advanced Settings", open=False):
+                    with gr.Row():
+                        ref_text_chat = gr.Textbox(
+                            label="Reference Text",
+                            info="Optional: Leave blank to auto-transcribe",
+                            lines=2,
+                            scale=3,
+                        )
+                        ref_text_file_chat = gr.File(
+                            label="Load Reference Text from File (.txt)", file_types=[".txt"], scale=1
+                        )
+                    with gr.Row():
+                        randomize_seed_chat = gr.Checkbox(
+                            label="Randomize Seed",
+                            value=True,
+                            info="Uncheck to use the seed specified.",
+                            scale=3,
+                        )
+                        seed_input_chat = gr.Number(show_label=False, value=0, precision=0, scale=1)
                     remove_silence_chat = gr.Checkbox(
                         label="Remove Silences",
                         value=True,
                     )
-                    ref_text_chat = gr.Textbox(
-                        label="Reference Text",
-                        info="Optional: Leave blank to auto-transcribe",
-                        lines=2,
-                    )
                     system_prompt_chat = gr.Textbox(
                         label="System Prompt",
                         value="You are not an AI assistant, you are whoever the user says you are. You must stay in character. Keep your responses concise since they will be spoken out loud.",
                         lines=2,
                     )
 
-        chatbot_interface = gr.Chatbot(label="Conversation")
+        chatbot_interface = gr.Chatbot(label="Conversation", type="messages")
 
         with gr.Row():
             with gr.Column():
@@ -618,132 +833,111 @@ Have a conversation with an AI using your reference voice!
                 send_btn_chat = gr.Button("Send Message")
                 clear_btn_chat = gr.Button("Clear Conversation")
 
-        conversation_state = gr.State(
-            value=[
-                {
-                    "role": "system",
-                    "content": "You are not an AI assistant, you are whoever the user says you are. You must stay in character. Keep your responses concise since they will be spoken out loud.",
-                }
-            ]
-        )
-
-        # Modify process_audio_input to use model and tokenizer from state
+        # Modify process_audio_input to generate user input
         @gpu_decorator
-        def process_audio_input(audio_path, text, history, conv_state):
+        def process_audio_input(conv_state, audio_path, text):
             """Handle audio or text input from user"""
 
             if not audio_path and not text.strip():
-                return history, conv_state, ""
+                return conv_state
 
             if audio_path:
                 text = preprocess_ref_audio_text(audio_path, text)[1]
-
             if not text.strip():
-                return history, conv_state, ""
+                return conv_state
 
             conv_state.append({"role": "user", "content": text})
-            history.append((text, None))
+            return conv_state
 
-            response = generate_response(conv_state, chat_model_state, chat_tokenizer_state)
+        # Use model and tokenizer from state to get text response
+        @gpu_decorator
+        def generate_text_response(conv_state, system_prompt):
+            """Generate text response from AI"""
+
+            system_prompt_state = [{"role": "system", "content": system_prompt}]
+            response = chat_model_inference(system_prompt_state + conv_state, chat_model_state, chat_tokenizer_state)
 
             conv_state.append({"role": "assistant", "content": response})
-            history[-1] = (text, response)
-
-            return history, conv_state, ""
+            return conv_state
 
         @gpu_decorator
-        def generate_audio_response(history, ref_audio, ref_text, remove_silence):
+        def generate_audio_response(conv_state, ref_audio, ref_text, remove_silence, randomize_seed, seed_input):
             """Generate TTS audio for AI response"""
-            if not history or not ref_audio:
-                return None
+            if not conv_state or not ref_audio:
+                return None, ref_text, seed_input
 
-            last_user_message, last_ai_response = history[-1]
-            if not last_ai_response:
-                return None
+            last_ai_response = conv_state[-1]["content"]
+            if not last_ai_response or conv_state[-1]["role"] != "assistant":
+                return None, ref_text, seed_input
 
-            audio_result, _, ref_text_out = infer(
+            if randomize_seed:
+                seed_input = np.random.randint(0, 2**31 - 1)
+
+            audio_result, _, ref_text_out, used_seed = infer(
                 ref_audio,
                 ref_text,
                 last_ai_response,
                 tts_model_choice,
                 remove_silence,
+                seed=seed_input,
                 cross_fade_duration=0.15,
                 speed=1.0,
                 show_info=print,  # show_info=print no pull to top when generating
             )
-            return audio_result, ref_text_out
+            return audio_result, ref_text_out, used_seed
 
         def clear_conversation():
             """Reset the conversation"""
-            return [], [
-                {
-                    "role": "system",
-                    "content": "You are not an AI assistant, you are whoever the user says you are. You must stay in character. Keep your responses concise since they will be spoken out loud.",
-                }
-            ]
+            return [], None
 
-        def update_system_prompt(new_prompt):
-            """Update the system prompt and reset the conversation"""
-            new_conv_state = [{"role": "system", "content": new_prompt}]
-            return [], new_conv_state
-
-        # Handle audio input
-        audio_input_chat.stop_recording(
-            process_audio_input,
-            inputs=[audio_input_chat, text_input_chat, chatbot_interface, conversation_state],
-            outputs=[chatbot_interface, conversation_state],
-        ).then(
-            generate_audio_response,
-            inputs=[chatbot_interface, ref_audio_chat, ref_text_chat, remove_silence_chat],
-            outputs=[audio_output_chat, ref_text_chat],
-        ).then(
-            lambda: None,
-            None,
-            audio_input_chat,
+        ref_text_file_chat.upload(
+            load_text_from_file,
+            inputs=[ref_text_file_chat],
+            outputs=[ref_text_chat],
         )
 
-        # Handle text input
-        text_input_chat.submit(
-            process_audio_input,
-            inputs=[audio_input_chat, text_input_chat, chatbot_interface, conversation_state],
-            outputs=[chatbot_interface, conversation_state],
-        ).then(
-            generate_audio_response,
-            inputs=[chatbot_interface, ref_audio_chat, ref_text_chat, remove_silence_chat],
-            outputs=[audio_output_chat, ref_text_chat],
-        ).then(
-            lambda: None,
-            None,
-            text_input_chat,
-        )
+        for user_operation in [audio_input_chat.stop_recording, text_input_chat.submit, send_btn_chat.click]:
+            user_operation(
+                process_audio_input,
+                inputs=[chatbot_interface, audio_input_chat, text_input_chat],
+                outputs=[chatbot_interface],
+            ).then(
+                generate_text_response,
+                inputs=[chatbot_interface, system_prompt_chat],
+                outputs=[chatbot_interface],
+            ).then(
+                generate_audio_response,
+                inputs=[
+                    chatbot_interface,
+                    ref_audio_chat,
+                    ref_text_chat,
+                    remove_silence_chat,
+                    randomize_seed_chat,
+                    seed_input_chat,
+                ],
+                outputs=[audio_output_chat, ref_text_chat, seed_input_chat],
+            ).then(
+                lambda: [None, None],
+                None,
+                [audio_input_chat, text_input_chat],
+            )
 
-        # Handle send button
-        send_btn_chat.click(
-            process_audio_input,
-            inputs=[audio_input_chat, text_input_chat, chatbot_interface, conversation_state],
-            outputs=[chatbot_interface, conversation_state],
-        ).then(
-            generate_audio_response,
-            inputs=[chatbot_interface, ref_audio_chat, ref_text_chat, remove_silence_chat],
-            outputs=[audio_output_chat, ref_text_chat],
-        ).then(
-            lambda: None,
-            None,
-            text_input_chat,
-        )
+        # Handle clear button or system prompt change and reset conversation
+        for user_operation in [clear_btn_chat.click, system_prompt_chat.change, chatbot_interface.clear]:
+            user_operation(
+                clear_conversation,
+                outputs=[chatbot_interface, audio_output_chat],
+            )
 
-        # Handle clear button
-        clear_btn_chat.click(
-            clear_conversation,
-            outputs=[chatbot_interface, conversation_state],
-        )
 
-        # Handle system prompt change and reset conversation
-        system_prompt_chat.change(
-            update_system_prompt,
-            inputs=system_prompt_chat,
-            outputs=[chatbot_interface, conversation_state],
-        )
+with gr.Blocks() as app_credits:
+    gr.Markdown("""
+# Credits
+
+* [mrfakename](https://github.com/fakerybakery) for the original [online demo](https://huggingface.co/spaces/mrfakename/E2-F5-TTS)
+* [RootingInLoad](https://github.com/RootingInLoad) for initial chunk generation and podcast app exploration
+* [jpgallegoar](https://github.com/jpgallegoar) for multiple speech-type generation & voice chat
+""")
 
 
 with gr.Blocks() as app:
@@ -781,7 +975,7 @@ If you're having issues, try converting your reference audio to WAV or MP3, clip
         global tts_model_choice
         if new_choice == "Custom":  # override in case webpage is refreshed
             custom_ckpt_path, custom_vocab_path, custom_model_cfg = load_last_used_custom()
-            tts_model_choice = ["Custom", custom_ckpt_path, custom_vocab_path, json.loads(custom_model_cfg)]
+            tts_model_choice = ("Custom", custom_ckpt_path, custom_vocab_path, custom_model_cfg)
             return (
                 gr.update(visible=True, value=custom_ckpt_path),
                 gr.update(visible=True, value=custom_vocab_path),
@@ -793,7 +987,7 @@ If you're having issues, try converting your reference audio to WAV or MP3, clip
 
     def set_custom_model(custom_ckpt_path, custom_vocab_path, custom_model_cfg):
         global tts_model_choice
-        tts_model_choice = ["Custom", custom_ckpt_path, custom_vocab_path, json.loads(custom_model_cfg)]
+        tts_model_choice = ("Custom", custom_ckpt_path, custom_vocab_path, custom_model_cfg)
         with open(last_used_custom, "w", encoding="utf-8") as f:
             f.write(custom_ckpt_path + "\n" + custom_vocab_path + "\n" + custom_model_cfg + "\n")
 

src/f5_tts/infer/speech_edit.py

@@ -1,5 +1,6 @@
 import os
 
+
 os.environ["PYTORCH_ENABLE_MPS_FALLBACK"] = "1"  # for MPS device compatibility
 
 from importlib.resources import files
@@ -7,6 +8,7 @@ from importlib.resources import files
 import torch
 import torch.nn.functional as F
 import torchaudio
+from cached_path import cached_path
 from hydra.utils import get_class
 from omegaconf import OmegaConf
 
@@ -14,6 +16,7 @@ from f5_tts.infer.utils_infer import load_checkpoint, load_vocoder, save_spectro
 from f5_tts.model import CFM
 from f5_tts.model.utils import convert_char_to_pinyin, get_tokenizer
 
+
 device = (
     "cuda"
     if torch.cuda.is_available()
@@ -55,7 +58,8 @@ win_length = model_cfg.model.mel_spec.win_length
 n_fft = model_cfg.model.mel_spec.n_fft
 
 
-ckpt_path = str(files("f5_tts").joinpath("../../")) + f"ckpts/{exp_name}/model_{ckpt_step}.safetensors"
+# ckpt_path = str(files("f5_tts").joinpath("../../")) + f"/ckpts/{exp_name}/model_{ckpt_step}.safetensors"
+ckpt_path = str(cached_path(f"hf://SWivid/F5-TTS/{exp_name}/model_{ckpt_step}.safetensors"))
 output_dir = "tests"
 
 
@@ -152,7 +156,7 @@ for part in parts_to_edit:
         dim=-1,
     )
     offset = end * target_sample_rate
-# audio = torch.cat((audio_, audio[:, round(offset):]), dim = -1)
+audio = torch.cat((audio_, audio[:, round(offset) :]), dim=-1)
 edit_mask = F.pad(edit_mask, (0, audio.shape[-1] // hop_length - edit_mask.shape[-1] + 1), value=True)
 audio = audio.to(device)
 edit_mask = edit_mask.to(device)

src/f5_tts/infer/utils_infer.py

@@ -4,6 +4,7 @@ import os
 import sys
 from concurrent.futures import ThreadPoolExecutor
 
+
 os.environ["PYTORCH_ENABLE_MPS_FALLBACK"] = "1"  # for MPS device compatibility
 sys.path.append(f"{os.path.dirname(os.path.abspath(__file__))}/../../third_party/BigVGAN/")
 
@@ -14,6 +15,7 @@ from importlib.resources import files
 
 import matplotlib
 
+
 matplotlib.use("Agg")
 
 import matplotlib.pylab as plt
@@ -21,18 +23,17 @@ import numpy as np
 import torch
 import torchaudio
 import tqdm
-from huggingface_hub import snapshot_download, hf_hub_download
+from huggingface_hub import hf_hub_download
 from pydub import AudioSegment, silence
 from transformers import pipeline
 from vocos import Vocos
 
 from f5_tts.model import CFM
-from f5_tts.model.utils import (
-    get_tokenizer,
-    convert_char_to_pinyin,
-)
+from f5_tts.model.utils import convert_char_to_pinyin, get_tokenizer
+
 
 _ref_audio_cache = {}
+_ref_text_cache = {}
 
 device = (
     "cuda"
@@ -44,6 +45,8 @@ device = (
     else "cpu"
 )
 
+tempfile_kwargs = {"delete_on_close": False} if sys.version_info >= (3, 12) else {"delete": False}
+
 # -----------------------------------------
 
 target_sample_rate = 24000
@@ -128,11 +131,12 @@ def load_vocoder(vocoder_name="vocos", is_local=False, local_path="", device=dev
         except ImportError:
             print("You need to follow the README to init submodule and change the BigVGAN source code.")
         if is_local:
-            """download from https://huggingface.co/nvidia/bigvgan_v2_24khz_100band_256x/tree/main"""
+            # download generator from https://huggingface.co/nvidia/bigvgan_v2_24khz_100band_256x/tree/main
             vocoder = bigvgan.BigVGAN.from_pretrained(local_path, use_cuda_kernel=False)
         else:
-            local_path = snapshot_download(repo_id="nvidia/bigvgan_v2_24khz_100band_256x", cache_dir=hf_cache_dir)
-            vocoder = bigvgan.BigVGAN.from_pretrained(local_path, use_cuda_kernel=False)
+            vocoder = bigvgan.BigVGAN.from_pretrained(
+                "nvidia/bigvgan_v2_24khz_100band_256x", use_cuda_kernel=False, cache_dir=hf_cache_dir
+            )
 
         vocoder.remove_weight_norm()
         vocoder = vocoder.eval().to(device)
@@ -289,62 +293,74 @@ def remove_silence_edges(audio, silence_threshold=-42):
 # preprocess reference audio and text
 
 
-def preprocess_ref_audio_text(ref_audio_orig, ref_text, clip_short=True, show_info=print):
+def preprocess_ref_audio_text(ref_audio_orig, ref_text, show_info=print):
     show_info("Converting audio...")
-    with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as f:
+
+    # Compute a hash of the reference audio file
+    with open(ref_audio_orig, "rb") as audio_file:
+        audio_data = audio_file.read()
+        audio_hash = hashlib.md5(audio_data).hexdigest()
+
+    global _ref_audio_cache
+
+    if audio_hash in _ref_audio_cache:
+        show_info("Using cached preprocessed reference audio...")
+        ref_audio = _ref_audio_cache[audio_hash]
+
+    else:  # first pass, do preprocess
+        with tempfile.NamedTemporaryFile(suffix=".wav", **tempfile_kwargs) as f:
+            temp_path = f.name
+
         aseg = AudioSegment.from_file(ref_audio_orig)
 
-        if clip_short:
-            # 1. try to find long silence for clipping
+        # 1. try to find long silence for clipping
+        non_silent_segs = silence.split_on_silence(
+            aseg, min_silence_len=1000, silence_thresh=-50, keep_silence=1000, seek_step=10
+        )
+        non_silent_wave = AudioSegment.silent(duration=0)
+        for non_silent_seg in non_silent_segs:
+            if len(non_silent_wave) > 6000 and len(non_silent_wave + non_silent_seg) > 12000:
+                show_info("Audio is over 12s, clipping short. (1)")
+                break
+            non_silent_wave += non_silent_seg
+
+        # 2. try to find short silence for clipping if 1. failed
+        if len(non_silent_wave) > 12000:
             non_silent_segs = silence.split_on_silence(
-                aseg, min_silence_len=1000, silence_thresh=-50, keep_silence=1000, seek_step=10
+                aseg, min_silence_len=100, silence_thresh=-40, keep_silence=1000, seek_step=10
             )
             non_silent_wave = AudioSegment.silent(duration=0)
             for non_silent_seg in non_silent_segs:
                 if len(non_silent_wave) > 6000 and len(non_silent_wave + non_silent_seg) > 12000:
-                    show_info("Audio is over 12s, clipping short. (1)")
+                    show_info("Audio is over 12s, clipping short. (2)")
                     break
                 non_silent_wave += non_silent_seg
 
-            # 2. try to find short silence for clipping if 1. failed
-            if len(non_silent_wave) > 12000:
-                non_silent_segs = silence.split_on_silence(
-                    aseg, min_silence_len=100, silence_thresh=-40, keep_silence=1000, seek_step=10
-                )
-                non_silent_wave = AudioSegment.silent(duration=0)
-                for non_silent_seg in non_silent_segs:
-                    if len(non_silent_wave) > 6000 and len(non_silent_wave + non_silent_seg) > 12000:
-                        show_info("Audio is over 12s, clipping short. (2)")
-                        break
-                    non_silent_wave += non_silent_seg
+        aseg = non_silent_wave
 
-            aseg = non_silent_wave
-
-            # 3. if no proper silence found for clipping
-            if len(aseg) > 12000:
-                aseg = aseg[:12000]
-                show_info("Audio is over 12s, clipping short. (3)")
+        # 3. if no proper silence found for clipping
+        if len(aseg) > 12000:
+            aseg = aseg[:12000]
+            show_info("Audio is over 12s, clipping short. (3)")
 
         aseg = remove_silence_edges(aseg) + AudioSegment.silent(duration=50)
-        aseg.export(f.name, format="wav")
-        ref_audio = f.name
+        aseg.export(temp_path, format="wav")
+        ref_audio = temp_path
 
-    # Compute a hash of the reference audio file
-    with open(ref_audio, "rb") as audio_file:
-        audio_data = audio_file.read()
-        audio_hash = hashlib.md5(audio_data).hexdigest()
+        # Cache the processed reference audio
+        _ref_audio_cache[audio_hash] = ref_audio
 
     if not ref_text.strip():
-        global _ref_audio_cache
-        if audio_hash in _ref_audio_cache:
+        global _ref_text_cache
+        if audio_hash in _ref_text_cache:
             # Use cached asr transcription
             show_info("Using cached reference text...")
-            ref_text = _ref_audio_cache[audio_hash]
+            ref_text = _ref_text_cache[audio_hash]
         else:
             show_info("No reference text provided, transcribing reference audio...")
             ref_text = transcribe(ref_audio)
             # Cache the transcribed text (not caching custom ref_text, enabling users to do manual tweak)
-            _ref_audio_cache[audio_hash] = ref_text
+            _ref_text_cache[audio_hash] = ref_text
     else:
         show_info("Using custom reference text...")
 
@@ -383,7 +399,7 @@ def infer_process(
 ):
     # Split the input text into batches
     audio, sr = torchaudio.load(ref_audio)
-    max_chars = int(len(ref_text.encode("utf-8")) / (audio.shape[-1] / sr) * (22 - audio.shape[-1] / sr))
+    max_chars = int(len(ref_text.encode("utf-8")) / (audio.shape[-1] / sr) * (22 - audio.shape[-1] / sr) * speed)
     gen_text_batches = chunk_text(gen_text, max_chars=max_chars)
     for i, gen_text in enumerate(gen_text_batches):
         print(f"gen_text {i}", gen_text)

src/f5_tts/model/__init__.py

@@ -1,9 +1,7 @@
-from f5_tts.model.cfm import CFM
-
-from f5_tts.model.backbones.unett import UNetT
 from f5_tts.model.backbones.dit import DiT
 from f5_tts.model.backbones.mmdit import MMDiT
-
+from f5_tts.model.backbones.unett import UNetT
+from f5_tts.model.cfm import CFM
 from f5_tts.model.trainer import Trainer
 
 

src/f5_tts/model/backbones/dit.py

@@ -10,19 +10,18 @@ d - dimension
 from __future__ import annotations
 
 import torch
-from torch import nn
 import torch.nn.functional as F
-
+from torch import nn
 from x_transformers.x_transformers import RotaryEmbedding
 
 from f5_tts.model.modules import (
-    TimestepEmbedding,
+    AdaLayerNorm_Final,
     ConvNeXtV2Block,
     ConvPositionEmbedding,
     DiTBlock,
-    AdaLayerNorm_Final,
-    precompute_freqs_cis,
+    TimestepEmbedding,
     get_pos_embed_indices,
+    precompute_freqs_cis,
 )
 
 
@@ -30,11 +29,16 @@ from f5_tts.model.modules import (
 
 
 class TextEmbedding(nn.Module):
-    def __init__(self, text_num_embeds, text_dim, mask_padding=True, conv_layers=0, conv_mult=2):
+    def __init__(
+        self, text_num_embeds, text_dim, mask_padding=True, average_upsampling=False, conv_layers=0, conv_mult=2
+    ):
         super().__init__()
         self.text_embed = nn.Embedding(text_num_embeds + 1, text_dim)  # use 0 as filler token
 
         self.mask_padding = mask_padding  # mask filler and batch padding tokens or not
+        self.average_upsampling = average_upsampling  # zipvoice-style text late average upsampling (after text encoder)
+        if average_upsampling:
+            assert mask_padding, "text_embedding_average_upsampling requires text_mask_padding to be True"
 
         if conv_layers > 0:
             self.extra_modeling = True
@@ -46,11 +50,47 @@ class TextEmbedding(nn.Module):
         else:
             self.extra_modeling = False
 
-    def forward(self, text: int["b nt"], seq_len, drop_text=False):  # noqa: F722
+    def average_upsample_text_by_mask(self, text, text_mask, audio_mask):
+        batch, text_len, text_dim = text.shape
+
+        if audio_mask is None:
+            audio_mask = torch.ones_like(text_mask, dtype=torch.bool)
+        valid_mask = audio_mask & text_mask
+        audio_lens = audio_mask.sum(dim=1)  # [batch]
+        valid_lens = valid_mask.sum(dim=1)  # [batch]
+
+        upsampled_text = torch.zeros_like(text)
+
+        for i in range(batch):
+            audio_len = audio_lens[i].item()
+            valid_len = valid_lens[i].item()
+
+            if valid_len == 0:
+                continue
+
+            valid_ind = torch.where(valid_mask[i])[0]
+            valid_data = text[i, valid_ind, :]  # [valid_len, text_dim]
+
+            base_repeat = audio_len // valid_len
+            remainder = audio_len % valid_len
+
+            indices = []
+            for j in range(valid_len):
+                repeat_count = base_repeat + (1 if j >= valid_len - remainder else 0)
+                indices.extend([j] * repeat_count)
+
+            indices = torch.tensor(indices[:audio_len], device=text.device, dtype=torch.long)
+            upsampled = valid_data[indices]  # [audio_len, text_dim]
+
+            upsampled_text[i, :audio_len, :] = upsampled
+
+        return upsampled_text
+
+    def forward(self, text: int["b nt"], seq_len, drop_text=False, audio_mask: bool["b n"] | None = None):  # noqa: F722
         text = text + 1  # use 0 as filler token. preprocess of batch pad -1, see list_str_to_idx()
         text = text[:, :seq_len]  # curtail if character tokens are more than the mel spec tokens
         batch, text_len = text.shape[0], text.shape[1]
-        text = F.pad(text, (0, seq_len - text_len), value=0)
+        text = F.pad(text, (0, seq_len - text_len), value=0)  # (opt.) if not self.average_upsampling:
         if self.mask_padding:
             text_mask = text == 0
 
@@ -62,7 +102,7 @@ class TextEmbedding(nn.Module):
         # possible extra modeling
         if self.extra_modeling:
             # sinus pos emb
-            batch_start = torch.zeros((batch,), dtype=torch.long)
+            batch_start = torch.zeros((batch,), device=text.device, dtype=torch.long)
             pos_idx = get_pos_embed_indices(batch_start, seq_len, max_pos=self.precompute_max_pos)
             text_pos_embed = self.freqs_cis[pos_idx]
             text = text + text_pos_embed
@@ -76,6 +116,9 @@ class TextEmbedding(nn.Module):
             else:
                 text = self.text_blocks(text)
 
+        if self.average_upsampling:
+            text = self.average_upsample_text_by_mask(text, ~text_mask, audio_mask)
+
         return text
 
 
@@ -114,9 +157,12 @@ class DiT(nn.Module):
         text_num_embeds=256,
         text_dim=None,
         text_mask_padding=True,
+        text_embedding_average_upsampling=False,
         qk_norm=None,
         conv_layers=0,
         pe_attn_head=None,
+        attn_backend="torch",  # "torch" | "flash_attn"
+        attn_mask_enabled=False,
         long_skip_connection=False,
         checkpoint_activations=False,
     ):
@@ -126,7 +172,11 @@ class DiT(nn.Module):
         if text_dim is None:
             text_dim = mel_dim
         self.text_embed = TextEmbedding(
-            text_num_embeds, text_dim, mask_padding=text_mask_padding, conv_layers=conv_layers
+            text_num_embeds,
+            text_dim,
+            mask_padding=text_mask_padding,
+            average_upsampling=text_embedding_average_upsampling,
+            conv_layers=conv_layers,
         )
         self.text_cond, self.text_uncond = None, None  # text cache
         self.input_embed = InputEmbedding(mel_dim, text_dim, dim)
@@ -146,6 +196,8 @@ class DiT(nn.Module):
                     dropout=dropout,
                     qk_norm=qk_norm,
                     pe_attn_head=pe_attn_head,
+                    attn_backend=attn_backend,
+                    attn_mask_enabled=attn_mask_enabled,
                 )
                 for _ in range(depth)
             ]
@@ -179,6 +231,33 @@ class DiT(nn.Module):
 
         return ckpt_forward
 
+    def get_input_embed(
+        self,
+        x,  # b n d
+        cond,  # b n d
+        text,  # b nt
+        drop_audio_cond: bool = False,
+        drop_text: bool = False,
+        cache: bool = True,
+        audio_mask: bool["b n"] | None = None,  # noqa: F722
+    ):
+        seq_len = x.shape[1]
+        if cache:
+            if drop_text:
+                if self.text_uncond is None:
+                    self.text_uncond = self.text_embed(text, seq_len, drop_text=True, audio_mask=audio_mask)
+                text_embed = self.text_uncond
+            else:
+                if self.text_cond is None:
+                    self.text_cond = self.text_embed(text, seq_len, drop_text=False, audio_mask=audio_mask)
+                text_embed = self.text_cond
+        else:
+            text_embed = self.text_embed(text, seq_len, drop_text=drop_text, audio_mask=audio_mask)
+
+        x = self.input_embed(x, cond, text_embed, drop_audio_cond=drop_audio_cond)
+
+        return x
+
     def clear_cache(self):
         self.text_cond, self.text_uncond = None, None
 
@@ -188,10 +267,11 @@ class DiT(nn.Module):
         cond: float["b n d"],  # masked cond audio  # noqa: F722
         text: int["b nt"],  # text  # noqa: F722
         time: float["b"] | float[""],  # time step  # noqa: F821 F722
-        drop_audio_cond,  # cfg for cond audio
-        drop_text,  # cfg for text
         mask: bool["b n"] | None = None,  # noqa: F722
-        cache=False,
+        drop_audio_cond: bool = False,  # cfg for cond audio
+        drop_text: bool = False,  # cfg for text
+        cfg_infer: bool = False,  # cfg inference, pack cond & uncond forward
+        cache: bool = False,
     ):
         batch, seq_len = x.shape[0], x.shape[1]
         if time.ndim == 0:
@@ -199,18 +279,20 @@ class DiT(nn.Module):
 
         # t: conditioning time, text: text, x: noised audio + cond audio + text
         t = self.time_embed(time)
-        if cache:
-            if drop_text:
-                if self.text_uncond is None:
-                    self.text_uncond = self.text_embed(text, seq_len, drop_text=True)
-                text_embed = self.text_uncond
-            else:
-                if self.text_cond is None:
-                    self.text_cond = self.text_embed(text, seq_len, drop_text=False)
-                text_embed = self.text_cond
+        if cfg_infer:  # pack cond & uncond forward: b n d -> 2b n d
+            x_cond = self.get_input_embed(
+                x, cond, text, drop_audio_cond=False, drop_text=False, cache=cache, audio_mask=mask
+            )
+            x_uncond = self.get_input_embed(
+                x, cond, text, drop_audio_cond=True, drop_text=True, cache=cache, audio_mask=mask
+            )
+            x = torch.cat((x_cond, x_uncond), dim=0)
+            t = torch.cat((t, t), dim=0)
+            mask = torch.cat((mask, mask), dim=0) if mask is not None else None
         else:
-            text_embed = self.text_embed(text, seq_len, drop_text=drop_text)
-        x = self.input_embed(x, cond, text_embed, drop_audio_cond=drop_audio_cond)
+            x = self.get_input_embed(
+                x, cond, text, drop_audio_cond=drop_audio_cond, drop_text=drop_text, cache=cache, audio_mask=mask
+            )
 
         rope = self.rotary_embed.forward_from_seq_len(seq_len)
 

src/f5_tts/model/backbones/mmdit.py

@@ -11,16 +11,15 @@ from __future__ import annotations
 
 import torch
 from torch import nn
-
 from x_transformers.x_transformers import RotaryEmbedding
 
 from f5_tts.model.modules import (
-    TimestepEmbedding,
+    AdaLayerNorm_Final,
     ConvPositionEmbedding,
     MMDiTBlock,
-    AdaLayerNorm_Final,
-    precompute_freqs_cis,
+    TimestepEmbedding,
     get_pos_embed_indices,
+    precompute_freqs_cis,
 )
 
 
@@ -142,26 +141,15 @@ class MMDiT(nn.Module):
         nn.init.constant_(self.proj_out.weight, 0)
         nn.init.constant_(self.proj_out.bias, 0)
 
-    def clear_cache(self):
-        self.text_cond, self.text_uncond = None, None
-
-    def forward(
+    def get_input_embed(
         self,
-        x: float["b n d"],  # nosied input audio  # noqa: F722
-        cond: float["b n d"],  # masked cond audio  # noqa: F722
-        text: int["b nt"],  # text  # noqa: F722
-        time: float["b"] | float[""],  # time step  # noqa: F821 F722
-        drop_audio_cond,  # cfg for cond audio
-        drop_text,  # cfg for text
-        mask: bool["b n"] | None = None,  # noqa: F722
-        cache=False,
+        x,  # b n d
+        cond,  # b n d
+        text,  # b nt
+        drop_audio_cond: bool = False,
+        drop_text: bool = False,
+        cache: bool = True,
     ):
-        batch = x.shape[0]
-        if time.ndim == 0:
-            time = time.repeat(batch)
-
-        # t: conditioning (time), c: context (text + masked cond audio), x: noised input audio
-        t = self.time_embed(time)
         if cache:
             if drop_text:
                 if self.text_uncond is None:
@@ -175,6 +163,41 @@ class MMDiT(nn.Module):
             c = self.text_embed(text, drop_text=drop_text)
         x = self.audio_embed(x, cond, drop_audio_cond=drop_audio_cond)
 
+        return x, c
+
+    def clear_cache(self):
+        self.text_cond, self.text_uncond = None, None
+
+    def forward(
+        self,
+        x: float["b n d"],  # nosied input audio  # noqa: F722
+        cond: float["b n d"],  # masked cond audio  # noqa: F722
+        text: int["b nt"],  # text  # noqa: F722
+        time: float["b"] | float[""],  # time step  # noqa: F821 F722
+        mask: bool["b n"] | None = None,  # noqa: F722
+        drop_audio_cond: bool = False,  # cfg for cond audio
+        drop_text: bool = False,  # cfg for text
+        cfg_infer: bool = False,  # cfg inference, pack cond & uncond forward
+        cache: bool = False,
+    ):
+        batch = x.shape[0]
+        if time.ndim == 0:
+            time = time.repeat(batch)
+
+        # t: conditioning (time), c: context (text + masked cond audio), x: noised input audio
+        t = self.time_embed(time)
+        if cfg_infer:  # pack cond & uncond forward: b n d -> 2b n d
+            x_cond, c_cond = self.get_input_embed(x, cond, text, drop_audio_cond=False, drop_text=False, cache=cache)
+            x_uncond, c_uncond = self.get_input_embed(x, cond, text, drop_audio_cond=True, drop_text=True, cache=cache)
+            x = torch.cat((x_cond, x_uncond), dim=0)
+            c = torch.cat((c_cond, c_uncond), dim=0)
+            t = torch.cat((t, t), dim=0)
+            mask = torch.cat((mask, mask), dim=0) if mask is not None else None
+        else:
+            x, c = self.get_input_embed(
+                x, cond, text, drop_audio_cond=drop_audio_cond, drop_text=drop_text, cache=cache
+            )
+
         seq_len = x.shape[1]
         text_len = text.shape[1]
         rope_audio = self.rotary_embed.forward_from_seq_len(seq_len)

src/f5_tts/model/backbones/unett.py

@@ -8,24 +8,24 @@ d - dimension
 """
 
 from __future__ import annotations
+
 from typing import Literal
 
 import torch
-from torch import nn
 import torch.nn.functional as F
-
+from torch import nn
 from x_transformers import RMSNorm
 from x_transformers.x_transformers import RotaryEmbedding
 
 from f5_tts.model.modules import (
-    TimestepEmbedding,
-    ConvNeXtV2Block,
-    ConvPositionEmbedding,
     Attention,
     AttnProcessor,
+    ConvNeXtV2Block,
+    ConvPositionEmbedding,
     FeedForward,
-    precompute_freqs_cis,
+    TimestepEmbedding,
     get_pos_embed_indices,
+    precompute_freqs_cis,
 )
 
 
@@ -178,26 +178,16 @@ class UNetT(nn.Module):
         self.norm_out = RMSNorm(dim)
         self.proj_out = nn.Linear(dim, mel_dim)
 
-    def clear_cache(self):
-        self.text_cond, self.text_uncond = None, None
-
-    def forward(
+    def get_input_embed(
         self,
-        x: float["b n d"],  # nosied input audio  # noqa: F722
-        cond: float["b n d"],  # masked cond audio  # noqa: F722
-        text: int["b nt"],  # text  # noqa: F722
-        time: float["b"] | float[""],  # time step  # noqa: F821 F722
-        drop_audio_cond,  # cfg for cond audio
-        drop_text,  # cfg for text
-        mask: bool["b n"] | None = None,  # noqa: F722
-        cache=False,
+        x,  # b n d
+        cond,  # b n d
+        text,  # b nt
+        drop_audio_cond: bool = False,
+        drop_text: bool = False,
+        cache: bool = True,
     ):
-        batch, seq_len = x.shape[0], x.shape[1]
-        if time.ndim == 0:
-            time = time.repeat(batch)
-
-        # t: conditioning time, c: context (text + masked cond audio), x: noised input audio
-        t = self.time_embed(time)
+        seq_len = x.shape[1]
         if cache:
             if drop_text:
                 if self.text_uncond is None:
@@ -209,8 +199,41 @@ class UNetT(nn.Module):
                 text_embed = self.text_cond
         else:
             text_embed = self.text_embed(text, seq_len, drop_text=drop_text)
+
         x = self.input_embed(x, cond, text_embed, drop_audio_cond=drop_audio_cond)
 
+        return x
+
+    def clear_cache(self):
+        self.text_cond, self.text_uncond = None, None
+
+    def forward(
+        self,
+        x: float["b n d"],  # nosied input audio  # noqa: F722
+        cond: float["b n d"],  # masked cond audio  # noqa: F722
+        text: int["b nt"],  # text  # noqa: F722
+        time: float["b"] | float[""],  # time step  # noqa: F821 F722
+        mask: bool["b n"] | None = None,  # noqa: F722
+        drop_audio_cond: bool = False,  # cfg for cond audio
+        drop_text: bool = False,  # cfg for text
+        cfg_infer: bool = False,  # cfg inference, pack cond & uncond forward
+        cache: bool = False,
+    ):
+        batch, seq_len = x.shape[0], x.shape[1]
+        if time.ndim == 0:
+            time = time.repeat(batch)
+
+        # t: conditioning time, c: context (text + masked cond audio), x: noised input audio
+        t = self.time_embed(time)
+        if cfg_infer:  # pack cond & uncond forward: b n d -> 2b n d
+            x_cond = self.get_input_embed(x, cond, text, drop_audio_cond=False, drop_text=False, cache=cache)
+            x_uncond = self.get_input_embed(x, cond, text, drop_audio_cond=True, drop_text=True, cache=cache)
+            x = torch.cat((x_cond, x_uncond), dim=0)
+            t = torch.cat((t, t), dim=0)
+            mask = torch.cat((mask, mask), dim=0) if mask is not None else None
+        else:
+            x = self.get_input_embed(x, cond, text, drop_audio_cond=drop_audio_cond, drop_text=drop_text, cache=cache)
+
         # postfix time t to input x, [b n d] -> [b n+1 d]
         x = torch.cat([t.unsqueeze(1), x], dim=1)  # pack t to x
         if mask is not None:

src/f5_tts/model/cfm.py

@@ -22,6 +22,7 @@ from f5_tts.model.modules import MelSpec
 from f5_tts.model.utils import (
     default,
     exists,
+    get_epss_timesteps,
     lens_to_mask,
     list_str_to_idx,
     list_str_to_tensor,
@@ -92,6 +93,7 @@ class CFM(nn.Module):
         seed: int | None = None,
         max_duration=4096,
         vocoder: Callable[[float["b d n"]], float["b nw"]] | None = None,  # noqa: F722
+        use_epss=True,
         no_ref_audio=False,
         duplicate_test=False,
         t_inter=0.1,
@@ -160,16 +162,31 @@ class CFM(nn.Module):
             # at each step, conditioning is fixed
             # step_cond = torch.where(cond_mask, cond, torch.zeros_like(cond))
 
-            # predict flow
-            pred = self.transformer(
-                x=x, cond=step_cond, text=text, time=t, mask=mask, drop_audio_cond=False, drop_text=False, cache=True
-            )
+            # predict flow (cond)
             if cfg_strength < 1e-5:
+                pred = self.transformer(
+                    x=x,
+                    cond=step_cond,
+                    text=text,
+                    time=t,
+                    mask=mask,
+                    drop_audio_cond=False,
+                    drop_text=False,
+                    cache=True,
+                )
                 return pred
 
-            null_pred = self.transformer(
-                x=x, cond=step_cond, text=text, time=t, mask=mask, drop_audio_cond=True, drop_text=True, cache=True
+            # predict flow (cond and uncond), for classifier-free guidance
+            pred_cfg = self.transformer(
+                x=x,
+                cond=step_cond,
+                text=text,
+                time=t,
+                mask=mask,
+                cfg_infer=True,
+                cache=True,
             )
+            pred, null_pred = torch.chunk(pred_cfg, 2, dim=0)
             return pred + (pred - null_pred) * cfg_strength
 
         # noise input
@@ -190,7 +207,10 @@ class CFM(nn.Module):
             y0 = (1 - t_start) * y0 + t_start * test_cond
             steps = int(steps * (1 - t_start))
 
-        t = torch.linspace(t_start, 1, steps + 1, device=self.device, dtype=step_cond.dtype)
+        if t_start == 0 and use_epss:  # use Empirically Pruned Step Sampling for low NFE
+            t = get_epss_timesteps(steps, device=self.device, dtype=step_cond.dtype)
+        else:
+            t = torch.linspace(t_start, 1, steps + 1, device=self.device, dtype=step_cond.dtype)
         if sway_sampling_coef is not None:
             t = t + sway_sampling_coef * (torch.cos(torch.pi / 2 * t) - 1 + t)
 
@@ -270,10 +290,9 @@ class CFM(nn.Module):
         else:
             drop_text = False
 
-        # if want rigourously mask out padding, record in collate_fn in dataset.py, and pass in here
-        # adding mask will use more memory, thus also need to adjust batchsampler with scaled down threshold for long sequences
+        # apply mask will use more memory; might adjust batchsize or batchsampler long sequence threshold
         pred = self.transformer(
-            x=φ, cond=cond, text=text, time=time, drop_audio_cond=drop_audio_cond, drop_text=drop_text
+            x=φ, cond=cond, text=text, time=time, drop_audio_cond=drop_audio_cond, drop_text=drop_text, mask=mask
         )
 
         # flow matching loss

src/f5_tts/model/dataset.py

@@ -312,7 +312,7 @@ def collate_fn(batch):
     max_mel_length = mel_lengths.amax()
 
     padded_mel_specs = []
-    for spec in mel_specs:  # TODO. maybe records mask for attention here
+    for spec in mel_specs:
         padding = (0, max_mel_length - spec.size(-1))
         padded_spec = F.pad(spec, padding, value=0)
         padded_mel_specs.append(padded_spec)
@@ -324,7 +324,7 @@ def collate_fn(batch):
 
     return dict(
         mel=mel_specs,
-        mel_lengths=mel_lengths,
+        mel_lengths=mel_lengths,  # records for padding mask
         text=text,
         text_lengths=text_lengths,
     )

src/f5_tts/model/modules.py

@@ -6,6 +6,7 @@ nt - text sequence
 nw - raw wave length
 d - dimension
 """
+# flake8: noqa
 
 from __future__ import annotations
 
@@ -19,6 +20,8 @@ from librosa.filters import mel as librosa_mel_fn
 from torch import nn
 from x_transformers.x_transformers import apply_rotary_pos_emb
 
+from f5_tts.model.utils import is_package_available
+
 
 # raw wav to mel spec
 
@@ -175,7 +178,7 @@ class ConvPositionEmbedding(nn.Module):
             nn.Mish(),
         )
 
-    def forward(self, x: float["b n d"], mask: bool["b n"] | None = None):  # noqa: F722
+    def forward(self, x: float["b n d"], mask: bool["b n"] | None = None):
         if mask is not None:
             mask = mask[..., None]
             x = x.masked_fill(~mask, 0.0)
@@ -417,9 +420,9 @@ class Attention(nn.Module):
 
     def forward(
         self,
-        x: float["b n d"],  # noised input x  # noqa: F722
-        c: float["b n d"] = None,  # context c  # noqa: F722
-        mask: bool["b n"] | None = None,  # noqa: F722
+        x: float["b n d"],  # noised input x
+        c: float["b n d"] = None,  # context c
+        mask: bool["b n"] | None = None,
         rope=None,  # rotary position embedding for x
         c_rope=None,  # rotary position embedding for c
     ) -> torch.Tensor:
@@ -431,19 +434,30 @@ class Attention(nn.Module):
 
 # Attention processor
 
+if is_package_available("flash_attn"):
+    from flash_attn.bert_padding import pad_input, unpad_input
+    from flash_attn import flash_attn_varlen_func, flash_attn_func
+
 
 class AttnProcessor:
     def __init__(
         self,
         pe_attn_head: int | None = None,  # number of attention head to apply rope, None for all
+        attn_backend: str = "torch",  # "torch" or "flash_attn"
+        attn_mask_enabled: bool = True,
     ):
+        if attn_backend == "flash_attn":
+            assert is_package_available("flash_attn"), "Please install flash-attn first."
+
         self.pe_attn_head = pe_attn_head
+        self.attn_backend = attn_backend
+        self.attn_mask_enabled = attn_mask_enabled
 
     def __call__(
         self,
         attn: Attention,
-        x: float["b n d"],  # noised input x  # noqa: F722
-        mask: bool["b n"] | None = None,  # noqa: F722
+        x: float["b n d"],  # noised input x
+        mask: bool["b n"] | None = None,
         rope=None,  # rotary position embedding
     ) -> torch.FloatTensor:
         batch_size = x.shape[0]
@@ -479,16 +493,40 @@ class AttnProcessor:
                 query = apply_rotary_pos_emb(query, freqs, q_xpos_scale)
                 key = apply_rotary_pos_emb(key, freqs, k_xpos_scale)
 
-        # mask. e.g. inference got a batch with different target durations, mask out the padding
-        if mask is not None:
-            attn_mask = mask
-            attn_mask = attn_mask.unsqueeze(1).unsqueeze(1)  # 'b n -> b 1 1 n'
-            attn_mask = attn_mask.expand(batch_size, attn.heads, query.shape[-2], key.shape[-2])
-        else:
-            attn_mask = None
+        if self.attn_backend == "torch":
+            # mask. e.g. inference got a batch with different target durations, mask out the padding
+            if self.attn_mask_enabled and mask is not None:
+                attn_mask = mask
+                attn_mask = attn_mask.unsqueeze(1).unsqueeze(1)  # 'b n -> b 1 1 n'
+                attn_mask = attn_mask.expand(batch_size, attn.heads, query.shape[-2], key.shape[-2])
+            else:
+                attn_mask = None
+            x = F.scaled_dot_product_attention(query, key, value, attn_mask=attn_mask, dropout_p=0.0, is_causal=False)
+            x = x.transpose(1, 2).reshape(batch_size, -1, attn.heads * head_dim)
+
+        elif self.attn_backend == "flash_attn":
+            query = query.transpose(1, 2)  # [b, h, n, d] -> [b, n, h, d]
+            key = key.transpose(1, 2)
+            value = value.transpose(1, 2)
+            if self.attn_mask_enabled and mask is not None:
+                query, indices, q_cu_seqlens, q_max_seqlen_in_batch, _ = unpad_input(query, mask)
+                key, _, k_cu_seqlens, k_max_seqlen_in_batch, _ = unpad_input(key, mask)
+                value, _, _, _, _ = unpad_input(value, mask)
+                x = flash_attn_varlen_func(
+                    query,
+                    key,
+                    value,
+                    q_cu_seqlens,
+                    k_cu_seqlens,
+                    q_max_seqlen_in_batch,
+                    k_max_seqlen_in_batch,
+                )
+                x = pad_input(x, indices, batch_size, q_max_seqlen_in_batch)
+                x = x.reshape(batch_size, -1, attn.heads * head_dim)
+            else:
+                x = flash_attn_func(query, key, value, dropout_p=0.0, causal=False)
+                x = x.reshape(batch_size, -1, attn.heads * head_dim)
 
-        x = F.scaled_dot_product_attention(query, key, value, attn_mask=attn_mask, dropout_p=0.0, is_causal=False)
-        x = x.transpose(1, 2).reshape(batch_size, -1, attn.heads * head_dim)
         x = x.to(query.dtype)
 
         # linear proj
@@ -514,9 +552,9 @@ class JointAttnProcessor:
     def __call__(
         self,
         attn: Attention,
-        x: float["b n d"],  # noised input x  # noqa: F722
-        c: float["b nt d"] = None,  # context c, here text # noqa: F722
-        mask: bool["b n"] | None = None,  # noqa: F722
+        x: float["b n d"],  # noised input x
+        c: float["b nt d"] = None,  # context c, here text
+        mask: bool["b n"] | None = None,
         rope=None,  # rotary position embedding for x
         c_rope=None,  # rotary position embedding for c
     ) -> torch.FloatTensor:
@@ -608,12 +646,27 @@ class JointAttnProcessor:
 
 
 class DiTBlock(nn.Module):
-    def __init__(self, dim, heads, dim_head, ff_mult=4, dropout=0.1, qk_norm=None, pe_attn_head=None):
+    def __init__(
+        self,
+        dim,
+        heads,
+        dim_head,
+        ff_mult=4,
+        dropout=0.1,
+        qk_norm=None,
+        pe_attn_head=None,
+        attn_backend="torch",  # "torch" or "flash_attn"
+        attn_mask_enabled=True,
+    ):
         super().__init__()
 
         self.attn_norm = AdaLayerNorm(dim)
         self.attn = Attention(
-            processor=AttnProcessor(pe_attn_head=pe_attn_head),
+            processor=AttnProcessor(
+                pe_attn_head=pe_attn_head,
+                attn_backend=attn_backend,
+                attn_mask_enabled=attn_mask_enabled,
+            ),
             dim=dim,
             heads=heads,
             dim_head=dim_head,
@@ -724,7 +777,7 @@ class TimestepEmbedding(nn.Module):
         self.time_embed = SinusPositionEmbedding(freq_embed_dim)
         self.time_mlp = nn.Sequential(nn.Linear(freq_embed_dim, dim), nn.SiLU(), nn.Linear(dim, dim))
 
-    def forward(self, timestep: float["b"]):  # noqa: F821
+    def forward(self, timestep: float["b"]):
         time_hidden = self.time_embed(timestep)
         time_hidden = time_hidden.to(timestep.dtype)
         time = self.time_mlp(time_hidden)  # b d

src/f5_tts/model/trainer.py

@@ -19,6 +19,7 @@ from f5_tts.model import CFM
 from f5_tts.model.dataset import DynamicBatchSampler, collate_fn
 from f5_tts.model.utils import default, exists
 
+
 # trainer
 
 
@@ -51,7 +52,7 @@ class Trainer:
         mel_spec_type: str = "vocos",  # "vocos" | "bigvgan"
         is_local_vocoder: bool = False,  # use local path vocoder
         local_vocoder_path: str = "",  # local vocoder path
-        cfg_dict: dict = dict(),  # training config
+        model_cfg_dict: dict = dict(),  # training config
     ):
         ddp_kwargs = DistributedDataParallelKwargs(find_unused_parameters=True)
 
@@ -73,8 +74,8 @@ class Trainer:
             else:
                 init_kwargs = {"wandb": {"resume": "allow", "name": wandb_run_name}}
 
-            if not cfg_dict:
-                cfg_dict = {
+            if not model_cfg_dict:
+                model_cfg_dict = {
                     "epochs": epochs,
                     "learning_rate": learning_rate,
                     "num_warmup_updates": num_warmup_updates,
@@ -85,11 +86,11 @@ class Trainer:
                     "max_grad_norm": max_grad_norm,
                     "noise_scheduler": noise_scheduler,
                 }
-            cfg_dict["gpus"] = self.accelerator.num_processes
+            model_cfg_dict["gpus"] = self.accelerator.num_processes
             self.accelerator.init_trackers(
                 project_name=wandb_project,
                 init_kwargs=init_kwargs,
-                config=cfg_dict,
+                config=model_cfg_dict,
             )
 
         elif self.logger == "tensorboard":
@@ -148,7 +149,7 @@ class Trainer:
         if self.is_main:
             checkpoint = dict(
                 model_state_dict=self.accelerator.unwrap_model(self.model).state_dict(),
-                optimizer_state_dict=self.accelerator.unwrap_model(self.optimizer).state_dict(),
+                optimizer_state_dict=self.optimizer.state_dict(),
                 ema_model_state_dict=self.ema_model.state_dict(),
                 scheduler_state_dict=self.scheduler.state_dict(),
                 update=update,
@@ -241,7 +242,7 @@ class Trainer:
                     del checkpoint["model_state_dict"][key]
 
             self.accelerator.unwrap_model(self.model).load_state_dict(checkpoint["model_state_dict"])
-            self.accelerator.unwrap_model(self.optimizer).load_state_dict(checkpoint["optimizer_state_dict"])
+            self.optimizer.load_state_dict(checkpoint["optimizer_state_dict"])
             if self.scheduler:
                 self.scheduler.load_state_dict(checkpoint["scheduler_state_dict"])
             update = checkpoint["update"]
@@ -395,6 +396,9 @@ class Trainer:
                         self.writer.add_scalar("loss", loss.item(), global_update)
                         self.writer.add_scalar("lr", self.scheduler.get_last_lr()[0], global_update)
 
+                if global_update % self.last_per_updates == 0 and self.accelerator.sync_gradients:
+                    self.save_checkpoint(global_update, last=True)
+
                 if global_update % self.save_per_updates == 0 and self.accelerator.sync_gradients:
                     self.save_checkpoint(global_update)
 
@@ -430,9 +434,6 @@ class Trainer:
                         )
                         self.model.train()
 
-                if global_update % self.last_per_updates == 0 and self.accelerator.sync_gradients:
-                    self.save_checkpoint(global_update, last=True)
-
         self.save_checkpoint(global_update, last=True)
 
         self.accelerator.end_training()

src/f5_tts/model/utils.py

@@ -5,11 +5,10 @@ import random
 from collections import defaultdict
 from importlib.resources import files
 
-import torch
-from torch.nn.utils.rnn import pad_sequence
-
 import jieba
-from pypinyin import lazy_pinyin, Style
+import torch
+from pypinyin import Style, lazy_pinyin
+from torch.nn.utils.rnn import pad_sequence
 
 
 # seed everything
@@ -36,6 +35,16 @@ def default(v, d):
     return v if exists(v) else d
 
 
+def is_package_available(package_name: str) -> bool:
+    try:
+        import importlib
+
+        package_exists = importlib.util.find_spec(package_name) is not None
+        return package_exists
+    except Exception:
+        return False
+
+
 # tensor helpers
 
 
@@ -190,3 +199,22 @@ def repetition_found(text, length=2, tolerance=10):
         if count > tolerance:
             return True
     return False
+
+
+# get the empirically pruned step for sampling
+
+
+def get_epss_timesteps(n, device, dtype):
+    dt = 1 / 32
+    predefined_timesteps = {
+        5: [0, 2, 4, 8, 16, 32],
+        6: [0, 2, 4, 6, 8, 16, 32],
+        7: [0, 2, 4, 6, 8, 16, 24, 32],
+        10: [0, 2, 4, 6, 8, 12, 16, 20, 24, 28, 32],
+        12: [0, 2, 4, 6, 8, 10, 12, 14, 16, 20, 24, 28, 32],
+        16: [0, 1, 2, 3, 4, 5, 6, 7, 8, 10, 12, 14, 16, 20, 24, 28, 32],
+    }
+    t = predefined_timesteps.get(n, [])
+    if not t:
+        return torch.linspace(0, 1, n + 1, device=device, dtype=dtype)
+    return dt * torch.tensor(t, device=device, dtype=dtype)

src/f5_tts/runtime/triton_trtllm/Dockerfile.server

@@ -0,0 +1,3 @@
+FROM nvcr.io/nvidia/tritonserver:24.12-py3
+RUN pip install tritonclient[grpc] tensorrt-llm==0.16.0 torchaudio==2.5.1 jieba pypinyin librosa vocos
+WORKDIR /workspace

src/f5_tts/runtime/triton_trtllm/README.md

@@ -0,0 +1,69 @@
+## Triton Inference Serving Best Practice for F5-TTS
+
+### Quick Start
+Directly launch the service using docker compose.
+```sh
+# TODO: support F5TTS_v1_Base
+MODEL=F5TTS_Base docker compose up
+```
+
+### Build Image
+Build the docker image from scratch. 
+```sh
+docker build . -f Dockerfile.server -t soar97/triton-f5-tts:24.12
+```
+
+### Create Docker Container
+```sh
+your_mount_dir=/mnt:/mnt
+docker run -it --name "f5-server" --gpus all --net host -v $your_mount_dir --shm-size=2g soar97/triton-f5-tts:24.12
+```
+
+### Export Models to TensorRT-LLM and Launch Server
+Inside docker container, we would follow the official guide of TensorRT-LLM to build qwen and whisper TensorRT-LLM engines. See [here](https://github.com/NVIDIA/TensorRT-LLM/tree/main/examples/whisper).
+```sh
+bash run.sh 0 4 F5TTS_Base
+```
+
+### HTTP Client
+```sh
+python3 client_http.py
+```
+
+### Benchmark using Client-Server Mode
+```sh
+num_task=2
+python3 client_grpc.py --num-tasks $num_task --huggingface-dataset yuekai/seed_tts --split-name wenetspeech4tts
+```
+
+### Benchmark using Offline TRT-LLM Mode
+```sh
+batch_size=1
+split_name=wenetspeech4tts
+backend_type=trt
+log_dir=./log_benchmark_batch_size_${batch_size}_${split_name}_${backend_type}
+rm -r $log_dir
+ln -s model_repo_f5_tts/f5_tts/1/f5_tts_trtllm.py ./
+torchrun --nproc_per_node=1 \
+benchmark.py --output-dir $log_dir \
+--batch-size $batch_size \
+--enable-warmup \
+--split-name $split_name \
+--model-path $F5_TTS_HF_DOWNLOAD_PATH/$model/model_1200000.pt \
+--vocab-file $F5_TTS_HF_DOWNLOAD_PATH/$model/vocab.txt \
+--vocoder-trt-engine-path $vocoder_trt_engine_path \
+--backend-type $backend_type \
+--tllm-model-dir $F5_TTS_TRT_LLM_ENGINE_PATH || exit 1
+```
+
+### Benchmark Results
+Decoding on a single L20 GPU, using 26 different prompt_audio & target_text pairs, 16 NFE.
+
+| Model               | Concurrency    | Avg Latency | RTF    | Mode            |
+|---------------------|----------------|-------------|--------|-----------------|
+| F5-TTS Base (Vocos) | 2              | 253 ms      | 0.0394 | Client-Server   |
+| F5-TTS Base (Vocos) | 1 (Batch_size) | -           | 0.0402 | Offline TRT-LLM |
+| F5-TTS Base (Vocos) | 1 (Batch_size) | -           | 0.1467 | Offline Pytorch |
+
+### Credits
+1. [F5-TTS-TRTLLM](https://github.com/Bigfishering/f5-tts-trtllm)

src/f5_tts/runtime/triton_trtllm/benchmark.py

@@ -0,0 +1,560 @@
+# Copyright (c) 2024 Tsinghua Univ. (authors: Xingchen Song)
+#               2025               （authors: Yuekai Zhang）
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# Modified from https://github.com/xingchensong/S3Tokenizer/blob/main/s3tokenizer/cli.py
+""" Example Usage
+torchrun --nproc_per_node=1 \
+benchmark.py --output-dir $log_dir \
+--batch-size $batch_size \
+--enable-warmup \
+--split-name $split_name \
+--model-path $F5_TTS_HF_DOWNLOAD_PATH/$model/model_1200000.pt \
+--vocab-file $F5_TTS_HF_DOWNLOAD_PATH/$model/vocab.txt \
+--vocoder-trt-engine-path $vocoder_trt_engine_path \
+--backend-type $backend_type \
+--tllm-model-dir $F5_TTS_TRT_LLM_ENGINE_PATH || exit 1
+"""
+
+import argparse
+import json
+import os
+import time
+from typing import Dict, List, Union
+
+import datasets
+import jieba
+import tensorrt as trt
+import torch
+import torch.distributed as dist
+import torch.nn.functional as F
+import torchaudio
+from datasets import load_dataset
+from f5_tts_trtllm import F5TTS
+from huggingface_hub import hf_hub_download
+from pypinyin import Style, lazy_pinyin
+from tensorrt_llm._utils import trt_dtype_to_torch
+from tensorrt_llm.logger import logger
+from tensorrt_llm.runtime.session import Session, TensorInfo
+from torch.nn.utils.rnn import pad_sequence
+from torch.utils.data import DataLoader, DistributedSampler
+from tqdm import tqdm
+from vocos import Vocos
+
+
+torch.manual_seed(0)
+
+
+def get_args():
+    parser = argparse.ArgumentParser(description="extract speech code")
+    parser.add_argument(
+        "--split-name",
+        type=str,
+        default="wenetspeech4tts",
+        choices=["wenetspeech4tts", "test_zh", "test_en", "test_hard"],
+        help="huggingface dataset split name",
+    )
+    parser.add_argument("--output-dir", required=True, type=str, help="dir to save result")
+    parser.add_argument(
+        "--vocab-file",
+        required=True,
+        type=str,
+        help="vocab file",
+    )
+    parser.add_argument(
+        "--model-path",
+        required=True,
+        type=str,
+        help="model path, to load text embedding",
+    )
+    parser.add_argument(
+        "--tllm-model-dir",
+        required=True,
+        type=str,
+        help="tllm model dir",
+    )
+    parser.add_argument(
+        "--batch-size",
+        required=True,
+        type=int,
+        help="batch size (per-device) for inference",
+    )
+    parser.add_argument("--num-workers", type=int, default=0, help="workers for dataloader")
+    parser.add_argument("--prefetch", type=int, default=None, help="prefetch for dataloader")
+    parser.add_argument(
+        "--vocoder",
+        default="vocos",
+        type=str,
+        help="vocoder name",
+    )
+    parser.add_argument(
+        "--vocoder-trt-engine-path",
+        default=None,
+        type=str,
+        help="vocoder trt engine path",
+    )
+    parser.add_argument("--enable-warmup", action="store_true")
+    parser.add_argument("--remove-input-padding", action="store_true")
+    parser.add_argument("--use-perf", action="store_true", help="use nvtx to record performance")
+    parser.add_argument("--backend-type", type=str, default="triton", choices=["trt", "pytorch"], help="backend type")
+    args = parser.parse_args()
+    return args
+
+
+def padded_mel_batch(ref_mels, max_seq_len):
+    padded_ref_mels = []
+    for mel in ref_mels:
+        # pad along the last dimension
+        padded_ref_mel = F.pad(mel, (0, 0, 0, max_seq_len - mel.shape[0]), value=0)
+        padded_ref_mels.append(padded_ref_mel)
+    padded_ref_mels = torch.stack(padded_ref_mels)
+    return padded_ref_mels
+
+
+def data_collator(batch, vocab_char_map, device="cuda", use_perf=False):
+    if use_perf:
+        torch.cuda.nvtx.range_push("data_collator")
+    target_sample_rate = 24000
+    target_rms = 0.1
+    ids, ref_mel_list, ref_mel_len_list, estimated_reference_target_mel_len, reference_target_texts_list = (
+        [],
+        [],
+        [],
+        [],
+        [],
+    )
+    for i, item in enumerate(batch):
+        item_id, prompt_text, target_text = (
+            item["id"],
+            item["prompt_text"],
+            item["target_text"],
+        )
+        ids.append(item_id)
+        reference_target_texts_list.append(prompt_text + target_text)
+
+        ref_audio_org, ref_sr = (
+            item["prompt_audio"]["array"],
+            item["prompt_audio"]["sampling_rate"],
+        )
+        ref_audio_org = torch.from_numpy(ref_audio_org).unsqueeze(0).float()
+        ref_rms = torch.sqrt(torch.mean(torch.square(ref_audio_org)))
+        if ref_rms < target_rms:
+            ref_audio_org = ref_audio_org * target_rms / ref_rms
+
+        if ref_sr != target_sample_rate:
+            resampler = torchaudio.transforms.Resample(ref_sr, target_sample_rate)
+            ref_audio = resampler(ref_audio_org)
+        else:
+            ref_audio = ref_audio_org
+
+        if use_perf:
+            torch.cuda.nvtx.range_push(f"mel_spectrogram {i}")
+        ref_mel = mel_spectrogram(ref_audio, vocoder="vocos", device="cuda")
+        if use_perf:
+            torch.cuda.nvtx.range_pop()
+        ref_mel = ref_mel.squeeze()
+        ref_mel_len = ref_mel.shape[0]
+        assert ref_mel.shape[1] == 100
+
+        ref_mel_list.append(ref_mel)
+        ref_mel_len_list.append(ref_mel_len)
+
+        estimated_reference_target_mel_len.append(
+            int(ref_mel.shape[0] * (1 + len(target_text.encode("utf-8")) / len(prompt_text.encode("utf-8"))))
+        )
+
+    max_seq_len = max(estimated_reference_target_mel_len)
+    ref_mel_batch = padded_mel_batch(ref_mel_list, max_seq_len)
+    ref_mel_len_batch = torch.LongTensor(ref_mel_len_list)
+
+    pinyin_list = convert_char_to_pinyin(reference_target_texts_list, polyphone=True)
+    text_pad_sequence = list_str_to_idx(pinyin_list, vocab_char_map)
+
+    for i, item in enumerate(text_pad_sequence):
+        text_pad_sequence[i] = F.pad(
+            item, (0, estimated_reference_target_mel_len[i] - len(item)), mode="constant", value=-1
+        )
+        text_pad_sequence[i] += 1  # WAR: 0 is reserved for padding token, hard coding in F5-TTS
+    text_pad_sequence = pad_sequence(text_pad_sequence, padding_value=-1, batch_first=True).to(device)
+    text_pad_sequence = F.pad(
+        text_pad_sequence, (0, max_seq_len - text_pad_sequence.shape[1]), mode="constant", value=-1
+    )
+    if use_perf:
+        torch.cuda.nvtx.range_pop()
+    return {
+        "ids": ids,
+        "ref_mel_batch": ref_mel_batch,
+        "ref_mel_len_batch": ref_mel_len_batch,
+        "text_pad_sequence": text_pad_sequence,
+        "estimated_reference_target_mel_len": estimated_reference_target_mel_len,
+    }
+
+
+def init_distributed():
+    world_size = int(os.environ.get("WORLD_SIZE", 1))
+    local_rank = int(os.environ.get("LOCAL_RANK", 0))
+    rank = int(os.environ.get("RANK", 0))
+    print(
+        "Inference on multiple gpus, this gpu {}".format(local_rank)
+        + ", rank {}, world_size {}".format(rank, world_size)
+    )
+    torch.cuda.set_device(local_rank)
+    # Initialize process group with explicit device IDs
+    dist.init_process_group(
+        "nccl",
+    )
+    return world_size, local_rank, rank
+
+
+def get_tokenizer(vocab_file_path: str):
+    """
+    tokenizer   - "pinyin" do g2p for only chinese characters, need .txt vocab_file
+                - "char" for char-wise tokenizer, need .txt vocab_file
+                - "byte" for utf-8 tokenizer
+                - "custom" if you're directly passing in a path to the vocab.txt you want to use
+    vocab_size  - if use "pinyin", all available pinyin types, common alphabets (also those with accent) and symbols
+                - if use "char", derived from unfiltered character & symbol counts of custom dataset
+                - if use "byte", set to 256 (unicode byte range)
+    """
+    with open(vocab_file_path, "r", encoding="utf-8") as f:
+        vocab_char_map = {}
+        for i, char in enumerate(f):
+            vocab_char_map[char[:-1]] = i
+    vocab_size = len(vocab_char_map)
+    return vocab_char_map, vocab_size
+
+
+def convert_char_to_pinyin(reference_target_texts_list, polyphone=True):
+    final_reference_target_texts_list = []
+    custom_trans = str.maketrans(
+        {";": ",", "“": '"', "”": '"', "‘": "'", "’": "'"}
+    )  # add custom trans here, to address oov
+
+    def is_chinese(c):
+        return "\u3100" <= c <= "\u9fff"  # common chinese characters
+
+    for text in reference_target_texts_list:
+        char_list = []
+        text = text.translate(custom_trans)
+        for seg in jieba.cut(text):
+            seg_byte_len = len(bytes(seg, "UTF-8"))
+            if seg_byte_len == len(seg):  # if pure alphabets and symbols
+                if char_list and seg_byte_len > 1 and char_list[-1] not in " :'\"":
+                    char_list.append(" ")
+                char_list.extend(seg)
+            elif polyphone and seg_byte_len == 3 * len(seg):  # if pure east asian characters
+                seg_ = lazy_pinyin(seg, style=Style.TONE3, tone_sandhi=True)
+                for i, c in enumerate(seg):
+                    if is_chinese(c):
+                        char_list.append(" ")
+                    char_list.append(seg_[i])
+            else:  # if mixed characters, alphabets and symbols
+                for c in seg:
+                    if ord(c) < 256:
+                        char_list.extend(c)
+                    elif is_chinese(c):
+                        char_list.append(" ")
+                        char_list.extend(lazy_pinyin(c, style=Style.TONE3, tone_sandhi=True))
+                    else:
+                        char_list.append(c)
+        final_reference_target_texts_list.append(char_list)
+
+    return final_reference_target_texts_list
+
+
+def list_str_to_idx(
+    text: Union[List[str], List[List[str]]],
+    vocab_char_map: Dict[str, int],  # {char: idx}
+    padding_value=-1,
+):
+    list_idx_tensors = [torch.tensor([vocab_char_map.get(c, 0) for c in t]) for t in text]  # pinyin or char style
+    # text = pad_sequence(list_idx_tensors, padding_value=padding_value, batch_first=True)
+    return list_idx_tensors
+
+
+def load_vocoder(
+    vocoder_name="vocos", is_local=False, local_path="", device="cuda", hf_cache_dir=None, vocoder_trt_engine_path=None
+):
+    if vocoder_name == "vocos":
+        if vocoder_trt_engine_path is not None:
+            vocoder = VocosTensorRT(engine_path=vocoder_trt_engine_path)
+        else:
+            # vocoder = Vocos.from_pretrained("charactr/vocos-mel-24khz").to(device)
+            if is_local:
+                print(f"Load vocos from local path {local_path}")
+                config_path = f"{local_path}/config.yaml"
+                model_path = f"{local_path}/pytorch_model.bin"
+            else:
+                print("Download Vocos from huggingface charactr/vocos-mel-24khz")
+                repo_id = "charactr/vocos-mel-24khz"
+                config_path = hf_hub_download(repo_id=repo_id, cache_dir=hf_cache_dir, filename="config.yaml")
+                model_path = hf_hub_download(repo_id=repo_id, cache_dir=hf_cache_dir, filename="pytorch_model.bin")
+            vocoder = Vocos.from_hparams(config_path)
+            state_dict = torch.load(model_path, map_location="cpu", weights_only=True)
+            from vocos.feature_extractors import EncodecFeatures
+
+            if isinstance(vocoder.feature_extractor, EncodecFeatures):
+                encodec_parameters = {
+                    "feature_extractor.encodec." + key: value
+                    for key, value in vocoder.feature_extractor.encodec.state_dict().items()
+                }
+                state_dict.update(encodec_parameters)
+            vocoder.load_state_dict(state_dict)
+            vocoder = vocoder.eval().to(device)
+    elif vocoder_name == "bigvgan":
+        raise NotImplementedError("BigVGAN is not implemented yet")
+    return vocoder
+
+
+def mel_spectrogram(waveform, vocoder="vocos", device="cuda"):
+    if vocoder == "vocos":
+        mel_stft = torchaudio.transforms.MelSpectrogram(
+            sample_rate=24000,
+            n_fft=1024,
+            win_length=1024,
+            hop_length=256,
+            n_mels=100,
+            power=1,
+            center=True,
+            normalized=False,
+            norm=None,
+        ).to(device)
+    mel = mel_stft(waveform.to(device))
+    mel = mel.clamp(min=1e-5).log()
+    return mel.transpose(1, 2)
+
+
+class VocosTensorRT:
+    def __init__(self, engine_path="./vocos_vocoder.plan", stream=None):
+        TRT_LOGGER = trt.Logger(trt.Logger.WARNING)
+        trt.init_libnvinfer_plugins(TRT_LOGGER, namespace="")
+        logger.info(f"Loading vae engine from {engine_path}")
+        self.engine_path = engine_path
+        with open(engine_path, "rb") as f:
+            engine_buffer = f.read()
+        self.session = Session.from_serialized_engine(engine_buffer)
+        self.stream = stream if stream is not None else torch.cuda.current_stream().cuda_stream
+
+    def decode(self, mels):
+        mels = mels.contiguous()
+        inputs = {"mel": mels}
+        output_info = self.session.infer_shapes([TensorInfo("mel", trt.DataType.FLOAT, mels.shape)])
+        outputs = {
+            t.name: torch.empty(tuple(t.shape), dtype=trt_dtype_to_torch(t.dtype), device="cuda") for t in output_info
+        }
+        ok = self.session.run(inputs, outputs, self.stream)
+
+        assert ok, "Runtime execution failed for vae session"
+
+        samples = outputs["waveform"]
+        return samples
+
+
+def main():
+    args = get_args()
+    os.makedirs(args.output_dir, exist_ok=True)
+
+    assert torch.cuda.is_available()
+    world_size, local_rank, rank = init_distributed()
+    device = torch.device(f"cuda:{local_rank}")
+
+    vocab_char_map, vocab_size = get_tokenizer(args.vocab_file)
+
+    tllm_model_dir = args.tllm_model_dir
+    config_file = os.path.join(tllm_model_dir, "config.json")
+    with open(config_file) as f:
+        config = json.load(f)
+    if args.backend_type == "trt":
+        model = F5TTS(
+            config, debug_mode=False, tllm_model_dir=tllm_model_dir, model_path=args.model_path, vocab_size=vocab_size
+        )
+    elif args.backend_type == "pytorch":
+        import sys
+
+        sys.path.append(f"{os.path.dirname(os.path.abspath(__file__))}/../../../../src/")
+        from f5_tts.infer.utils_infer import load_model
+        from f5_tts.model import DiT
+
+        F5TTS_model_cfg = dict(
+            dim=1024,
+            depth=22,
+            heads=16,
+            ff_mult=2,
+            text_dim=512,
+            conv_layers=4,
+            pe_attn_head=1,
+            text_mask_padding=False,
+        )
+        model = load_model(DiT, F5TTS_model_cfg, args.model_path)
+
+    vocoder = load_vocoder(
+        vocoder_name=args.vocoder, device=device, vocoder_trt_engine_path=args.vocoder_trt_engine_path
+    )
+
+    dataset = load_dataset(
+        "yuekai/seed_tts",
+        split=args.split_name,
+        trust_remote_code=True,
+    )
+
+    def add_estimated_duration(example):
+        prompt_audio_len = example["prompt_audio"]["array"].shape[0]
+        scale_factor = 1 + len(example["target_text"]) / len(example["prompt_text"])
+        estimated_duration = prompt_audio_len * scale_factor
+        example["estimated_duration"] = estimated_duration / example["prompt_audio"]["sampling_rate"]
+        return example
+
+    dataset = dataset.map(add_estimated_duration)
+    dataset = dataset.sort("estimated_duration", reverse=True)
+    if args.use_perf:
+        # dataset_list = [dataset.select(range(1)) for i in range(16)]  # seq_len 1000
+        dataset_list_short = [dataset.select([24]) for i in range(8)]  # seq_len 719
+        # dataset_list_long = [dataset.select([23]) for i in range(8)] # seq_len 2002
+        # dataset = datasets.concatenate_datasets(dataset_list_short + dataset_list_long)
+        dataset = datasets.concatenate_datasets(dataset_list_short)
+    if world_size > 1:
+        sampler = DistributedSampler(dataset, num_replicas=world_size, rank=rank)
+    else:
+        # This would disable shuffling
+        sampler = None
+
+    dataloader = DataLoader(
+        dataset,
+        batch_size=args.batch_size,
+        sampler=sampler,
+        shuffle=False,
+        num_workers=args.num_workers,
+        prefetch_factor=args.prefetch,
+        collate_fn=lambda x: data_collator(x, vocab_char_map, use_perf=args.use_perf),
+    )
+
+    total_steps = len(dataset)
+
+    if args.enable_warmup:
+        for batch in dataloader:
+            ref_mels, ref_mel_lens = batch["ref_mel_batch"].to(device), batch["ref_mel_len_batch"].to(device)
+            text_pad_seq = batch["text_pad_sequence"].to(device)
+            total_mel_lens = batch["estimated_reference_target_mel_len"]
+            if args.backend_type == "trt":
+                _ = model.sample(
+                    text_pad_seq, ref_mels, ref_mel_lens, total_mel_lens, remove_input_padding=args.remove_input_padding
+                )
+            elif args.backend_type == "pytorch":
+                with torch.inference_mode():
+                    text_pad_seq -= 1
+                    text_pad_seq[text_pad_seq == -2] = -1
+                    total_mel_lens = torch.tensor(total_mel_lens, device=device)
+                    generated, _ = model.sample(
+                        cond=ref_mels,
+                        text=text_pad_seq,
+                        duration=total_mel_lens,
+                        steps=16,
+                        cfg_strength=2.0,
+                        sway_sampling_coef=-1,
+                    )
+
+    if rank == 0:
+        progress_bar = tqdm(total=total_steps, desc="Processing", unit="wavs")
+
+    decoding_time = 0
+    vocoder_time = 0
+    total_duration = 0
+    if args.use_perf:
+        torch.cuda.cudart().cudaProfilerStart()
+    total_decoding_time = time.time()
+    for batch in dataloader:
+        if args.use_perf:
+            torch.cuda.nvtx.range_push("data sample")
+        ref_mels, ref_mel_lens = batch["ref_mel_batch"].to(device), batch["ref_mel_len_batch"].to(device)
+        text_pad_seq = batch["text_pad_sequence"].to(device)
+        total_mel_lens = batch["estimated_reference_target_mel_len"]
+
+        if args.use_perf:
+            torch.cuda.nvtx.range_pop()
+        if args.backend_type == "trt":
+            generated, cost_time = model.sample(
+                text_pad_seq,
+                ref_mels,
+                ref_mel_lens,
+                total_mel_lens,
+                remove_input_padding=args.remove_input_padding,
+                use_perf=args.use_perf,
+            )
+        elif args.backend_type == "pytorch":
+            total_mel_lens = torch.tensor(total_mel_lens, device=device)
+            with torch.inference_mode():
+                start_time = time.time()
+                text_pad_seq -= 1
+                text_pad_seq[text_pad_seq == -2] = -1
+                generated, _ = model.sample(
+                    cond=ref_mels,
+                    text=text_pad_seq,
+                    duration=total_mel_lens,
+                    lens=ref_mel_lens,
+                    steps=16,
+                    cfg_strength=2.0,
+                    sway_sampling_coef=-1,
+                )
+                cost_time = time.time() - start_time
+        decoding_time += cost_time
+        vocoder_start_time = time.time()
+        for i, gen in enumerate(generated):
+            gen = gen[ref_mel_lens[i] : total_mel_lens[i], :].unsqueeze(0)
+            gen_mel_spec = gen.permute(0, 2, 1).to(torch.float32)
+            if args.vocoder == "vocos":
+                if args.use_perf:
+                    torch.cuda.nvtx.range_push("vocoder decode")
+                generated_wave = vocoder.decode(gen_mel_spec).cpu()
+                if args.use_perf:
+                    torch.cuda.nvtx.range_pop()
+            else:
+                generated_wave = vocoder(gen_mel_spec).squeeze(0).cpu()
+            target_rms = 0.1
+            target_sample_rate = 24_000
+            # if ref_rms_list[i] < target_rms:
+            #     generated_wave = generated_wave * ref_rms_list[i] / target_rms
+            rms = torch.sqrt(torch.mean(torch.square(generated_wave)))
+            if rms < target_rms:
+                generated_wave = generated_wave * target_rms / rms
+            utt = batch["ids"][i]
+            torchaudio.save(
+                f"{args.output_dir}/{utt}.wav",
+                generated_wave,
+                target_sample_rate,
+            )
+            total_duration += generated_wave.shape[1] / target_sample_rate
+        vocoder_time += time.time() - vocoder_start_time
+        if rank == 0:
+            progress_bar.update(world_size * len(batch["ids"]))
+    total_decoding_time = time.time() - total_decoding_time
+    if rank == 0:
+        progress_bar.close()
+    rtf = total_decoding_time / total_duration
+    s = f"RTF: {rtf:.4f}\n"
+    s += f"total_duration: {total_duration:.3f} seconds\n"
+    s += f"({total_duration / 3600:.2f} hours)\n"
+    s += f"DiT time: {decoding_time:.3f} seconds ({decoding_time / 3600:.2f} hours)\n"
+    s += f"Vocoder time: {vocoder_time:.3f} seconds ({vocoder_time / 3600:.2f} hours)\n"
+    s += f"total decoding time: {total_decoding_time:.3f} seconds ({total_decoding_time / 3600:.2f} hours)\n"
+    s += f"batch size: {args.batch_size}\n"
+    print(s)
+
+    with open(f"{args.output_dir}/rtf.txt", "w") as f:
+        f.write(s)
+
+    dist.barrier()
+    dist.destroy_process_group()
+
+
+if __name__ == "__main__":
+    main()

src/f5_tts/runtime/triton_trtllm/client_grpc.py

@@ -0,0 +1,470 @@
+#!/usr/bin/env python3
+# Copyright      2022  Xiaomi Corp.        (authors: Fangjun Kuang)
+#                2023  Nvidia              (authors: Yuekai Zhang)
+#                2023  Recurrent.ai        (authors: Songtao Shi)
+# See LICENSE for clarification regarding multiple authors
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""
+This script supports to load dataset from huggingface and sends it to the server
+for decoding, in parallel.
+
+Usage:
+num_task=2
+
+# For offline F5-TTS
+python3 client_grpc.py \
+    --server-addr localhost \
+    --model-name f5_tts \
+    --num-tasks $num_task \
+    --huggingface-dataset yuekai/seed_tts \
+    --split-name test_zh \
+    --log-dir ./log_concurrent_tasks_${num_task}
+
+# For offline Spark-TTS-0.5B
+python3 client_grpc.py \
+    --server-addr localhost \
+    --model-name spark_tts \
+    --num-tasks $num_task \
+    --huggingface-dataset yuekai/seed_tts \
+    --split-name wenetspeech4tts \
+    --log-dir ./log_concurrent_tasks_${num_task}
+"""
+
+import argparse
+import asyncio
+import json
+import os
+import time
+import types
+from pathlib import Path
+
+import numpy as np
+import soundfile as sf
+import tritonclient
+import tritonclient.grpc.aio as grpcclient
+from tritonclient.utils import np_to_triton_dtype
+
+
+def write_triton_stats(stats, summary_file):
+    with open(summary_file, "w") as summary_f:
+        model_stats = stats["model_stats"]
+        # write a note, the log is from triton_client.get_inference_statistics(), to better human readability
+        summary_f.write(
+            "The log is parsing from triton_client.get_inference_statistics(), to better human readability. \n"
+        )
+        summary_f.write("To learn more about the log, please refer to: \n")
+        summary_f.write("1. https://github.com/triton-inference-server/server/blob/main/docs/user_guide/metrics.md \n")
+        summary_f.write("2. https://github.com/triton-inference-server/server/issues/5374 \n\n")
+        summary_f.write(
+            "To better improve throughput, we always would like let requests wait in the queue for a while, and then execute them with a larger batch size. \n"
+        )
+        summary_f.write(
+            "However, there is a trade-off between the increased queue time and the increased batch size. \n"
+        )
+        summary_f.write(
+            "You may change 'max_queue_delay_microseconds' and 'preferred_batch_size' in the model configuration file to achieve this. \n"
+        )
+        summary_f.write(
+            "See https://github.com/triton-inference-server/server/blob/main/docs/user_guide/model_configuration.md#delayed-batching for more details. \n\n"
+        )
+        for model_state in model_stats:
+            if "last_inference" not in model_state:
+                continue
+            summary_f.write(f"model name is {model_state['name']} \n")
+            model_inference_stats = model_state["inference_stats"]
+            total_queue_time_s = int(model_inference_stats["queue"]["ns"]) / 1e9
+            total_infer_time_s = int(model_inference_stats["compute_infer"]["ns"]) / 1e9
+            total_input_time_s = int(model_inference_stats["compute_input"]["ns"]) / 1e9
+            total_output_time_s = int(model_inference_stats["compute_output"]["ns"]) / 1e9
+            summary_f.write(
+                f"queue time {total_queue_time_s:<5.2f} s, compute infer time {total_infer_time_s:<5.2f} s, compute input time {total_input_time_s:<5.2f} s, compute output time {total_output_time_s:<5.2f} s \n"  # noqa
+            )
+            model_batch_stats = model_state["batch_stats"]
+            for batch in model_batch_stats:
+                batch_size = int(batch["batch_size"])
+                compute_input = batch["compute_input"]
+                compute_output = batch["compute_output"]
+                compute_infer = batch["compute_infer"]
+                batch_count = int(compute_infer["count"])
+                assert compute_infer["count"] == compute_output["count"] == compute_input["count"]
+                compute_infer_time_ms = int(compute_infer["ns"]) / 1e6
+                compute_input_time_ms = int(compute_input["ns"]) / 1e6
+                compute_output_time_ms = int(compute_output["ns"]) / 1e6
+                summary_f.write(
+                    f"execuate inference with batch_size {batch_size:<2} total {batch_count:<5} times, total_infer_time {compute_infer_time_ms:<9.2f} ms, avg_infer_time {compute_infer_time_ms:<9.2f}/{batch_count:<5}={compute_infer_time_ms / batch_count:.2f} ms, avg_infer_time_per_sample {compute_infer_time_ms:<9.2f}/{batch_count:<5}/{batch_size}={compute_infer_time_ms / batch_count / batch_size:.2f} ms \n"  # noqa
+                )
+                summary_f.write(
+                    f"input {compute_input_time_ms:<9.2f} ms, avg {compute_input_time_ms / batch_count:.2f} ms, "  # noqa
+                )
+                summary_f.write(
+                    f"output {compute_output_time_ms:<9.2f} ms, avg {compute_output_time_ms / batch_count:.2f} ms \n"  # noqa
+                )
+
+
+def get_args():
+    parser = argparse.ArgumentParser(formatter_class=argparse.ArgumentDefaultsHelpFormatter)
+
+    parser.add_argument(
+        "--server-addr",
+        type=str,
+        default="localhost",
+        help="Address of the server",
+    )
+
+    parser.add_argument(
+        "--server-port",
+        type=int,
+        default=8001,
+        help="Grpc port of the triton server, default is 8001",
+    )
+
+    parser.add_argument(
+        "--reference-audio",
+        type=str,
+        default=None,
+        help="Path to a single audio file. It can't be specified at the same time with --manifest-dir",
+    )
+
+    parser.add_argument(
+        "--reference-text",
+        type=str,
+        default="",
+        help="",
+    )
+
+    parser.add_argument(
+        "--target-text",
+        type=str,
+        default="",
+        help="",
+    )
+
+    parser.add_argument(
+        "--huggingface-dataset",
+        type=str,
+        default="yuekai/seed_tts",
+        help="dataset name in huggingface dataset hub",
+    )
+
+    parser.add_argument(
+        "--split-name",
+        type=str,
+        default="wenetspeech4tts",
+        choices=["wenetspeech4tts", "test_zh", "test_en", "test_hard"],
+        help="dataset split name, default is 'test'",
+    )
+
+    parser.add_argument(
+        "--manifest-path",
+        type=str,
+        default=None,
+        help="Path to the manifest dir which includes wav.scp trans.txt files.",
+    )
+
+    parser.add_argument(
+        "--model-name",
+        type=str,
+        default="f5_tts",
+        choices=["f5_tts", "spark_tts"],
+        help="triton model_repo module name to request: transducer for k2, attention_rescoring for wenet offline, streaming_wenet for wenet streaming, infer_pipeline for paraformer large offline",
+    )
+
+    parser.add_argument(
+        "--num-tasks",
+        type=int,
+        default=1,
+        help="Number of concurrent tasks for sending",
+    )
+
+    parser.add_argument(
+        "--log-interval",
+        type=int,
+        default=5,
+        help="Controls how frequently we print the log.",
+    )
+
+    parser.add_argument(
+        "--compute-wer",
+        action="store_true",
+        default=False,
+        help="""True to compute WER.
+        """,
+    )
+
+    parser.add_argument(
+        "--log-dir",
+        type=str,
+        required=False,
+        default="./tmp",
+        help="log directory",
+    )
+
+    parser.add_argument(
+        "--batch-size",
+        type=int,
+        default=1,
+        help="Inference batch_size per request for offline mode.",
+    )
+
+    return parser.parse_args()
+
+
+def load_audio(wav_path, target_sample_rate=24000):
+    assert target_sample_rate == 24000, "hard coding in server"
+    if isinstance(wav_path, dict):
+        waveform = wav_path["array"]
+        sample_rate = wav_path["sampling_rate"]
+    else:
+        waveform, sample_rate = sf.read(wav_path)
+    if sample_rate != target_sample_rate:
+        from scipy.signal import resample
+
+        num_samples = int(len(waveform) * (target_sample_rate / sample_rate))
+        waveform = resample(waveform, num_samples)
+    return waveform, target_sample_rate
+
+
+async def send(
+    manifest_item_list: list,
+    name: str,
+    triton_client: tritonclient.grpc.aio.InferenceServerClient,
+    protocol_client: types.ModuleType,
+    log_interval: int,
+    model_name: str,
+    padding_duration: int = None,
+    audio_save_dir: str = "./",
+    save_sample_rate: int = 24000,
+):
+    total_duration = 0.0
+    latency_data = []
+    task_id = int(name[5:])
+
+    print(f"manifest_item_list: {manifest_item_list}")
+    for i, item in enumerate(manifest_item_list):
+        if i % log_interval == 0:
+            print(f"{name}: {i}/{len(manifest_item_list)}")
+        waveform, sample_rate = load_audio(item["audio_filepath"], target_sample_rate=24000)
+        duration = len(waveform) / sample_rate
+        lengths = np.array([[len(waveform)]], dtype=np.int32)
+
+        reference_text, target_text = item["reference_text"], item["target_text"]
+
+        estimated_target_duration = duration / len(reference_text) * len(target_text)
+
+        if padding_duration:
+            # padding to nearset 10 seconds
+            samples = np.zeros(
+                (
+                    1,
+                    padding_duration
+                    * sample_rate
+                    * ((int(estimated_target_duration + duration) // padding_duration) + 1),
+                ),
+                dtype=np.float32,
+            )
+
+            samples[0, : len(waveform)] = waveform
+        else:
+            samples = waveform
+
+        samples = samples.reshape(1, -1).astype(np.float32)
+
+        inputs = [
+            protocol_client.InferInput("reference_wav", samples.shape, np_to_triton_dtype(samples.dtype)),
+            protocol_client.InferInput("reference_wav_len", lengths.shape, np_to_triton_dtype(lengths.dtype)),
+            protocol_client.InferInput("reference_text", [1, 1], "BYTES"),
+            protocol_client.InferInput("target_text", [1, 1], "BYTES"),
+        ]
+        inputs[0].set_data_from_numpy(samples)
+        inputs[1].set_data_from_numpy(lengths)
+
+        input_data_numpy = np.array([reference_text], dtype=object)
+        input_data_numpy = input_data_numpy.reshape((1, 1))
+        inputs[2].set_data_from_numpy(input_data_numpy)
+
+        input_data_numpy = np.array([target_text], dtype=object)
+        input_data_numpy = input_data_numpy.reshape((1, 1))
+        inputs[3].set_data_from_numpy(input_data_numpy)
+
+        outputs = [protocol_client.InferRequestedOutput("waveform")]
+
+        sequence_id = 100000000 + i + task_id * 10
+        start = time.time()
+        response = await triton_client.infer(model_name, inputs, request_id=str(sequence_id), outputs=outputs)
+
+        audio = response.as_numpy("waveform").reshape(-1)
+
+        end = time.time() - start
+
+        audio_save_path = os.path.join(audio_save_dir, f"{item['target_audio_path']}.wav")
+        sf.write(audio_save_path, audio, save_sample_rate, "PCM_16")
+
+        actual_duration = len(audio) / save_sample_rate
+        latency_data.append((end, actual_duration))
+        total_duration += actual_duration
+
+    return total_duration, latency_data
+
+
+def load_manifests(manifest_path):
+    with open(manifest_path, "r") as f:
+        manifest_list = []
+        for line in f:
+            assert len(line.strip().split("|")) == 4
+            utt, prompt_text, prompt_wav, gt_text = line.strip().split("|")
+            utt = Path(utt).stem
+            # gt_wav = os.path.join(os.path.dirname(manifest_path), "wavs", utt + ".wav")
+            if not os.path.isabs(prompt_wav):
+                prompt_wav = os.path.join(os.path.dirname(manifest_path), prompt_wav)
+            manifest_list.append(
+                {
+                    "audio_filepath": prompt_wav,
+                    "reference_text": prompt_text,
+                    "target_text": gt_text,
+                    "target_audio_path": utt,
+                }
+            )
+    return manifest_list
+
+
+def split_data(data, k):
+    n = len(data)
+    if n < k:
+        print(f"Warning: the length of the input list ({n}) is less than k ({k}). Setting k to {n}.")
+        k = n
+
+    quotient = n // k
+    remainder = n % k
+
+    result = []
+    start = 0
+    for i in range(k):
+        if i < remainder:
+            end = start + quotient + 1
+        else:
+            end = start + quotient
+
+        result.append(data[start:end])
+        start = end
+
+    return result
+
+
+async def main():
+    args = get_args()
+    url = f"{args.server_addr}:{args.server_port}"
+
+    triton_client = grpcclient.InferenceServerClient(url=url, verbose=False)
+    protocol_client = grpcclient
+
+    if args.reference_audio:
+        args.num_tasks = 1
+        args.log_interval = 1
+        manifest_item_list = [
+            {
+                "reference_text": args.reference_text,
+                "target_text": args.target_text,
+                "audio_filepath": args.reference_audio,
+                "target_audio_path": "test",
+            }
+        ]
+    elif args.huggingface_dataset:
+        import datasets
+
+        dataset = datasets.load_dataset(
+            args.huggingface_dataset,
+            split=args.split_name,
+            trust_remote_code=True,
+        )
+        manifest_item_list = []
+        for i in range(len(dataset)):
+            manifest_item_list.append(
+                {
+                    "audio_filepath": dataset[i]["prompt_audio"],
+                    "reference_text": dataset[i]["prompt_text"],
+                    "target_audio_path": dataset[i]["id"],
+                    "target_text": dataset[i]["target_text"],
+                }
+            )
+    else:
+        manifest_item_list = load_manifests(args.manifest_path)
+
+    args.num_tasks = min(args.num_tasks, len(manifest_item_list))
+    manifest_item_list = split_data(manifest_item_list, args.num_tasks)
+
+    os.makedirs(args.log_dir, exist_ok=True)
+    tasks = []
+    start_time = time.time()
+    for i in range(args.num_tasks):
+        task = asyncio.create_task(
+            send(
+                manifest_item_list[i],
+                name=f"task-{i}",
+                triton_client=triton_client,
+                protocol_client=protocol_client,
+                log_interval=args.log_interval,
+                model_name=args.model_name,
+                audio_save_dir=args.log_dir,
+                padding_duration=1,
+                save_sample_rate=24000,
+            )
+        )
+        tasks.append(task)
+
+    ans_list = await asyncio.gather(*tasks)
+
+    end_time = time.time()
+    elapsed = end_time - start_time
+
+    total_duration = 0.0
+    latency_data = []
+    for ans in ans_list:
+        total_duration += ans[0]
+        latency_data += ans[1]
+
+    rtf = elapsed / total_duration
+
+    s = f"RTF: {rtf:.4f}\n"
+    s += f"total_duration: {total_duration:.3f} seconds\n"
+    s += f"({total_duration / 3600:.2f} hours)\n"
+    s += f"processing time: {elapsed:.3f} seconds ({elapsed / 3600:.2f} hours)\n"
+
+    latency_list = [chunk_end for (chunk_end, chunk_duration) in latency_data]
+    latency_ms = sum(latency_list) / float(len(latency_list)) * 1000.0
+    latency_variance = np.var(latency_list, dtype=np.float64) * 1000.0
+    s += f"latency_variance: {latency_variance:.2f}\n"
+    s += f"latency_50_percentile_ms: {np.percentile(latency_list, 50) * 1000.0:.2f}\n"
+    s += f"latency_90_percentile_ms: {np.percentile(latency_list, 90) * 1000.0:.2f}\n"
+    s += f"latency_95_percentile_ms: {np.percentile(latency_list, 95) * 1000.0:.2f}\n"
+    s += f"latency_99_percentile_ms: {np.percentile(latency_list, 99) * 1000.0:.2f}\n"
+    s += f"average_latency_ms: {latency_ms:.2f}\n"
+
+    print(s)
+    if args.manifest_path:
+        name = Path(args.manifest_path).stem
+    elif args.split_name:
+        name = args.split_name
+    with open(f"{args.log_dir}/rtf-{name}.txt", "w") as f:
+        f.write(s)
+
+    stats = await triton_client.get_inference_statistics(model_name="", as_json=True)
+    write_triton_stats(stats, f"{args.log_dir}/stats_summary-{name}.txt")
+
+    metadata = await triton_client.get_model_config(model_name=args.model_name, as_json=True)
+    with open(f"{args.log_dir}/model_config-{name}.json", "w") as f:
+        json.dump(metadata, f, indent=4)
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

src/f5_tts/runtime/triton_trtllm/client_http.py

@@ -0,0 +1,143 @@
+# Copyright 2025, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions
+# are met:
+#  * Redistributions of source code must retain the above copyright
+#    notice, this list of conditions and the following disclaimer.
+#  * Redistributions in binary form must reproduce the above copyright
+#    notice, this list of conditions and the following disclaimer in the
+#    documentation and/or other materials provided with the distribution.
+#  * Neither the name of NVIDIA CORPORATION nor the names of its
+#    contributors may be used to endorse or promote products derived
+#    from this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
+# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+# PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT OWNER OR
+# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+# PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
+# OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+import argparse
+
+import numpy as np
+import requests
+import soundfile as sf
+
+
+def get_args():
+    parser = argparse.ArgumentParser(formatter_class=argparse.ArgumentDefaultsHelpFormatter)
+
+    parser.add_argument(
+        "--server-url",
+        type=str,
+        default="localhost:8000",
+        help="Address of the server",
+    )
+
+    parser.add_argument(
+        "--reference-audio",
+        type=str,
+        default="../../infer/examples/basic/basic_ref_en.wav",
+        help="Path to a single audio file. It can't be specified at the same time with --manifest-dir",
+    )
+
+    parser.add_argument(
+        "--reference-text",
+        type=str,
+        default="Some call me nature, others call me mother nature.",
+        help="",
+    )
+
+    parser.add_argument(
+        "--target-text",
+        type=str,
+        default="I don't really care what you call me. I've been a silent spectator, watching species evolve, empires rise and fall. But always remember, I am mighty and enduring.",
+        help="",
+    )
+
+    parser.add_argument(
+        "--model-name",
+        type=str,
+        default="f5_tts",
+        choices=["f5_tts", "spark_tts"],
+        help="triton model_repo module name to request",
+    )
+
+    parser.add_argument(
+        "--output-audio",
+        type=str,
+        default="output.wav",
+        help="Path to save the output audio",
+    )
+    return parser.parse_args()
+
+
+def prepare_request(
+    samples,
+    reference_text,
+    target_text,
+    sample_rate=24000,
+    audio_save_dir: str = "./",
+):
+    assert len(samples.shape) == 1, "samples should be 1D"
+    lengths = np.array([[len(samples)]], dtype=np.int32)
+    samples = samples.reshape(1, -1).astype(np.float32)
+
+    data = {
+        "inputs": [
+            {"name": "reference_wav", "shape": samples.shape, "datatype": "FP32", "data": samples.tolist()},
+            {
+                "name": "reference_wav_len",
+                "shape": lengths.shape,
+                "datatype": "INT32",
+                "data": lengths.tolist(),
+            },
+            {"name": "reference_text", "shape": [1, 1], "datatype": "BYTES", "data": [reference_text]},
+            {"name": "target_text", "shape": [1, 1], "datatype": "BYTES", "data": [target_text]},
+        ]
+    }
+
+    return data
+
+
+def load_audio(wav_path, target_sample_rate=24000):
+    assert target_sample_rate == 24000, "hard coding in server"
+    if isinstance(wav_path, dict):
+        samples = wav_path["array"]
+        sample_rate = wav_path["sampling_rate"]
+    else:
+        samples, sample_rate = sf.read(wav_path)
+    if sample_rate != target_sample_rate:
+        from scipy.signal import resample
+
+        num_samples = int(len(samples) * (target_sample_rate / sample_rate))
+        samples = resample(samples, num_samples)
+    return samples, target_sample_rate
+
+
+if __name__ == "__main__":
+    args = get_args()
+    server_url = args.server_url
+    if not server_url.startswith(("http://", "https://")):
+        server_url = f"http://{server_url}"
+
+    url = f"{server_url}/v2/models/{args.model_name}/infer"
+    samples, sr = load_audio(args.reference_audio)
+    assert sr == 24000, "sample rate hardcoded in server"
+
+    samples = np.array(samples, dtype=np.float32)
+    data = prepare_request(samples, args.reference_text, args.target_text)
+
+    rsp = requests.post(
+        url, headers={"Content-Type": "application/json"}, json=data, verify=False, params={"request_id": "0"}
+    )
+    result = rsp.json()
+    audio = result["outputs"][0]["data"]
+    audio = np.array(audio, dtype=np.float32)
+    sf.write(args.output_audio, audio, 24000, "PCM_16")

src/f5_tts/runtime/triton_trtllm/docker-compose.yml

@@ -0,0 +1,20 @@
+services:
+  tts:
+    image: soar97/triton-f5-tts:24.12
+    shm_size: '1gb'
+    ports:
+      - "8000:8000"
+      - "8001:8001"
+      - "8002:8002"
+    environment:
+      - PYTHONIOENCODING=utf-8
+      - MODEL_ID=${MODEL_ID}
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              device_ids: ['0']
+              capabilities: [gpu]
+    command: >
+      /bin/bash -c "pip install vocos && rm -rf F5-TTS && git clone https://github.com/SWivid/F5-TTS.git && cd F5-TTS/src/f5_tts/runtime/triton_trtllm/ && bash run.sh 0 4 $MODEL"

src/f5_tts/runtime/triton_trtllm/model_repo_f5_tts/f5_tts/1/f5_tts_trtllm.py

@@ -0,0 +1,430 @@
+import math
+import os
+import time
+from functools import wraps
+from typing import List, Optional
+
+import tensorrt as trt
+import tensorrt_llm
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from tensorrt_llm._utils import str_dtype_to_torch, trt_dtype_to_torch
+from tensorrt_llm.logger import logger
+from tensorrt_llm.runtime.session import Session
+
+
+def remove_tensor_padding(input_tensor, input_tensor_lengths=None):
+    # Audio tensor case: batch, seq_len, feature_len
+    # position_ids case: batch, seq_len
+    assert input_tensor_lengths is not None, "input_tensor_lengths must be provided for 3D input_tensor"
+
+    # Initialize a list to collect valid sequences
+    valid_sequences = []
+
+    for i in range(input_tensor.shape[0]):
+        valid_length = input_tensor_lengths[i]
+        valid_sequences.append(input_tensor[i, :valid_length])
+
+    # Concatenate all valid sequences along the batch dimension
+    output_tensor = torch.cat(valid_sequences, dim=0).contiguous()
+    return output_tensor
+
+
+class TextEmbedding(nn.Module):
+    def __init__(self, text_num_embeds, text_dim, conv_layers=0, conv_mult=2, precompute_max_pos=4096):
+        super().__init__()
+        self.text_embed = nn.Embedding(text_num_embeds + 1, text_dim)  # use 0 as filler token
+        self.register_buffer("freqs_cis", precompute_freqs_cis(text_dim, precompute_max_pos), persistent=False)
+        self.text_blocks = nn.Sequential(*[ConvNeXtV2Block(text_dim, text_dim * conv_mult) for _ in range(conv_layers)])
+
+    def forward(self, text):
+        # only keep tensors with value not -1
+        text_mask = text != -1
+        text_pad_cut_off_index = text_mask.sum(dim=1).max()
+
+        text = text[:, :text_pad_cut_off_index]
+        text = self.text_embed(text)
+        text = text + self.freqs_cis[: text.shape[1], :]
+        for block in self.text_blocks:
+            text = block(text)
+        # padding text to the original length
+        # text shape: B,seq_len,C
+        # pad at the second dimension
+        text = F.pad(text, (0, 0, 0, text_mask.shape[1] - text.shape[1], 0, 0), value=0)
+        return text
+
+
+class GRN(nn.Module):
+    def __init__(self, dim):
+        super().__init__()
+        self.gamma = nn.Parameter(torch.zeros(1, 1, dim))
+        self.beta = nn.Parameter(torch.zeros(1, 1, dim))
+
+    def forward(self, x):
+        Gx = torch.norm(x, p=2, dim=1, keepdim=True)
+        Nx = Gx / (Gx.mean(dim=-1, keepdim=True) + 1e-6)
+        return self.gamma * (x * Nx) + self.beta + x
+
+
+class ConvNeXtV2Block(nn.Module):
+    def __init__(
+        self,
+        dim: int,
+        intermediate_dim: int,
+        dilation: int = 1,
+    ):
+        super().__init__()
+        padding = (dilation * (7 - 1)) // 2
+        self.dwconv = nn.Conv1d(
+            dim, dim, kernel_size=7, padding=padding, groups=dim, dilation=dilation
+        )  # depthwise conv
+        self.norm = nn.LayerNorm(dim, eps=1e-6)
+        self.pwconv1 = nn.Linear(dim, intermediate_dim)  # pointwise/1x1 convs, implemented with linear layers
+        self.act = nn.GELU()
+        self.grn = GRN(intermediate_dim)
+        self.pwconv2 = nn.Linear(intermediate_dim, dim)
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        residual = x
+        x = x.transpose(1, 2)  # b n d -> b d n
+        x = self.dwconv(x)
+        x = x.transpose(1, 2)  # b d n -> b n d
+        x = self.norm(x)
+        x = self.pwconv1(x)
+        x = self.act(x)
+        x = self.grn(x)
+        x = self.pwconv2(x)
+        return residual + x
+
+
+def precompute_freqs_cis(dim: int, end: int, theta: float = 10000.0, theta_rescale_factor=1.0):
+    # proposed by reddit user bloc97, to rescale rotary embeddings to longer sequence length without fine-tuning
+    # has some connection to NTK literature
+    # https://www.reddit.com/r/LocalLLaMA/comments/14lz7j5/ntkaware_scaled_rope_allows_llama_models_to_have/
+    # https://github.com/lucidrains/rotary-embedding-torch/blob/main/rotary_embedding_torch/rotary_embedding_torch.py
+    theta *= theta_rescale_factor ** (dim / (dim - 2))
+    freqs = 1.0 / (theta ** (torch.arange(0, dim, 2)[: (dim // 2)].float() / dim))
+    t = torch.arange(end, device=freqs.device)  # type: ignore
+    freqs = torch.outer(t, freqs).float()  # type: ignore
+    freqs_cos = torch.cos(freqs)  # real part
+    freqs_sin = torch.sin(freqs)  # imaginary part
+    return torch.cat([freqs_cos, freqs_sin], dim=-1)
+
+
+def load_checkpoint(ckpt_path, use_ema=True):
+    checkpoint = torch.load(ckpt_path, weights_only=True)
+    if use_ema:
+        checkpoint["model_state_dict"] = {
+            k.replace("ema_model.", ""): v
+            for k, v in checkpoint["ema_model_state_dict"].items()
+            if k not in ["initted", "step"]
+        }
+    dict_state = checkpoint["model_state_dict"]
+    text_embed_dict = {}
+    for key in dict_state.keys():
+        # transformer.text_embed.text_embed.weight -> text_embed.weight
+        if "text_embed" in key:
+            text_embed_dict[key.replace("transformer.text_embed.", "")] = dict_state[key]
+    return text_embed_dict
+
+
+class F5TTS(object):
+    def __init__(
+        self,
+        config,
+        debug_mode=True,
+        stream: Optional[torch.cuda.Stream] = None,
+        tllm_model_dir: Optional[str] = None,
+        model_path: Optional[str] = None,
+        vocab_size: Optional[int] = None,
+    ):
+        self.dtype = config["pretrained_config"]["dtype"]
+
+        rank = tensorrt_llm.mpi_rank()
+        world_size = config["pretrained_config"]["mapping"]["world_size"]
+        cp_size = config["pretrained_config"]["mapping"]["cp_size"]
+        tp_size = config["pretrained_config"]["mapping"]["tp_size"]
+        pp_size = config["pretrained_config"]["mapping"]["pp_size"]
+        assert pp_size == 1
+        self.mapping = tensorrt_llm.Mapping(
+            world_size=world_size, rank=rank, cp_size=cp_size, tp_size=tp_size, pp_size=1, gpus_per_node=1
+        )
+
+        local_rank = rank % self.mapping.gpus_per_node
+        self.device = torch.device(f"cuda:{local_rank}")
+
+        torch.cuda.set_device(self.device)
+
+        self.stream = stream
+        if self.stream is None:
+            self.stream = torch.cuda.Stream(self.device)
+        torch.cuda.set_stream(self.stream)
+
+        engine_file = os.path.join(tllm_model_dir, f"rank{rank}.engine")
+        logger.info(f"Loading engine from {engine_file}")
+        with open(engine_file, "rb") as f:
+            engine_buffer = f.read()
+
+        assert engine_buffer is not None
+
+        self.session = Session.from_serialized_engine(engine_buffer)
+
+        self.debug_mode = debug_mode
+
+        self.inputs = {}
+        self.outputs = {}
+        self.buffer_allocated = False
+
+        expected_tensor_names = ["noise", "cond", "time", "rope_cos", "rope_sin", "input_lengths", "denoised"]
+
+        found_tensor_names = [self.session.engine.get_tensor_name(i) for i in range(self.session.engine.num_io_tensors)]
+        if not self.debug_mode and set(expected_tensor_names) != set(found_tensor_names):
+            logger.error(
+                f"The following expected tensors are not found: {set(expected_tensor_names).difference(set(found_tensor_names))}"
+            )
+            logger.error(
+                f"Those tensors in engine are not expected: {set(found_tensor_names).difference(set(expected_tensor_names))}"
+            )
+            logger.error(f"Expected tensor names: {expected_tensor_names}")
+            logger.error(f"Found tensor names: {found_tensor_names}")
+            raise RuntimeError("Tensor names in engine are not the same as expected.")
+        if self.debug_mode:
+            self.debug_tensors = list(set(found_tensor_names) - set(expected_tensor_names))
+
+        self.max_mel_len = 4096
+        self.text_embedding = TextEmbedding(
+            text_num_embeds=vocab_size, text_dim=512, conv_layers=4, precompute_max_pos=self.max_mel_len
+        ).to(self.device)
+        self.text_embedding.load_state_dict(load_checkpoint(model_path), strict=True)
+
+        self.target_audio_sample_rate = 24000
+        self.target_rms = 0.15  # target rms for audio
+        self.n_fft = 1024
+        self.win_length = 1024
+        self.hop_length = 256
+        self.n_mel_channels = 100
+        # self.max_mel_len = 3000
+        self.head_dim = 64
+        self.base_rescale_factor = 1.0
+        self.interpolation_factor = 1.0
+        base = 10000.0 * self.base_rescale_factor ** (self.head_dim / (self.head_dim - 2))
+        inv_freq = 1.0 / (base ** (torch.arange(0, self.head_dim, 2).float() / self.head_dim))
+        freqs = torch.outer(torch.arange(self.max_mel_len, dtype=torch.float32), inv_freq) / self.interpolation_factor
+        self.freqs = freqs.repeat_interleave(2, dim=-1).unsqueeze(0)
+        self.rope_cos = self.freqs.cos().half()
+        self.rope_sin = self.freqs.sin().half()
+        self.nfe_steps = 16
+        t = torch.linspace(0, 1, self.nfe_steps + 1, dtype=torch.float32)
+        time_step = t + (-1.0) * (torch.cos(torch.pi * 0.5 * t) - 1 + t)
+        delta_t = torch.diff(time_step)
+        # WAR: hard coding 256 here
+        tmp_dim = 256
+        time_expand = torch.zeros((1, self.nfe_steps, tmp_dim), dtype=torch.float32)
+        half_dim = tmp_dim // 2
+        emb_factor = math.log(10000) / (half_dim - 1)
+        emb_factor = 1000.0 * torch.exp(torch.arange(half_dim, dtype=torch.float32) * -emb_factor)
+        for i in range(self.nfe_steps):
+            emb = time_step[i] * emb_factor
+            time_expand[:, i, :] = torch.cat((emb.sin(), emb.cos()), dim=-1)
+        self.time_expand = time_expand.to(self.device)
+        self.delta_t = torch.cat((delta_t, delta_t), dim=0).contiguous().to(self.device)
+
+    def _tensor_dtype(self, name):
+        # return torch dtype given tensor name for convenience
+        dtype = trt_dtype_to_torch(self.session.engine.get_tensor_dtype(name))
+        return dtype
+
+    def _setup(self, batch_size, seq_len):
+        for i in range(self.session.engine.num_io_tensors):
+            name = self.session.engine.get_tensor_name(i)
+            if self.session.engine.get_tensor_mode(name) == trt.TensorIOMode.OUTPUT:
+                shape = list(self.session.engine.get_tensor_shape(name))
+                shape[0] = batch_size
+                shape[1] = seq_len
+                self.outputs[name] = torch.empty(shape, dtype=self._tensor_dtype(name), device=self.device)
+
+        self.buffer_allocated = True
+
+    def cuda_stream_guard(func):
+        """Sync external stream and set current stream to the one bound to the session. Reset on exit."""
+
+        @wraps(func)
+        def wrapper(self, *args, **kwargs):
+            external_stream = torch.cuda.current_stream()
+            if external_stream != self.stream:
+                external_stream.synchronize()
+                torch.cuda.set_stream(self.stream)
+            ret = func(self, *args, **kwargs)
+            if external_stream != self.stream:
+                self.stream.synchronize()
+                torch.cuda.set_stream(external_stream)
+            return ret
+
+        return wrapper
+
+    @cuda_stream_guard
+    def forward(
+        self,
+        noise: torch.Tensor,
+        cond: torch.Tensor,
+        time_expand: torch.Tensor,
+        rope_cos: torch.Tensor,
+        rope_sin: torch.Tensor,
+        input_lengths: torch.Tensor,
+        delta_t: torch.Tensor,
+        use_perf: bool = False,
+    ):
+        if use_perf:
+            torch.cuda.nvtx.range_push("flow matching")
+        cfg_strength = 2.0
+        batch_size = noise.shape[0]
+        half_batch = batch_size // 2
+        noise_half = noise[:half_batch]  # Store the initial half of noise
+
+        input_type = str_dtype_to_torch(self.dtype)
+
+        # Keep a copy of the initial tensors
+        cond = cond.to(input_type)
+        rope_cos = rope_cos.to(input_type)
+        rope_sin = rope_sin.to(input_type)
+        input_lengths = input_lengths.to(str_dtype_to_torch("int32"))
+
+        # Instead of iteratively updating noise within a single model context,
+        # we'll do a single forward pass for each iteration with fresh context setup
+        for i in range(self.nfe_steps):
+            # Re-setup the buffers for clean execution
+            self._setup(batch_size, noise.shape[1])
+            if not self.buffer_allocated:
+                raise RuntimeError("Buffer not allocated, please call setup first!")
+
+            # Re-create combined noises for this iteration
+            current_noise = torch.cat([noise_half, noise_half], dim=0).to(input_type)
+
+            # Get time step for this iteration
+            current_time = time_expand[:, i].to(input_type)
+
+            # Create fresh input dictionary for this iteration
+            current_inputs = {
+                "noise": current_noise,
+                "cond": cond,
+                "time": current_time,
+                "rope_cos": rope_cos,
+                "rope_sin": rope_sin,
+                "input_lengths": input_lengths,
+            }
+
+            # Update inputs and set shapes
+            self.inputs.clear()  # Clear previous inputs
+            self.inputs.update(**current_inputs)
+            self.session.set_shapes(self.inputs)
+
+            if use_perf:
+                torch.cuda.nvtx.range_push(f"execute {i}")
+            ok = self.session.run(self.inputs, self.outputs, self.stream.cuda_stream)
+            assert ok, "Failed to execute model"
+            # self.session.context.execute_async_v3(self.stream.cuda_stream)
+            if use_perf:
+                torch.cuda.nvtx.range_pop()
+            # Process results
+            t_scale = delta_t[i].unsqueeze(0).to(input_type)
+
+            # Extract predictions
+            pred_cond = self.outputs["denoised"][:half_batch]
+            pred_uncond = self.outputs["denoised"][half_batch:]
+
+            # Apply classifier-free guidance with safeguards
+            guidance = pred_cond + (pred_cond - pred_uncond) * cfg_strength
+            # Calculate update for noise
+            noise_half = noise_half + guidance * t_scale
+        if use_perf:
+            torch.cuda.nvtx.range_pop()
+        return noise_half
+
+    def sample(
+        self,
+        text_pad_sequence: torch.Tensor,
+        ref_mel_batch: torch.Tensor,
+        ref_mel_len_batch: torch.Tensor,
+        estimated_reference_target_mel_len: List[int],
+        remove_input_padding: bool = False,
+        use_perf: bool = False,
+    ):
+        if use_perf:
+            torch.cuda.nvtx.range_push("text embedding")
+        batch = text_pad_sequence.shape[0]
+        max_seq_len = ref_mel_batch.shape[1]
+
+        text_pad_sequence_drop = torch.cat(
+            (text_pad_sequence, torch.zeros((1, text_pad_sequence.shape[1]), dtype=torch.int32).to(self.device)), dim=0
+        )
+
+        text_embedding_drop_list = []
+        for i in range(batch + 1):
+            text_embedding_drop_list.append(self.text_embedding(text_pad_sequence_drop[i].unsqueeze(0).to(self.device)))
+        text_embedding_drop_condition = torch.cat(text_embedding_drop_list, dim=0)
+
+        text_embedding = text_embedding_drop_condition[:-1]
+        # text_embedding_drop B,T,C batch should be the same
+        text_embedding_drop = text_embedding_drop_condition[-1].unsqueeze(0).repeat(batch, 1, 1)
+
+        noise = torch.randn_like(ref_mel_batch).to(self.device)
+        rope_cos = self.rope_cos[:, :max_seq_len, :].float().repeat(batch, 1, 1)
+        rope_sin = self.rope_sin[:, :max_seq_len, :].float().repeat(batch, 1, 1)
+
+        cat_mel_text = torch.cat((ref_mel_batch, text_embedding), dim=-1)
+        cat_mel_text_drop = torch.cat(
+            (
+                torch.zeros((batch, max_seq_len, self.n_mel_channels), dtype=torch.float32).to(self.device),
+                text_embedding_drop,
+            ),
+            dim=-1,
+        )
+
+        time_expand = self.time_expand.repeat(2 * batch, 1, 1).contiguous()
+
+        # Convert estimated_reference_target_mel_len to tensor
+        input_lengths = torch.tensor(estimated_reference_target_mel_len, dtype=torch.int32)
+
+        # combine above along the batch dimension
+        inputs = {
+            "noise": torch.cat((noise, noise), dim=0).contiguous(),
+            "cond": torch.cat((cat_mel_text, cat_mel_text_drop), dim=0).contiguous(),
+            "time_expand": time_expand,
+            "rope_cos": torch.cat((rope_cos, rope_cos), dim=0).contiguous(),
+            "rope_sin": torch.cat((rope_sin, rope_sin), dim=0).contiguous(),
+            "input_lengths": torch.cat((input_lengths, input_lengths), dim=0).contiguous(),
+            "delta_t": self.delta_t,
+        }
+        if use_perf and remove_input_padding:
+            torch.cuda.nvtx.range_push("remove input padding")
+        if remove_input_padding:
+            max_seq_len = inputs["cond"].shape[1]
+            inputs["noise"] = remove_tensor_padding(inputs["noise"], inputs["input_lengths"])
+            inputs["cond"] = remove_tensor_padding(inputs["cond"], inputs["input_lengths"])
+            # for time_expand, convert from B,D to B,T,D by repeat
+            inputs["time_expand"] = inputs["time_expand"].unsqueeze(1).repeat(1, max_seq_len, 1, 1)
+            inputs["time_expand"] = remove_tensor_padding(inputs["time_expand"], inputs["input_lengths"])
+            inputs["rope_cos"] = remove_tensor_padding(inputs["rope_cos"], inputs["input_lengths"])
+            inputs["rope_sin"] = remove_tensor_padding(inputs["rope_sin"], inputs["input_lengths"])
+        if use_perf and remove_input_padding:
+            torch.cuda.nvtx.range_pop()
+        for key in inputs:
+            inputs[key] = inputs[key].to(self.device)
+        if use_perf:
+            torch.cuda.nvtx.range_pop()
+        start_time = time.time()
+        denoised = self.forward(**inputs, use_perf=use_perf)
+        cost_time = time.time() - start_time
+        if use_perf and remove_input_padding:
+            torch.cuda.nvtx.range_push("remove input padding output")
+        if remove_input_padding:
+            denoised_list = []
+            start_idx = 0
+            for i in range(batch):
+                denoised_list.append(denoised[start_idx : start_idx + inputs["input_lengths"][i]])
+                start_idx += inputs["input_lengths"][i]
+            if use_perf and remove_input_padding:
+                torch.cuda.nvtx.range_pop()
+            return denoised_list, cost_time
+        return denoised, cost_time

src/f5_tts/runtime/triton_trtllm/model_repo_f5_tts/f5_tts/1/model.py

@@ -0,0 +1,278 @@
+# Copyright 2025, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions
+# are met:
+#  * Redistributions of source code must retain the above copyright
+#    notice, this list of conditions and the following disclaimer.
+#  * Redistributions in binary form must reproduce the above copyright
+#    notice, this list of conditions and the following disclaimer in the
+#    documentation and/or other materials provided with the distribution.
+#  * Neither the name of NVIDIA CORPORATION nor the names of its
+#    contributors may be used to endorse or promote products derived
+#    from this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
+# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+# PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT OWNER OR
+# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+# PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
+# OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+import json
+import os
+
+import jieba
+import torch
+import torch.nn.functional as F
+import torchaudio
+import triton_python_backend_utils as pb_utils
+from f5_tts_trtllm import F5TTS
+from pypinyin import Style, lazy_pinyin
+from torch.nn.utils.rnn import pad_sequence
+from torch.utils.dlpack import from_dlpack, to_dlpack
+
+
+def get_tokenizer(vocab_file_path: str):
+    """
+    tokenizer   - "pinyin" do g2p for only chinese characters, need .txt vocab_file
+                - "char" for char-wise tokenizer, need .txt vocab_file
+                - "byte" for utf-8 tokenizer
+                - "custom" if you're directly passing in a path to the vocab.txt you want to use
+    vocab_size  - if use "pinyin", all available pinyin types, common alphabets (also those with accent) and symbols
+                - if use "char", derived from unfiltered character & symbol counts of custom dataset
+                - if use "byte", set to 256 (unicode byte range)
+    """
+    with open(vocab_file_path, "r", encoding="utf-8") as f:
+        vocab_char_map = {}
+        for i, char in enumerate(f):
+            vocab_char_map[char[:-1]] = i
+    vocab_size = len(vocab_char_map)
+    return vocab_char_map, vocab_size
+
+
+def convert_char_to_pinyin(reference_target_texts_list, polyphone=True):
+    final_reference_target_texts_list = []
+    custom_trans = str.maketrans(
+        {";": ",", "“": '"', "”": '"', "‘": "'", "’": "'"}
+    )  # add custom trans here, to address oov
+
+    def is_chinese(c):
+        return "\u3100" <= c <= "\u9fff"  # common chinese characters
+
+    for text in reference_target_texts_list:
+        char_list = []
+        text = text.translate(custom_trans)
+        for seg in jieba.cut(text):
+            seg_byte_len = len(bytes(seg, "UTF-8"))
+            if seg_byte_len == len(seg):  # if pure alphabets and symbols
+                if char_list and seg_byte_len > 1 and char_list[-1] not in " :'\"":
+                    char_list.append(" ")
+                char_list.extend(seg)
+            elif polyphone and seg_byte_len == 3 * len(seg):  # if pure east asian characters
+                seg_ = lazy_pinyin(seg, style=Style.TONE3, tone_sandhi=True)
+                for i, c in enumerate(seg):
+                    if is_chinese(c):
+                        char_list.append(" ")
+                    char_list.append(seg_[i])
+            else:  # if mixed characters, alphabets and symbols
+                for c in seg:
+                    if ord(c) < 256:
+                        char_list.extend(c)
+                    elif is_chinese(c):
+                        char_list.append(" ")
+                        char_list.extend(lazy_pinyin(c, style=Style.TONE3, tone_sandhi=True))
+                    else:
+                        char_list.append(c)
+        final_reference_target_texts_list.append(char_list)
+
+    return final_reference_target_texts_list
+
+
+def list_str_to_idx(
+    text: list[str] | list[list[str]],
+    vocab_char_map: dict[str, int],  # {char: idx}
+    padding_value=-1,
+):  # noqa: F722
+    list_idx_tensors = [torch.tensor([vocab_char_map.get(c, 0) for c in t]) for t in text]  # pinyin or char style
+    return list_idx_tensors
+
+
+class TritonPythonModel:
+    def initialize(self, args):
+        self.use_perf = True
+        self.device = torch.device("cuda")
+        self.target_audio_sample_rate = 24000
+        self.target_rms = 0.15  # target rms for audio
+        self.n_fft = 1024
+        self.win_length = 1024
+        self.hop_length = 256
+        self.n_mel_channels = 100
+        self.max_mel_len = 3000
+        self.head_dim = 64
+
+        parameters = json.loads(args["model_config"])["parameters"]
+        for key, value in parameters.items():
+            parameters[key] = value["string_value"]
+
+        self.vocab_char_map, self.vocab_size = get_tokenizer(parameters["vocab_file"])
+        self.reference_sample_rate = int(parameters["reference_audio_sample_rate"])
+        self.resampler = torchaudio.transforms.Resample(self.reference_sample_rate, self.target_audio_sample_rate)
+
+        self.tllm_model_dir = parameters["tllm_model_dir"]
+        config_file = os.path.join(self.tllm_model_dir, "config.json")
+        with open(config_file) as f:
+            config = json.load(f)
+        self.model = F5TTS(
+            config,
+            debug_mode=False,
+            tllm_model_dir=self.tllm_model_dir,
+            model_path=parameters["model_path"],
+            vocab_size=self.vocab_size,
+        )
+
+        self.vocoder = parameters["vocoder"]
+        assert self.vocoder in ["vocos", "bigvgan"]
+        if self.vocoder == "vocos":
+            self.mel_stft = torchaudio.transforms.MelSpectrogram(
+                sample_rate=self.target_audio_sample_rate,
+                n_fft=self.n_fft,
+                win_length=self.win_length,
+                hop_length=self.hop_length,
+                n_mels=self.n_mel_channels,
+                power=1,
+                center=True,
+                normalized=False,
+                norm=None,
+            ).to(self.device)
+            self.compute_mel_fn = self.get_vocos_mel_spectrogram
+        elif self.vocoder == "bigvgan":
+            self.compute_mel_fn = self.get_bigvgan_mel_spectrogram
+
+    def get_vocos_mel_spectrogram(self, waveform):
+        mel = self.mel_stft(waveform)
+        mel = mel.clamp(min=1e-5).log()
+        return mel.transpose(1, 2)
+
+    def forward_vocoder(self, mel):
+        mel = mel.to(torch.float32).contiguous().cpu()
+        input_tensor_0 = pb_utils.Tensor.from_dlpack("mel", to_dlpack(mel))
+
+        inference_request = pb_utils.InferenceRequest(
+            model_name="vocoder", requested_output_names=["waveform"], inputs=[input_tensor_0]
+        )
+        inference_response = inference_request.exec()
+        if inference_response.has_error():
+            raise pb_utils.TritonModelException(inference_response.error().message())
+        else:
+            waveform = pb_utils.get_output_tensor_by_name(inference_response, "waveform")
+            waveform = torch.utils.dlpack.from_dlpack(waveform.to_dlpack()).cpu()
+
+            return waveform
+
+    def execute(self, requests):
+        (
+            reference_text_list,
+            target_text_list,
+            reference_target_texts_list,
+            estimated_reference_target_mel_len,
+            reference_mel_len,
+        ) = [], [], [], [], []
+        mel_features_list = []
+        if self.use_perf:
+            torch.cuda.nvtx.range_push("preprocess")
+        for request in requests:
+            wav_tensor = pb_utils.get_input_tensor_by_name(request, "reference_wav")
+            wav_lens = pb_utils.get_input_tensor_by_name(request, "reference_wav_len")
+
+            reference_text = pb_utils.get_input_tensor_by_name(request, "reference_text").as_numpy()
+            reference_text = reference_text[0][0].decode("utf-8")
+            reference_text_list.append(reference_text)
+            target_text = pb_utils.get_input_tensor_by_name(request, "target_text").as_numpy()
+            target_text = target_text[0][0].decode("utf-8")
+            target_text_list.append(target_text)
+
+            text = reference_text + target_text
+            reference_target_texts_list.append(text)
+
+            wav = from_dlpack(wav_tensor.to_dlpack())
+            wav_len = from_dlpack(wav_lens.to_dlpack())
+            wav_len = wav_len.squeeze()
+            assert wav.shape[0] == 1, "Only support batch size 1 for now."
+            wav = wav[:, :wav_len]
+
+            ref_rms = torch.sqrt(torch.mean(torch.square(wav)))
+            if ref_rms < self.target_rms:
+                wav = wav * self.target_rms / ref_rms
+            if self.reference_sample_rate != self.target_audio_sample_rate:
+                wav = self.resampler(wav)
+            wav = wav.to(self.device)
+            if self.use_perf:
+                torch.cuda.nvtx.range_push("compute_mel")
+            mel_features = self.compute_mel_fn(wav)
+            if self.use_perf:
+                torch.cuda.nvtx.range_pop()
+            mel_features_list.append(mel_features)
+
+            reference_mel_len.append(mel_features.shape[1])
+            estimated_reference_target_mel_len.append(
+                int(
+                    mel_features.shape[1] * (1 + len(target_text.encode("utf-8")) / len(reference_text.encode("utf-8")))
+                )
+            )
+
+        max_seq_len = min(max(estimated_reference_target_mel_len), self.max_mel_len)
+
+        batch = len(requests)
+        mel_features = torch.zeros((batch, max_seq_len, self.n_mel_channels), dtype=torch.float16).to(self.device)
+        for i, mel in enumerate(mel_features_list):
+            mel_features[i, : mel.shape[1], :] = mel
+
+        reference_mel_len_tensor = torch.LongTensor(reference_mel_len).to(self.device)
+
+        pinyin_list = convert_char_to_pinyin(reference_target_texts_list, polyphone=True)
+        text_pad_sequence = list_str_to_idx(pinyin_list, self.vocab_char_map)
+
+        for i, item in enumerate(text_pad_sequence):
+            text_pad_sequence[i] = F.pad(
+                item, (0, estimated_reference_target_mel_len[i] - len(item)), mode="constant", value=-1
+            )
+            text_pad_sequence[i] += 1  # WAR: 0 is reserved for padding token, hard coding in F5-TTS
+        text_pad_sequence = pad_sequence(text_pad_sequence, padding_value=-1, batch_first=True).to(self.device)
+        text_pad_sequence = F.pad(
+            text_pad_sequence, (0, max_seq_len - text_pad_sequence.shape[1]), mode="constant", value=-1
+        )
+        if self.use_perf:
+            torch.cuda.nvtx.range_pop()
+
+        denoised, cost_time = self.model.sample(
+            text_pad_sequence,
+            mel_features,
+            reference_mel_len_tensor,
+            estimated_reference_target_mel_len,
+            remove_input_padding=False,
+            use_perf=self.use_perf,
+        )
+        if self.use_perf:
+            torch.cuda.nvtx.range_push("vocoder")
+
+        responses = []
+        for i in range(batch):
+            ref_me_len = reference_mel_len[i]
+            estimated_mel_len = estimated_reference_target_mel_len[i]
+            denoised_one_item = denoised[i, ref_me_len:estimated_mel_len, :].unsqueeze(0).transpose(1, 2)
+            audio = self.forward_vocoder(denoised_one_item)
+            rms = torch.sqrt(torch.mean(torch.square(audio)))
+            if rms < self.target_rms:
+                audio = audio * self.target_rms / rms
+
+            audio = pb_utils.Tensor.from_dlpack("waveform", to_dlpack(audio))
+            inference_response = pb_utils.InferenceResponse(output_tensors=[audio])
+            responses.append(inference_response)
+        if self.use_perf:
+            torch.cuda.nvtx.range_pop()
+        return responses

src/f5_tts/runtime/triton_trtllm/model_repo_f5_tts/f5_tts/config.pbtxt

@@ -0,0 +1,81 @@
+# Copyright (c) 2025, NVIDIA CORPORATION.  All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+name: "f5_tts"
+backend: "python"
+max_batch_size: 4
+dynamic_batching {
+    max_queue_delay_microseconds: 1000
+}
+parameters [
+  {
+    key: "vocab_file"
+    value: { string_value: "${vocab}"}
+  },
+  {
+   key: "model_path", 
+   value: {string_value:"${model}"}
+  },
+  {
+   key: "tllm_model_dir", 
+   value: {string_value:"${trtllm}"}
+  },
+  {
+   key: "reference_audio_sample_rate", 
+   value: {string_value:"24000"}
+  },
+  {
+   key: "vocoder", 
+   value: {string_value:"${vocoder}"}
+  }
+]
+
+input [
+  {
+    name: "reference_wav"
+    data_type: TYPE_FP32
+    dims: [-1]
+    optional: True
+  },
+  {
+    name: "reference_wav_len"
+    data_type: TYPE_INT32
+    dims: [1]
+    optional: True
+  },
+  {
+    name: "reference_text"
+    data_type: TYPE_STRING
+    dims: [1]
+  },
+  {
+    name: "target_text"
+    data_type: TYPE_STRING
+    dims: [1]
+  }
+]
+output [
+  {
+    name: "waveform"
+    data_type: TYPE_FP32
+    dims: [ -1 ]
+  }
+]
+
+instance_group [
+  {
+    count: 1
+    kind: KIND_GPU
+  }
+]

src/f5_tts/runtime/triton_trtllm/model_repo_f5_tts/vocoder/config.pbtxt

@@ -0,0 +1,32 @@
+name: "vocoder"
+backend: "tensorrt"
+default_model_filename: "vocoder.plan"
+max_batch_size: 4
+
+input [
+  {
+    name: "mel"
+    data_type: TYPE_FP32
+    dims: [ 100, -1 ]
+  }
+]
+
+output [
+  {
+    name: "waveform"
+    data_type: TYPE_FP32
+    dims: [ -1 ]
+  }
+]
+
+dynamic_batching {
+    preferred_batch_size: [1, 2, 4]
+    max_queue_delay_microseconds: 1
+}
+
+instance_group [
+  {
+    count: 1
+    kind: KIND_GPU 
+  }
+]

src/f5_tts/runtime/triton_trtllm/patch/__init__.py

@@ -0,0 +1,199 @@
+# SPDX-FileCopyrightText: Copyright (c) 2022-2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+from .baichuan.model import BaichuanForCausalLM
+from .bert.model import (
+    BertForQuestionAnswering,
+    BertForSequenceClassification,
+    BertModel,
+    RobertaForQuestionAnswering,
+    RobertaForSequenceClassification,
+    RobertaModel,
+)
+from .bloom.model import BloomForCausalLM, BloomModel
+from .chatglm.config import ChatGLMConfig
+from .chatglm.model import ChatGLMForCausalLM, ChatGLMModel
+from .cogvlm.config import CogVLMConfig
+from .cogvlm.model import CogVLMForCausalLM
+from .commandr.model import CohereForCausalLM
+from .dbrx.config import DbrxConfig
+from .dbrx.model import DbrxForCausalLM
+from .deepseek_v1.model import DeepseekForCausalLM
+from .deepseek_v2.model import DeepseekV2ForCausalLM
+from .dit.model import DiT
+from .eagle.model import EagleForCausalLM
+from .enc_dec.model import DecoderModel, EncoderModel, WhisperEncoder
+from .f5tts.model import F5TTS
+from .falcon.config import FalconConfig
+from .falcon.model import FalconForCausalLM, FalconModel
+from .gemma.config import GEMMA2_ARCHITECTURE, GEMMA_ARCHITECTURE, GemmaConfig
+from .gemma.model import GemmaForCausalLM
+from .gpt.config import GPTConfig
+from .gpt.model import GPTForCausalLM, GPTModel
+from .gptj.config import GPTJConfig
+from .gptj.model import GPTJForCausalLM, GPTJModel
+from .gptneox.model import GPTNeoXForCausalLM, GPTNeoXModel
+from .grok.model import GrokForCausalLM
+from .llama.config import LLaMAConfig
+from .llama.model import LLaMAForCausalLM, LLaMAModel
+from .mamba.model import MambaForCausalLM
+from .medusa.config import MedusaConfig
+from .medusa.model import MedusaForCausalLm
+from .mllama.model import MLLaMAModel
+from .modeling_utils import PretrainedConfig, PretrainedModel, SpeculativeDecodingMode
+from .mpt.model import MPTForCausalLM, MPTModel
+from .nemotron_nas.model import DeciLMForCausalLM
+from .opt.model import OPTForCausalLM, OPTModel
+from .phi.model import PhiForCausalLM, PhiModel
+from .phi3.model import Phi3ForCausalLM, Phi3Model
+from .qwen.model import QWenForCausalLM
+from .recurrentgemma.model import RecurrentGemmaForCausalLM
+from .redrafter.model import ReDrafterForCausalLM
+
+
+__all__ = [
+    "BertModel",
+    "BertForQuestionAnswering",
+    "BertForSequenceClassification",
+    "RobertaModel",
+    "RobertaForQuestionAnswering",
+    "RobertaForSequenceClassification",
+    "BloomModel",
+    "BloomForCausalLM",
+    "DiT",
+    "DeepseekForCausalLM",
+    "FalconConfig",
+    "DeepseekV2ForCausalLM",
+    "FalconForCausalLM",
+    "FalconModel",
+    "GPTConfig",
+    "GPTModel",
+    "GPTForCausalLM",
+    "OPTForCausalLM",
+    "OPTModel",
+    "LLaMAConfig",
+    "LLaMAForCausalLM",
+    "LLaMAModel",
+    "MedusaConfig",
+    "MedusaForCausalLm",
+    "ReDrafterForCausalLM",
+    "GPTJConfig",
+    "GPTJModel",
+    "GPTJForCausalLM",
+    "GPTNeoXModel",
+    "GPTNeoXForCausalLM",
+    "PhiModel",
+    "PhiConfig",
+    "Phi3Model",
+    "Phi3Config",
+    "PhiForCausalLM",
+    "Phi3ForCausalLM",
+    "ChatGLMConfig",
+    "ChatGLMForCausalLM",
+    "ChatGLMModel",
+    "BaichuanForCausalLM",
+    "QWenConfigQWenForCausalLM",
+    "QWenModel",
+    "EncoderModel",
+    "DecoderModel",
+    "PretrainedConfig",
+    "PretrainedModel",
+    "WhisperEncoder",
+    "MambaForCausalLM",
+    "MambaConfig",
+    "MPTForCausalLM",
+    "MPTModel",
+    "SkyworkForCausalLM",
+    "GemmaConfig",
+    "GemmaForCausalLM",
+    "DbrxConfig",
+    "DbrxForCausalLM",
+    "RecurrentGemmaForCausalLM",
+    "CogVLMConfig",
+    "CogVLMForCausalLM",
+    "EagleForCausalLM",
+    "SpeculativeDecodingMode",
+    "CohereForCausalLM",
+    "MLLaMAModel",
+    "F5TTS",
+]
+
+MODEL_MAP = {
+    "GPT2LMHeadModel": GPTForCausalLM,
+    "GPT2LMHeadCustomModel": GPTForCausalLM,
+    "GPTBigCodeForCausalLM": GPTForCausalLM,
+    "Starcoder2ForCausalLM": GPTForCausalLM,
+    "FuyuForCausalLM": GPTForCausalLM,
+    "Kosmos2ForConditionalGeneration": GPTForCausalLM,
+    "JAISLMHeadModel": GPTForCausalLM,
+    "GPTForCausalLM": GPTForCausalLM,
+    "NemotronForCausalLM": GPTForCausalLM,
+    "OPTForCausalLM": OPTForCausalLM,
+    "BloomForCausalLM": BloomForCausalLM,
+    "RWForCausalLM": FalconForCausalLM,
+    "FalconForCausalLM": FalconForCausalLM,
+    "PhiForCausalLM": PhiForCausalLM,
+    "Phi3ForCausalLM": Phi3ForCausalLM,
+    "Phi3VForCausalLM": Phi3ForCausalLM,
+    "Phi3SmallForCausalLM": Phi3ForCausalLM,
+    "PhiMoEForCausalLM": Phi3ForCausalLM,
+    "MambaForCausalLM": MambaForCausalLM,
+    "GPTNeoXForCausalLM": GPTNeoXForCausalLM,
+    "GPTJForCausalLM": GPTJForCausalLM,
+    "MPTForCausalLM": MPTForCausalLM,
+    "GLMModel": ChatGLMForCausalLM,
+    "ChatGLMModel": ChatGLMForCausalLM,
+    "ChatGLMForCausalLM": ChatGLMForCausalLM,
+    "LlamaForCausalLM": LLaMAForCausalLM,
+    "ExaoneForCausalLM": LLaMAForCausalLM,
+    "MistralForCausalLM": LLaMAForCausalLM,
+    "MixtralForCausalLM": LLaMAForCausalLM,
+    "ArcticForCausalLM": LLaMAForCausalLM,
+    "Grok1ModelForCausalLM": GrokForCausalLM,
+    "InternLMForCausalLM": LLaMAForCausalLM,
+    "InternLM2ForCausalLM": LLaMAForCausalLM,
+    "MedusaForCausalLM": MedusaForCausalLm,
+    "ReDrafterForCausalLM": ReDrafterForCausalLM,
+    "BaichuanForCausalLM": BaichuanForCausalLM,
+    "BaiChuanForCausalLM": BaichuanForCausalLM,
+    "SkyworkForCausalLM": LLaMAForCausalLM,
+    GEMMA_ARCHITECTURE: GemmaForCausalLM,
+    GEMMA2_ARCHITECTURE: GemmaForCausalLM,
+    "QWenLMHeadModel": QWenForCausalLM,
+    "QWenForCausalLM": QWenForCausalLM,
+    "Qwen2ForCausalLM": QWenForCausalLM,
+    "Qwen2MoeForCausalLM": QWenForCausalLM,
+    "Qwen2ForSequenceClassification": QWenForCausalLM,
+    "Qwen2VLForConditionalGeneration": QWenForCausalLM,
+    "WhisperEncoder": WhisperEncoder,
+    "EncoderModel": EncoderModel,
+    "DecoderModel": DecoderModel,
+    "DbrxForCausalLM": DbrxForCausalLM,
+    "RecurrentGemmaForCausalLM": RecurrentGemmaForCausalLM,
+    "CogVLMForCausalLM": CogVLMForCausalLM,
+    "DiT": DiT,
+    "DeepseekForCausalLM": DeepseekForCausalLM,
+    "DeciLMForCausalLM": DeciLMForCausalLM,
+    "DeepseekV2ForCausalLM": DeepseekV2ForCausalLM,
+    "EagleForCausalLM": EagleForCausalLM,
+    "CohereForCausalLM": CohereForCausalLM,
+    "MllamaForConditionalGeneration": MLLaMAModel,
+    "BertForQuestionAnswering": BertForQuestionAnswering,
+    "BertForSequenceClassification": BertForSequenceClassification,
+    "BertModel": BertModel,
+    "RobertaModel": RobertaModel,
+    "RobertaForQuestionAnswering": RobertaForQuestionAnswering,
+    "RobertaForSequenceClassification": RobertaForSequenceClassification,
+    "F5TTS": F5TTS,
+}

src/f5_tts/runtime/triton_trtllm/patch/f5tts/model.py

@@ -0,0 +1,222 @@
+from __future__ import annotations
+
+import os
+import sys
+from collections import OrderedDict
+
+import tensorrt as trt
+from tensorrt_llm._common import default_net
+
+from ..._utils import str_dtype_to_trt
+from ...functional import Tensor, concat
+from ...layers import Linear
+from ...module import Module, ModuleList
+from ...plugin import current_all_reduce_helper
+from ..modeling_utils import PretrainedConfig, PretrainedModel
+from .modules import AdaLayerNormZero_Final, ConvPositionEmbedding, DiTBlock, TimestepEmbedding
+
+
+current_file_path = os.path.abspath(__file__)
+parent_dir = os.path.dirname(current_file_path)
+sys.path.append(parent_dir)
+
+
+class InputEmbedding(Module):
+    def __init__(self, mel_dim, text_dim, out_dim):
+        super().__init__()
+        self.proj = Linear(mel_dim * 2 + text_dim, out_dim)
+        self.conv_pos_embed = ConvPositionEmbedding(dim=out_dim)
+
+    def forward(self, x, cond):
+        x = self.proj(concat([x, cond], dim=-1))
+        return self.conv_pos_embed(x) + x
+
+
+class F5TTS(PretrainedModel):
+    def __init__(self, config: PretrainedConfig):
+        super().__init__(config)
+        self.dtype = str_dtype_to_trt(config.dtype)
+
+        self.time_embed = TimestepEmbedding(config.hidden_size)
+        self.input_embed = InputEmbedding(config.mel_dim, config.text_dim, config.hidden_size)
+
+        self.dim = config.hidden_size
+        self.depth = config.num_hidden_layers
+        self.transformer_blocks = ModuleList(
+            [
+                DiTBlock(
+                    dim=self.dim,
+                    heads=config.num_attention_heads,
+                    dim_head=config.dim_head,
+                    ff_mult=config.ff_mult,
+                    dropout=config.dropout,
+                )
+                for _ in range(self.depth)
+            ]
+        )
+
+        self.norm_out = AdaLayerNormZero_Final(config.hidden_size)  # final modulation
+        self.proj_out = Linear(config.hidden_size, config.mel_dim)
+
+    def forward(
+        self,
+        noise,  # nosied input audio
+        cond,  # masked cond audio
+        time,  # time step
+        rope_cos,
+        rope_sin,
+        input_lengths,
+        scale=1.0,
+    ):
+        t = self.time_embed(time)
+        x = self.input_embed(noise, cond)
+        for block in self.transformer_blocks:
+            x = block(x, t, rope_cos=rope_cos, rope_sin=rope_sin, input_lengths=input_lengths, scale=scale)
+        denoise = self.proj_out(self.norm_out(x, t))
+        denoise.mark_output("denoised", self.dtype)
+        return denoise
+
+    def prepare_inputs(self, **kwargs):
+        max_batch_size = kwargs["max_batch_size"]
+        batch_size_range = [2, 2, max_batch_size]
+        mel_size = 100
+        max_seq_len = 3000
+        num_frames_range = [200, 2 * max_seq_len, max_seq_len * max_batch_size]
+        hidden_size = 512
+        concat_feature_dim = mel_size + hidden_size
+        freq_embed_dim = 256
+        head_dim = 64
+        mapping = self.config.mapping
+        if mapping.tp_size > 1:
+            current_all_reduce_helper().set_workspace_tensor(mapping, 1)
+        if default_net().plugin_config.remove_input_padding:
+            noise = Tensor(
+                name="noise",
+                dtype=self.dtype,
+                shape=[-1, mel_size],
+                dim_range=OrderedDict(
+                    [
+                        ("num_frames", [num_frames_range]),
+                        ("n_mels", [mel_size]),
+                    ]
+                ),
+            )
+            cond = Tensor(
+                name="cond",
+                dtype=self.dtype,
+                shape=[-1, concat_feature_dim],
+                dim_range=OrderedDict(
+                    [
+                        ("num_frames", [num_frames_range]),
+                        ("embeded_length", [concat_feature_dim]),
+                    ]
+                ),
+            )
+            time = Tensor(
+                name="time",
+                dtype=self.dtype,
+                shape=[-1, freq_embed_dim],
+                dim_range=OrderedDict(
+                    [
+                        ("num_frames", [num_frames_range]),
+                        ("freq_dim", [freq_embed_dim]),
+                    ]
+                ),
+            )
+            rope_cos = Tensor(
+                name="rope_cos",
+                dtype=self.dtype,
+                shape=[-1, head_dim],
+                dim_range=OrderedDict(
+                    [
+                        ("num_frames", [num_frames_range]),
+                        ("head_dim", [head_dim]),
+                    ]
+                ),
+            )
+            rope_sin = Tensor(
+                name="rope_sin",
+                dtype=self.dtype,
+                shape=[-1, head_dim],
+                dim_range=OrderedDict(
+                    [
+                        ("num_frames", [num_frames_range]),
+                        ("head_dim", [head_dim]),
+                    ]
+                ),
+            )
+
+        else:
+            noise = Tensor(
+                name="noise",
+                dtype=self.dtype,
+                shape=[-1, -1, mel_size],
+                dim_range=OrderedDict(
+                    [
+                        ("batch_size", [batch_size_range]),
+                        ("max_duratuion", [[100, max_seq_len // 2, max_seq_len]]),
+                        ("n_mels", [mel_size]),
+                    ]
+                ),
+            )
+            cond = Tensor(
+                name="cond",
+                dtype=self.dtype,
+                shape=[-1, -1, concat_feature_dim],
+                dim_range=OrderedDict(
+                    [
+                        ("batch_size", [batch_size_range]),
+                        ("max_duratuion", [[100, max_seq_len // 2, max_seq_len]]),
+                        ("embeded_length", [concat_feature_dim]),
+                    ]
+                ),
+            )
+            time = Tensor(
+                name="time",
+                dtype=self.dtype,
+                shape=[-1, freq_embed_dim],
+                dim_range=OrderedDict(
+                    [
+                        ("batch_size", [batch_size_range]),
+                        ("freq_dim", [freq_embed_dim]),
+                    ]
+                ),
+            )
+            rope_cos = Tensor(
+                name="rope_cos",
+                dtype=self.dtype,
+                shape=[-1, -1, head_dim],
+                dim_range=OrderedDict(
+                    [
+                        ("batch_size", [batch_size_range]),
+                        ("max_duratuion", [[100, max_seq_len // 2, max_seq_len]]),
+                        ("head_dim", [head_dim]),
+                    ]
+                ),
+            )
+            rope_sin = Tensor(
+                name="rope_sin",
+                dtype=self.dtype,
+                shape=[-1, -1, head_dim],
+                dim_range=OrderedDict(
+                    [
+                        ("batch_size", [batch_size_range]),
+                        ("max_duratuion", [[100, max_seq_len // 2, max_seq_len]]),
+                        ("head_dim", [head_dim]),
+                    ]
+                ),
+            )
+        input_lengths = Tensor(
+            name="input_lengths",
+            dtype=trt.int32,
+            shape=[-1],
+            dim_range=OrderedDict([("batch_size", [batch_size_range])]),
+        )
+        return {
+            "noise": noise,
+            "cond": cond,
+            "time": time,
+            "rope_cos": rope_cos,
+            "rope_sin": rope_sin,
+            "input_lengths": input_lengths,
+        }

src/f5_tts/runtime/triton_trtllm/patch/f5tts/modules.py

@@ -0,0 +1,412 @@
+from __future__ import annotations
+
+import math
+from typing import Optional
+
+import numpy as np
+import torch
+import torch.nn.functional as F
+from tensorrt_llm._common import default_net
+
+from ..._utils import str_dtype_to_trt, trt_dtype_to_np
+from ...functional import (
+    Tensor,
+    bert_attention,
+    cast,
+    chunk,
+    concat,
+    constant,
+    expand,
+    expand_dims,
+    expand_dims_like,
+    expand_mask,
+    gelu,
+    matmul,
+    permute,
+    shape,
+    silu,
+    slice,
+    softmax,
+    squeeze,
+    unsqueeze,
+    view,
+)
+from ...layers import ColumnLinear, Conv1d, LayerNorm, Linear, Mish, RowLinear
+from ...module import Module
+
+
+class FeedForward(Module):
+    def __init__(self, dim, dim_out=None, mult=4, dropout=0.0):
+        super().__init__()
+        inner_dim = int(dim * mult)
+        dim_out = dim_out if dim_out is not None else dim
+
+        self.project_in = Linear(dim, inner_dim)
+        self.ff = Linear(inner_dim, dim_out)
+
+    def forward(self, x):
+        return self.ff(gelu(self.project_in(x)))
+
+
+class AdaLayerNormZero(Module):
+    def __init__(self, dim):
+        super().__init__()
+
+        self.linear = Linear(dim, dim * 6)
+        self.norm = LayerNorm(dim, elementwise_affine=False, eps=1e-6)
+
+    def forward(self, x, emb=None):
+        emb = self.linear(silu(emb))
+        shift_msa, scale_msa, gate_msa, shift_mlp, scale_mlp, gate_mlp = chunk(emb, 6, dim=1)
+        x = self.norm(x)
+        ones = constant(np.ones(1, dtype=np.float32)).cast(x.dtype)
+        if default_net().plugin_config.remove_input_padding:
+            x = x * (ones + scale_msa) + shift_msa
+        else:
+            x = x * (ones + unsqueeze(scale_msa, 1)) + unsqueeze(shift_msa, 1)
+        return x, gate_msa, shift_mlp, scale_mlp, gate_mlp
+
+
+class AdaLayerNormZero_Final(Module):
+    def __init__(self, dim):
+        super().__init__()
+
+        self.linear = Linear(dim, dim * 2)
+
+        self.norm = LayerNorm(dim, elementwise_affine=False, eps=1e-6)
+
+    def forward(self, x, emb):
+        emb = self.linear(silu(emb))
+        scale, shift = chunk(emb, 2, dim=1)
+        ones = constant(np.ones(1, dtype=np.float32)).cast(x.dtype)
+        if default_net().plugin_config.remove_input_padding:
+            x = self.norm(x) * (ones + scale) + shift
+        else:
+            x = self.norm(x) * unsqueeze((ones + scale), 1)
+            x = x + unsqueeze(shift, 1)
+        return x
+
+
+class ConvPositionEmbedding(Module):
+    def __init__(self, dim, kernel_size=31, groups=16):
+        super().__init__()
+        assert kernel_size % 2 != 0
+        self.conv1d1 = Conv1d(dim, dim, kernel_size, groups=groups, padding=kernel_size // 2)
+        self.conv1d2 = Conv1d(dim, dim, kernel_size, groups=groups, padding=kernel_size // 2)
+        self.mish = Mish()
+
+    def forward(self, x, mask=None):  # noqa: F722
+        if default_net().plugin_config.remove_input_padding:
+            x = unsqueeze(x, 0)
+        x = permute(x, [0, 2, 1])
+        x = self.mish(self.conv1d2(self.mish(self.conv1d1(x))))
+        out = permute(x, [0, 2, 1])
+        if default_net().plugin_config.remove_input_padding:
+            out = squeeze(out, 0)
+        return out
+
+
+class Attention(Module):
+    def __init__(
+        self,
+        processor: AttnProcessor,
+        dim: int,
+        heads: int = 16,
+        dim_head: int = 64,
+        dropout: float = 0.0,
+        context_dim: Optional[int] = None,  # if not None -> joint attention
+        context_pre_only=None,
+    ):
+        super().__init__()
+
+        if not hasattr(F, "scaled_dot_product_attention"):
+            raise ImportError("Attention equires PyTorch 2.0, to use it, please upgrade PyTorch to 2.0.")
+
+        self.processor = processor
+
+        self.dim = dim  # hidden_size
+        self.heads = heads
+        self.inner_dim = dim_head * heads
+        self.dropout = dropout
+        self.attention_head_size = dim_head
+        self.context_dim = context_dim
+        self.context_pre_only = context_pre_only
+        self.tp_size = 1
+        self.num_attention_heads = heads // self.tp_size
+        self.num_attention_kv_heads = heads // self.tp_size  # 8
+        self.dtype = str_dtype_to_trt("float32")
+        self.attention_hidden_size = self.attention_head_size * self.num_attention_heads
+        self.to_q = ColumnLinear(
+            dim,
+            self.tp_size * self.num_attention_heads * self.attention_head_size,
+            bias=True,
+            dtype=self.dtype,
+            tp_group=None,
+            tp_size=self.tp_size,
+        )
+        self.to_k = ColumnLinear(
+            dim,
+            self.tp_size * self.num_attention_heads * self.attention_head_size,
+            bias=True,
+            dtype=self.dtype,
+            tp_group=None,
+            tp_size=self.tp_size,
+        )
+        self.to_v = ColumnLinear(
+            dim,
+            self.tp_size * self.num_attention_heads * self.attention_head_size,
+            bias=True,
+            dtype=self.dtype,
+            tp_group=None,
+            tp_size=self.tp_size,
+        )
+
+        if self.context_dim is not None:
+            self.to_k_c = Linear(context_dim, self.inner_dim)
+            self.to_v_c = Linear(context_dim, self.inner_dim)
+            if self.context_pre_only is not None:
+                self.to_q_c = Linear(context_dim, self.inner_dim)
+
+        self.to_out = RowLinear(
+            self.tp_size * self.num_attention_heads * self.attention_head_size,
+            dim,
+            bias=True,
+            dtype=self.dtype,
+            tp_group=None,
+            tp_size=self.tp_size,
+        )
+
+        if self.context_pre_only is not None and not self.context_pre_only:
+            self.to_out_c = Linear(self.inner_dim, dim)
+
+    def forward(
+        self,
+        x,  # noised input x
+        rope_cos,
+        rope_sin,
+        input_lengths,
+        c=None,  # context c
+        scale=1.0,
+        rope=None,
+        c_rope=None,  # rotary position embedding for c
+    ) -> torch.Tensor:
+        if c is not None:
+            return self.processor(self, x, c=c, input_lengths=input_lengths, scale=scale, rope=rope, c_rope=c_rope)
+        else:
+            return self.processor(
+                self, x, rope_cos=rope_cos, rope_sin=rope_sin, input_lengths=input_lengths, scale=scale
+            )
+
+
+def rotate_every_two_3dim(tensor: Tensor) -> Tensor:
+    shape_tensor = concat(
+        [shape(tensor, i) / 2 if i == (tensor.ndim() - 1) else shape(tensor, i) for i in range(tensor.ndim())]
+    )
+    if default_net().plugin_config.remove_input_padding:
+        assert tensor.ndim() == 2
+        x1 = slice(tensor, [0, 0], shape_tensor, [1, 2])
+        x2 = slice(tensor, [0, 1], shape_tensor, [1, 2])
+        x1 = expand_dims(x1, 2)
+        x2 = expand_dims(x2, 2)
+        zero = constant(np.ascontiguousarray(np.zeros([1], dtype=trt_dtype_to_np(tensor.dtype))))
+        x2 = zero - x2
+        x = concat([x2, x1], 2)
+        out = view(x, concat([shape(x, 0), shape(x, 1) * 2]))
+    else:
+        assert tensor.ndim() == 3
+
+        x1 = slice(tensor, [0, 0, 0], shape_tensor, [1, 1, 2])
+        x2 = slice(tensor, [0, 0, 1], shape_tensor, [1, 1, 2])
+        x1 = expand_dims(x1, 3)
+        x2 = expand_dims(x2, 3)
+        zero = constant(np.ascontiguousarray(np.zeros([1], dtype=trt_dtype_to_np(tensor.dtype))))
+        x2 = zero - x2
+        x = concat([x2, x1], 3)
+        out = view(x, concat([shape(x, 0), shape(x, 1), shape(x, 2) * 2]))
+
+    return out
+
+
+def apply_rotary_pos_emb_3dim(x, rope_cos, rope_sin):
+    if default_net().plugin_config.remove_input_padding:
+        rot_dim = shape(rope_cos, -1)  # 64
+        new_t_shape = concat([shape(x, 0), rot_dim])  # (-1, 64)
+        x_ = slice(x, [0, 0], new_t_shape, [1, 1])
+        end_dim = shape(x, -1) - shape(rope_cos, -1)
+        new_t_unrotated_shape = concat([shape(x, 0), end_dim])  # (2, -1, 960)
+        x_unrotated = slice(x, concat([0, rot_dim]), new_t_unrotated_shape, [1, 1])
+        out = concat([x_ * rope_cos + rotate_every_two_3dim(x_) * rope_sin, x_unrotated], dim=-1)
+    else:
+        rot_dim = shape(rope_cos, 2)  # 64
+        new_t_shape = concat([shape(x, 0), shape(x, 1), rot_dim])  # (2, -1, 64)
+        x_ = slice(x, [0, 0, 0], new_t_shape, [1, 1, 1])
+        end_dim = shape(x, 2) - shape(rope_cos, 2)
+        new_t_unrotated_shape = concat([shape(x, 0), shape(x, 1), end_dim])  # (2, -1, 960)
+        x_unrotated = slice(x, concat([0, 0, rot_dim]), new_t_unrotated_shape, [1, 1, 1])
+        out = concat([x_ * rope_cos + rotate_every_two_3dim(x_) * rope_sin, x_unrotated], dim=-1)
+    return out
+
+
+class AttnProcessor:
+    def __init__(self):
+        pass
+
+    def __call__(
+        self,
+        attn,
+        x,  # noised input x
+        rope_cos,
+        rope_sin,
+        input_lengths,
+        scale=1.0,
+        rope=None,
+    ) -> torch.FloatTensor:
+        query = attn.to_q(x)
+        key = attn.to_k(x)
+        value = attn.to_v(x)
+        # k,v,q all (2,1226,1024)
+        query = apply_rotary_pos_emb_3dim(query, rope_cos, rope_sin)
+        key = apply_rotary_pos_emb_3dim(key, rope_cos, rope_sin)
+
+        # attention
+        inner_dim = key.shape[-1]
+        norm_factor = math.sqrt(attn.attention_head_size)
+        q_scaling = 1.0 / norm_factor
+        mask = None
+        if not default_net().plugin_config.remove_input_padding:
+            N = shape(x, 1)
+            B = shape(x, 0)
+            seq_len_2d = concat([1, N])
+            max_position_embeddings = 4096
+            # create position ids
+            position_ids_buffer = constant(np.expand_dims(np.arange(max_position_embeddings).astype(np.int32), 0))
+            tmp_position_ids = slice(position_ids_buffer, starts=[0, 0], sizes=seq_len_2d)
+            tmp_position_ids = expand(tmp_position_ids, concat([B, N]))  # BxL
+            tmp_input_lengths = unsqueeze(input_lengths, 1)  # Bx1
+            tmp_input_lengths = expand(tmp_input_lengths, concat([B, N]))  # BxL
+            mask = tmp_position_ids < tmp_input_lengths  # BxL
+            mask = mask.cast("int32")
+
+        if default_net().plugin_config.bert_attention_plugin:
+            qkv = concat([query, key, value], dim=-1)
+            # TRT plugin mode
+            assert input_lengths is not None
+            if default_net().plugin_config.remove_input_padding:
+                qkv = qkv.view(concat([-1, 3 * inner_dim]))
+                max_input_length = constant(
+                    np.zeros(
+                        [
+                            2048,
+                        ],
+                        dtype=np.int32,
+                    )
+                )
+            else:
+                max_input_length = None
+            context = bert_attention(
+                qkv,
+                input_lengths,
+                attn.num_attention_heads,
+                attn.attention_head_size,
+                q_scaling=q_scaling,
+                max_input_length=max_input_length,
+            )
+        else:
+            assert not default_net().plugin_config.remove_input_padding
+
+            def transpose_for_scores(x):
+                new_x_shape = concat([shape(x, 0), shape(x, 1), attn.num_attention_heads, attn.attention_head_size])
+
+                y = x.view(new_x_shape)
+                y = y.transpose(1, 2)
+                return y
+
+            def transpose_for_scores_k(x):
+                new_x_shape = concat([shape(x, 0), shape(x, 1), attn.num_attention_heads, attn.attention_head_size])
+
+                y = x.view(new_x_shape)
+                y = y.permute([0, 2, 3, 1])
+                return y
+
+            query = transpose_for_scores(query)
+            key = transpose_for_scores_k(key)
+            value = transpose_for_scores(value)
+
+            attention_scores = matmul(query, key, use_fp32_acc=False)
+
+            if mask is not None:
+                attention_mask = expand_mask(mask, shape(query, 2))
+                attention_mask = cast(attention_mask, attention_scores.dtype)
+                attention_scores = attention_scores + attention_mask
+
+            attention_probs = softmax(attention_scores, dim=-1)
+
+            context = matmul(attention_probs, value, use_fp32_acc=False).transpose(1, 2)
+            context = context.view(concat([shape(context, 0), shape(context, 1), attn.attention_hidden_size]))
+        context = attn.to_out(context)
+        if mask is not None:
+            mask = mask.view(concat([shape(mask, 0), shape(mask, 1), 1]))
+            mask = expand_dims_like(mask, context)
+            mask = cast(mask, context.dtype)
+            context = context * mask
+        return context
+
+
+# DiT Block
+class DiTBlock(Module):
+    def __init__(self, dim, heads, dim_head, ff_mult=2, dropout=0.1):
+        super().__init__()
+
+        self.attn_norm = AdaLayerNormZero(dim)
+        self.attn = Attention(
+            processor=AttnProcessor(),
+            dim=dim,
+            heads=heads,
+            dim_head=dim_head,
+            dropout=dropout,
+        )
+
+        self.ff_norm = LayerNorm(dim, elementwise_affine=False, eps=1e-6)
+        self.ff = FeedForward(dim=dim, mult=ff_mult, dropout=dropout)
+
+    def forward(
+        self, x, t, rope_cos, rope_sin, input_lengths, scale=1.0, rope=ModuleNotFoundError
+    ):  # x: noised input, t: time embedding
+        # pre-norm & modulation for attention input
+        norm, gate_msa, shift_mlp, scale_mlp, gate_mlp = self.attn_norm(x, emb=t)
+        # attention
+        # norm ----> (2,1226,1024)
+        attn_output = self.attn(x=norm, rope_cos=rope_cos, rope_sin=rope_sin, input_lengths=input_lengths, scale=scale)
+
+        # process attention output for input x
+        if default_net().plugin_config.remove_input_padding:
+            x = x + gate_msa * attn_output
+        else:
+            x = x + unsqueeze(gate_msa, 1) * attn_output
+        ones = constant(np.ones(1, dtype=np.float32)).cast(x.dtype)
+        if default_net().plugin_config.remove_input_padding:
+            norm = self.ff_norm(x) * (ones + scale_mlp) + shift_mlp
+        else:
+            norm = self.ff_norm(x) * (ones + unsqueeze(scale_mlp, 1)) + unsqueeze(shift_mlp, 1)
+            # norm = self.ff_norm(x) * (ones + scale_mlp) + shift_mlp
+        ff_output = self.ff(norm)
+        if default_net().plugin_config.remove_input_padding:
+            x = x + gate_mlp * ff_output
+        else:
+            x = x + unsqueeze(gate_mlp, 1) * ff_output
+
+        return x
+
+
+class TimestepEmbedding(Module):
+    def __init__(self, dim, freq_embed_dim=256, dtype=None):
+        super().__init__()
+        # self.time_embed = SinusPositionEmbedding(freq_embed_dim)
+        self.mlp1 = Linear(freq_embed_dim, dim, bias=True, dtype=dtype)
+        self.mlp2 = Linear(dim, dim, bias=True, dtype=dtype)
+
+    def forward(self, timestep):
+        t_freq = self.mlp1(timestep)
+        t_freq = silu(t_freq)
+        t_emb = self.mlp2(t_freq)
+        return t_emb

src/f5_tts/runtime/triton_trtllm/requirements-pytorch.txt

@@ -0,0 +1,24 @@
+accelerate>=0.33.0
+bitsandbytes>0.37.0
+cached_path
+click
+datasets
+ema_pytorch>=0.5.2
+gradio>=3.45.2
+hydra-core>=1.3.0
+jieba
+librosa
+matplotlib
+numpy<=1.26.4
+pydub
+pypinyin
+safetensors
+soundfile
+tomli
+torch>=2.0.0
+# torchaudio>=2.0.0
+torchdiffeq
+tqdm>=4.65.0
+transformers
+x_transformers>=1.31.14
+packaging>=24.2

src/f5_tts/runtime/triton_trtllm/run.sh

@@ -0,0 +1,110 @@
+stage=$1
+stop_stage=$2
+model=$3 # F5TTS_Base
+if [ -z "$model" ]; then
+    echo "Model is none, using default model F5TTS_Base"
+    model=F5TTS_Base
+fi
+echo "Start stage: $stage, Stop stage: $stop_stage, Model: $model"
+export CUDA_VISIBLE_DEVICES=0
+
+F5_TTS_HF_DOWNLOAD_PATH=./F5-TTS
+F5_TTS_TRT_LLM_CHECKPOINT_PATH=./trtllm_ckpt
+F5_TTS_TRT_LLM_ENGINE_PATH=./f5_trt_llm_engine
+
+vocoder_trt_engine_path=vocos_vocoder.plan
+model_repo=./model_repo
+
+if [ $stage -le 0 ] && [ $stop_stage -ge 0 ]; then
+    echo "Downloading f5 tts from huggingface"
+    huggingface-cli download SWivid/F5-TTS --local-dir $F5_TTS_HF_DOWNLOAD_PATH
+
+fi
+
+if [ $stage -le 1 ] && [ $stop_stage -ge 1 ]; then
+    echo "Converting checkpoint"
+    python3 ./scripts/convert_checkpoint.py \
+        --timm_ckpt "$F5_TTS_HF_DOWNLOAD_PATH/$model/model_1200000.pt" \
+        --output_dir "$F5_TTS_TRT_LLM_CHECKPOINT_PATH" --model_name $model
+    python_package_path=/usr/local/lib/python3.12/dist-packages
+    cp -r patch/* $python_package_path/tensorrt_llm/models
+    trtllm-build --checkpoint_dir $F5_TTS_TRT_LLM_CHECKPOINT_PATH \
+      --max_batch_size 8 \
+      --output_dir $F5_TTS_TRT_LLM_ENGINE_PATH --remove_input_padding disable
+fi
+
+if [ $stage -le 2 ] && [ $stop_stage -ge 2 ]; then
+    echo "Exporting vocos vocoder"
+    onnx_vocoder_path=vocos_vocoder.onnx
+    python3 scripts/export_vocoder_to_onnx.py --vocoder vocos --output-path $onnx_vocoder_path
+    bash scripts/export_vocos_trt.sh $onnx_vocoder_path $vocoder_trt_engine_path
+fi
+
+if [ $stage -le 3 ] && [ $stop_stage -ge 3 ]; then
+    echo "Building triton server"
+    rm -r $model_repo
+    cp -r ./model_repo_f5_tts $model_repo
+    python3 scripts/fill_template.py -i $model_repo/f5_tts/config.pbtxt vocab:$F5_TTS_HF_DOWNLOAD_PATH/$model/vocab.txt,model:$F5_TTS_HF_DOWNLOAD_PATH/$model/model_1200000.pt,trtllm:$F5_TTS_TRT_LLM_ENGINE_PATH,vocoder:vocos
+    cp $vocoder_trt_engine_path $model_repo/vocoder/1/vocoder.plan
+fi
+
+if [ $stage -le 4 ] && [ $stop_stage -ge 4 ]; then
+    echo "Starting triton server"
+    tritonserver --model-repository=$model_repo
+fi
+
+if [ $stage -le 5 ] && [ $stop_stage -ge 5 ]; then
+    echo "Testing triton server"
+    num_task=1
+    log_dir=./log_concurrent_tasks_${num_task}
+    rm -r $log_dir
+    python3 client_grpc.py --num-tasks $num_task --huggingface-dataset yuekai/seed_tts --split-name wenetspeech4tts --log-dir $log_dir
+fi
+
+if [ $stage -le 6 ] && [ $stop_stage -ge 6 ]; then
+    echo "Testing http client"
+    audio=../../infer/examples/basic/basic_ref_en.wav
+    reference_text="Some call me nature, others call me mother nature."
+    target_text="I don't really care what you call me. I've been a silent spectator, watching species evolve, empires rise and fall. But always remember, I am mighty and enduring."
+    python3 client_http.py --reference-audio $audio --reference-text "$reference_text" --target-text "$target_text"
+fi
+
+if [ $stage -le 7 ] && [ $stop_stage -ge 7 ]; then
+    echo "TRT-LLM: offline decoding benchmark test"
+    batch_size=1
+    split_name=wenetspeech4tts
+    backend_type=trt
+    log_dir=./log_benchmark_batch_size_${batch_size}_${split_name}_${backend_type}
+    rm -r $log_dir
+    ln -s model_repo_f5_tts/f5_tts/1/f5_tts_trtllm.py ./
+    torchrun --nproc_per_node=1 \
+    benchmark.py --output-dir $log_dir \
+    --batch-size $batch_size \
+    --enable-warmup \
+    --split-name $split_name \
+    --model-path $F5_TTS_HF_DOWNLOAD_PATH/$model/model_1200000.pt \
+    --vocab-file $F5_TTS_HF_DOWNLOAD_PATH/$model/vocab.txt \
+    --vocoder-trt-engine-path $vocoder_trt_engine_path \
+    --backend-type $backend_type \
+    --tllm-model-dir $F5_TTS_TRT_LLM_ENGINE_PATH || exit 1
+fi
+
+if [ $stage -le 8 ] && [ $stop_stage -ge 8 ]; then
+    echo "Native Pytorch: offline decoding benchmark test"
+    pip install -r requirements-pytorch.txt
+    batch_size=1
+    split_name=wenetspeech4tts
+    backend_type=pytorch
+    log_dir=./log_benchmark_batch_size_${batch_size}_${split_name}_${backend_type}
+    rm -r $log_dir
+    ln -s model_repo_f5_tts/f5_tts/1/f5_tts_trtllm.py ./
+    torchrun --nproc_per_node=1 \
+    benchmark.py --output-dir $log_dir \
+    --batch-size $batch_size \
+    --split-name $split_name \
+    --enable-warmup \
+    --model-path $F5_TTS_HF_DOWNLOAD_PATH/$model/model_1200000.pt \
+    --vocab-file $F5_TTS_HF_DOWNLOAD_PATH/$model/vocab.txt \
+    --backend-type $backend_type \
+    --tllm-model-dir $F5_TTS_TRT_LLM_ENGINE_PATH || exit 1
+fi

src/f5_tts/runtime/triton_trtllm/scripts/conv_stft.py

@@ -0,0 +1,248 @@
+# Modified from https://github.com/echocatzh/conv-stft/blob/master/conv_stft/conv_stft.py
+
+# Copyright (c) 2024, NVIDIA CORPORATION.  All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# MIT License
+
+# Copyright (c) 2020 Shimin Zhang
+
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+
+# The above copyright notice and this permission notice shall be included in all
+# copies or substantial portions of the Software.
+
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+
+import torch as th
+import torch.nn.functional as F
+from scipy.signal import check_COLA, get_window
+
+
+support_clp_op = None
+if th.__version__ >= "1.7.0":
+    from torch.fft import rfft as fft
+
+    support_clp_op = True
+else:
+    from torch import rfft as fft
+
+
+class STFT(th.nn.Module):
+    def __init__(
+        self,
+        win_len=1024,
+        win_hop=512,
+        fft_len=1024,
+        enframe_mode="continue",
+        win_type="hann",
+        win_sqrt=False,
+        pad_center=True,
+    ):
+        """
+        Implement of STFT using 1D convolution and 1D transpose convolutions.
+        Implement of framing the signal in 2 ways, `break` and `continue`.
+        `break` method is a kaldi-like framing.
+        `continue` method is a librosa-like framing.
+
+        More information about `perfect reconstruction`:
+        1. https://ww2.mathworks.cn/help/signal/ref/stft.html
+        2. https://docs.scipy.org/doc/scipy/reference/generated/scipy.signal.get_window.html
+
+        Args:
+            win_len (int): Number of points in one frame.  Defaults to 1024.
+            win_hop (int): Number of framing stride. Defaults to 512.
+            fft_len (int): Number of DFT points. Defaults to 1024.
+            enframe_mode (str, optional): `break` and `continue`. Defaults to 'continue'.
+            win_type (str, optional): The type of window to create. Defaults to 'hann'.
+            win_sqrt (bool, optional): using square root window. Defaults to True.
+            pad_center (bool, optional): `perfect reconstruction` opts. Defaults to True.
+        """
+        super(STFT, self).__init__()
+        assert enframe_mode in ["break", "continue"]
+        assert fft_len >= win_len
+        self.win_len = win_len
+        self.win_hop = win_hop
+        self.fft_len = fft_len
+        self.mode = enframe_mode
+        self.win_type = win_type
+        self.win_sqrt = win_sqrt
+        self.pad_center = pad_center
+        self.pad_amount = self.fft_len // 2
+
+        en_k, fft_k, ifft_k, ola_k = self.__init_kernel__()
+        self.register_buffer("en_k", en_k)
+        self.register_buffer("fft_k", fft_k)
+        self.register_buffer("ifft_k", ifft_k)
+        self.register_buffer("ola_k", ola_k)
+
+    def __init_kernel__(self):
+        """
+        Generate enframe_kernel, fft_kernel, ifft_kernel and overlap-add kernel.
+        ** enframe_kernel: Using conv1d layer and identity matrix.
+        ** fft_kernel: Using linear layer for matrix multiplication. In fact,
+        enframe_kernel and fft_kernel can be combined, But for the sake of
+        readability, I took the two apart.
+        ** ifft_kernel, pinv of fft_kernel.
+        ** overlap-add kernel, just like enframe_kernel, but transposed.
+
+        Returns:
+            tuple: four kernels.
+        """
+        enframed_kernel = th.eye(self.fft_len)[:, None, :]
+        if support_clp_op:
+            tmp = fft(th.eye(self.fft_len))
+            fft_kernel = th.stack([tmp.real, tmp.imag], dim=2)
+        else:
+            fft_kernel = fft(th.eye(self.fft_len), 1)
+        if self.mode == "break":
+            enframed_kernel = th.eye(self.win_len)[:, None, :]
+            fft_kernel = fft_kernel[: self.win_len]
+        fft_kernel = th.cat((fft_kernel[:, :, 0], fft_kernel[:, :, 1]), dim=1)
+        ifft_kernel = th.pinverse(fft_kernel)[:, None, :]
+        window = get_window(self.win_type, self.win_len)
+
+        self.perfect_reconstruct = check_COLA(window, self.win_len, self.win_len - self.win_hop)
+        window = th.FloatTensor(window)
+        if self.mode == "continue":
+            left_pad = (self.fft_len - self.win_len) // 2
+            right_pad = left_pad + (self.fft_len - self.win_len) % 2
+            window = F.pad(window, (left_pad, right_pad))
+        if self.win_sqrt:
+            self.padded_window = window
+            window = th.sqrt(window)
+        else:
+            self.padded_window = window**2
+
+        fft_kernel = fft_kernel.T * window
+        ifft_kernel = ifft_kernel * window
+        ola_kernel = th.eye(self.fft_len)[: self.win_len, None, :]
+        if self.mode == "continue":
+            ola_kernel = th.eye(self.fft_len)[:, None, : self.fft_len]
+        return enframed_kernel, fft_kernel, ifft_kernel, ola_kernel
+
+    def is_perfect(self):
+        """
+        Whether the parameters win_len, win_hop and win_sqrt
+        obey constants overlap-add(COLA)
+
+        Returns:
+            bool: Return true if parameters obey COLA.
+        """
+        return self.perfect_reconstruct and self.pad_center
+
+    def transform(self, inputs, return_type="complex"):
+        """Take input data (audio) to STFT domain.
+
+        Args:
+            inputs (tensor): Tensor of floats, with shape (num_batch, num_samples)
+            return_type (str, optional): return (mag, phase) when `magphase`,
+            return (real, imag) when `realimag` and complex(real, imag) when `complex`.
+            Defaults to 'complex'.
+
+        Returns:
+            tuple: (mag, phase) when `magphase`, return (real, imag) when
+            `realimag`. Defaults to 'complex', each elements with shape
+            [num_batch, num_frequencies, num_frames]
+        """
+        assert return_type in ["magphase", "realimag", "complex"]
+        if inputs.dim() == 2:
+            inputs = th.unsqueeze(inputs, 1)
+        self.num_samples = inputs.size(-1)
+        if self.pad_center:
+            inputs = F.pad(inputs, (self.pad_amount, self.pad_amount), mode="reflect")
+        enframe_inputs = F.conv1d(inputs, self.en_k, stride=self.win_hop)
+        outputs = th.transpose(enframe_inputs, 1, 2)
+        outputs = F.linear(outputs, self.fft_k)
+        outputs = th.transpose(outputs, 1, 2)
+        dim = self.fft_len // 2 + 1
+        real = outputs[:, :dim, :]
+        imag = outputs[:, dim:, :]
+        if return_type == "realimag":
+            return real, imag
+        elif return_type == "complex":
+            assert support_clp_op
+            return th.complex(real, imag)
+        else:
+            mags = th.sqrt(real**2 + imag**2)
+            phase = th.atan2(imag, real)
+            return mags, phase
+
+    def inverse(self, input1, input2=None, input_type="magphase"):
+        """Call the inverse STFT (iSTFT), given tensors produced
+        by the `transform` function.
+
+        Args:
+            input1 (tensors): Magnitude/Real-part of STFT with shape
+            [num_batch, num_frequencies, num_frames]
+            input2 (tensors): Phase/Imag-part of STFT with shape
+            [num_batch, num_frequencies, num_frames]
+            input_type (str, optional): Mathematical meaning of input tensor's.
+            Defaults to 'magphase'.
+
+        Returns:
+            tensors: Reconstructed audio given magnitude and phase. Of
+                shape [num_batch, num_samples]
+        """
+        assert input_type in ["magphase", "realimag"]
+        if input_type == "realimag":
+            real, imag = None, None
+            if support_clp_op and th.is_complex(input1):
+                real, imag = input1.real, input1.imag
+            else:
+                real, imag = input1, input2
+        else:
+            real = input1 * th.cos(input2)
+            imag = input1 * th.sin(input2)
+        inputs = th.cat([real, imag], dim=1)
+        outputs = F.conv_transpose1d(inputs, self.ifft_k, stride=self.win_hop)
+        t = (self.padded_window[None, :, None]).repeat(1, 1, inputs.size(-1))
+        t = t.to(inputs.device)
+        coff = F.conv_transpose1d(t, self.ola_k, stride=self.win_hop)
+
+        num_frames = input1.size(-1)
+        num_samples = num_frames * self.win_hop
+
+        rm_start, rm_end = self.pad_amount, self.pad_amount + num_samples
+
+        outputs = outputs[..., rm_start:rm_end]
+        coff = coff[..., rm_start:rm_end]
+        coffidx = th.where(coff > 1e-8)
+        outputs[coffidx] = outputs[coffidx] / (coff[coffidx])
+        return outputs.squeeze(dim=1)
+
+    def forward(self, inputs):
+        """Take input data (audio) to STFT domain and then back to audio.
+
+        Args:
+            inputs (tensor): Tensor of floats, with shape [num_batch, num_samples]
+
+        Returns:
+            tensor: Reconstructed audio given magnitude and phase.
+            Of shape [num_batch, num_samples]
+        """
+        mag, phase = self.transform(inputs)
+        rec_wav = self.inverse(mag, phase)
+        return rec_wav

src/f5_tts/runtime/triton_trtllm/scripts/convert_checkpoint.py

@@ -0,0 +1,358 @@
+import argparse
+import json
+import os
+import re
+import time
+import traceback
+from concurrent.futures import ThreadPoolExecutor, as_completed
+
+import safetensors.torch
+import torch
+from tensorrt_llm import str_dtype_to_torch
+from tensorrt_llm.mapping import Mapping
+from tensorrt_llm.models.convert_utils import split, split_matrix_tp
+
+
+def split_q_tp(v, n_head, n_hidden, tensor_parallel, rank):
+    split_v = split(v, tensor_parallel, rank, dim=1)
+    return split_v.contiguous()
+
+
+def split_q_bias_tp(v, n_head, n_hidden, tensor_parallel, rank):
+    split_v = split(v, tensor_parallel, rank, dim=0)
+    return split_v.contiguous()
+
+
+FACEBOOK_DIT_NAME_MAPPING = {
+    "^time_embed.time_mlp.0.weight$": "time_embed.mlp1.weight",
+    "^time_embed.time_mlp.0.bias$": "time_embed.mlp1.bias",
+    "^time_embed.time_mlp.2.weight$": "time_embed.mlp2.weight",
+    "^time_embed.time_mlp.2.bias$": "time_embed.mlp2.bias",
+    "^input_embed.conv_pos_embed.conv1d.0.weight$": "input_embed.conv_pos_embed.conv1d1.weight",
+    "^input_embed.conv_pos_embed.conv1d.0.bias$": "input_embed.conv_pos_embed.conv1d1.bias",
+    "^input_embed.conv_pos_embed.conv1d.2.weight$": "input_embed.conv_pos_embed.conv1d2.weight",
+    "^input_embed.conv_pos_embed.conv1d.2.bias$": "input_embed.conv_pos_embed.conv1d2.bias",
+    "^transformer_blocks.0.attn.to_out.0.weight$": "transformer_blocks.0.attn.to_out.weight",
+    "^transformer_blocks.0.attn.to_out.0.bias$": "transformer_blocks.0.attn.to_out.bias",
+    "^transformer_blocks.1.attn.to_out.0.weight$": "transformer_blocks.1.attn.to_out.weight",
+    "^transformer_blocks.1.attn.to_out.0.bias$": "transformer_blocks.1.attn.to_out.bias",
+    "^transformer_blocks.2.attn.to_out.0.weight$": "transformer_blocks.2.attn.to_out.weight",
+    "^transformer_blocks.2.attn.to_out.0.bias$": "transformer_blocks.2.attn.to_out.bias",
+    "^transformer_blocks.3.attn.to_out.0.weight$": "transformer_blocks.3.attn.to_out.weight",
+    "^transformer_blocks.3.attn.to_out.0.bias$": "transformer_blocks.3.attn.to_out.bias",
+    "^transformer_blocks.4.attn.to_out.0.weight$": "transformer_blocks.4.attn.to_out.weight",
+    "^transformer_blocks.4.attn.to_out.0.bias$": "transformer_blocks.4.attn.to_out.bias",
+    "^transformer_blocks.5.attn.to_out.0.weight$": "transformer_blocks.5.attn.to_out.weight",
+    "^transformer_blocks.5.attn.to_out.0.bias$": "transformer_blocks.5.attn.to_out.bias",
+    "^transformer_blocks.6.attn.to_out.0.weight$": "transformer_blocks.6.attn.to_out.weight",
+    "^transformer_blocks.6.attn.to_out.0.bias$": "transformer_blocks.6.attn.to_out.bias",
+    "^transformer_blocks.7.attn.to_out.0.weight$": "transformer_blocks.7.attn.to_out.weight",
+    "^transformer_blocks.7.attn.to_out.0.bias$": "transformer_blocks.7.attn.to_out.bias",
+    "^transformer_blocks.8.attn.to_out.0.weight$": "transformer_blocks.8.attn.to_out.weight",
+    "^transformer_blocks.8.attn.to_out.0.bias$": "transformer_blocks.8.attn.to_out.bias",
+    "^transformer_blocks.9.attn.to_out.0.weight$": "transformer_blocks.9.attn.to_out.weight",
+    "^transformer_blocks.9.attn.to_out.0.bias$": "transformer_blocks.9.attn.to_out.bias",
+    "^transformer_blocks.10.attn.to_out.0.weight$": "transformer_blocks.10.attn.to_out.weight",
+    "^transformer_blocks.10.attn.to_out.0.bias$": "transformer_blocks.10.attn.to_out.bias",
+    "^transformer_blocks.11.attn.to_out.0.weight$": "transformer_blocks.11.attn.to_out.weight",
+    "^transformer_blocks.11.attn.to_out.0.bias$": "transformer_blocks.11.attn.to_out.bias",
+    "^transformer_blocks.12.attn.to_out.0.weight$": "transformer_blocks.12.attn.to_out.weight",
+    "^transformer_blocks.12.attn.to_out.0.bias$": "transformer_blocks.12.attn.to_out.bias",
+    "^transformer_blocks.13.attn.to_out.0.weight$": "transformer_blocks.13.attn.to_out.weight",
+    "^transformer_blocks.13.attn.to_out.0.bias$": "transformer_blocks.13.attn.to_out.bias",
+    "^transformer_blocks.14.attn.to_out.0.weight$": "transformer_blocks.14.attn.to_out.weight",
+    "^transformer_blocks.14.attn.to_out.0.bias$": "transformer_blocks.14.attn.to_out.bias",
+    "^transformer_blocks.15.attn.to_out.0.weight$": "transformer_blocks.15.attn.to_out.weight",
+    "^transformer_blocks.15.attn.to_out.0.bias$": "transformer_blocks.15.attn.to_out.bias",
+    "^transformer_blocks.16.attn.to_out.0.weight$": "transformer_blocks.16.attn.to_out.weight",
+    "^transformer_blocks.16.attn.to_out.0.bias$": "transformer_blocks.16.attn.to_out.bias",
+    "^transformer_blocks.17.attn.to_out.0.weight$": "transformer_blocks.17.attn.to_out.weight",
+    "^transformer_blocks.17.attn.to_out.0.bias$": "transformer_blocks.17.attn.to_out.bias",
+    "^transformer_blocks.18.attn.to_out.0.weight$": "transformer_blocks.18.attn.to_out.weight",
+    "^transformer_blocks.18.attn.to_out.0.bias$": "transformer_blocks.18.attn.to_out.bias",
+    "^transformer_blocks.19.attn.to_out.0.weight$": "transformer_blocks.19.attn.to_out.weight",
+    "^transformer_blocks.19.attn.to_out.0.bias$": "transformer_blocks.19.attn.to_out.bias",
+    "^transformer_blocks.20.attn.to_out.0.weight$": "transformer_blocks.20.attn.to_out.weight",
+    "^transformer_blocks.20.attn.to_out.0.bias$": "transformer_blocks.20.attn.to_out.bias",
+    "^transformer_blocks.21.attn.to_out.0.weight$": "transformer_blocks.21.attn.to_out.weight",
+    "^transformer_blocks.21.attn.to_out.0.bias$": "transformer_blocks.21.attn.to_out.bias",
+    "^transformer_blocks.0.ff.ff.0.0.weight$": "transformer_blocks.0.ff.project_in.weight",
+    "^transformer_blocks.0.ff.ff.0.0.bias$": "transformer_blocks.0.ff.project_in.bias",
+    "^transformer_blocks.0.ff.ff.2.weight$": "transformer_blocks.0.ff.ff.weight",
+    "^transformer_blocks.0.ff.ff.2.bias$": "transformer_blocks.0.ff.ff.bias",
+    "^transformer_blocks.1.ff.ff.0.0.weight$": "transformer_blocks.1.ff.project_in.weight",
+    "^transformer_blocks.1.ff.ff.0.0.bias$": "transformer_blocks.1.ff.project_in.bias",
+    "^transformer_blocks.1.ff.ff.2.weight$": "transformer_blocks.1.ff.ff.weight",
+    "^transformer_blocks.1.ff.ff.2.bias$": "transformer_blocks.1.ff.ff.bias",
+    "^transformer_blocks.2.ff.ff.0.0.weight$": "transformer_blocks.2.ff.project_in.weight",
+    "^transformer_blocks.2.ff.ff.0.0.bias$": "transformer_blocks.2.ff.project_in.bias",
+    "^transformer_blocks.2.ff.ff.2.weight$": "transformer_blocks.2.ff.ff.weight",
+    "^transformer_blocks.2.ff.ff.2.bias$": "transformer_blocks.2.ff.ff.bias",
+    "^transformer_blocks.3.ff.ff.0.0.weight$": "transformer_blocks.3.ff.project_in.weight",
+    "^transformer_blocks.3.ff.ff.0.0.bias$": "transformer_blocks.3.ff.project_in.bias",
+    "^transformer_blocks.3.ff.ff.2.weight$": "transformer_blocks.3.ff.ff.weight",
+    "^transformer_blocks.3.ff.ff.2.bias$": "transformer_blocks.3.ff.ff.bias",
+    "^transformer_blocks.4.ff.ff.0.0.weight$": "transformer_blocks.4.ff.project_in.weight",
+    "^transformer_blocks.4.ff.ff.0.0.bias$": "transformer_blocks.4.ff.project_in.bias",
+    "^transformer_blocks.4.ff.ff.2.weight$": "transformer_blocks.4.ff.ff.weight",
+    "^transformer_blocks.4.ff.ff.2.bias$": "transformer_blocks.4.ff.ff.bias",
+    "^transformer_blocks.5.ff.ff.0.0.weight$": "transformer_blocks.5.ff.project_in.weight",
+    "^transformer_blocks.5.ff.ff.0.0.bias$": "transformer_blocks.5.ff.project_in.bias",
+    "^transformer_blocks.5.ff.ff.2.weight$": "transformer_blocks.5.ff.ff.weight",
+    "^transformer_blocks.5.ff.ff.2.bias$": "transformer_blocks.5.ff.ff.bias",
+    "^transformer_blocks.6.ff.ff.0.0.weight$": "transformer_blocks.6.ff.project_in.weight",
+    "^transformer_blocks.6.ff.ff.0.0.bias$": "transformer_blocks.6.ff.project_in.bias",
+    "^transformer_blocks.6.ff.ff.2.weight$": "transformer_blocks.6.ff.ff.weight",
+    "^transformer_blocks.6.ff.ff.2.bias$": "transformer_blocks.6.ff.ff.bias",
+    "^transformer_blocks.7.ff.ff.0.0.weight$": "transformer_blocks.7.ff.project_in.weight",
+    "^transformer_blocks.7.ff.ff.0.0.bias$": "transformer_blocks.7.ff.project_in.bias",
+    "^transformer_blocks.7.ff.ff.2.weight$": "transformer_blocks.7.ff.ff.weight",
+    "^transformer_blocks.7.ff.ff.2.bias$": "transformer_blocks.7.ff.ff.bias",
+    "^transformer_blocks.8.ff.ff.0.0.weight$": "transformer_blocks.8.ff.project_in.weight",
+    "^transformer_blocks.8.ff.ff.0.0.bias$": "transformer_blocks.8.ff.project_in.bias",
+    "^transformer_blocks.8.ff.ff.2.weight$": "transformer_blocks.8.ff.ff.weight",
+    "^transformer_blocks.8.ff.ff.2.bias$": "transformer_blocks.8.ff.ff.bias",
+    "^transformer_blocks.9.ff.ff.0.0.weight$": "transformer_blocks.9.ff.project_in.weight",
+    "^transformer_blocks.9.ff.ff.0.0.bias$": "transformer_blocks.9.ff.project_in.bias",
+    "^transformer_blocks.9.ff.ff.2.weight$": "transformer_blocks.9.ff.ff.weight",
+    "^transformer_blocks.9.ff.ff.2.bias$": "transformer_blocks.9.ff.ff.bias",
+    "^transformer_blocks.10.ff.ff.0.0.weight$": "transformer_blocks.10.ff.project_in.weight",
+    "^transformer_blocks.10.ff.ff.0.0.bias$": "transformer_blocks.10.ff.project_in.bias",
+    "^transformer_blocks.10.ff.ff.2.weight$": "transformer_blocks.10.ff.ff.weight",
+    "^transformer_blocks.10.ff.ff.2.bias$": "transformer_blocks.10.ff.ff.bias",
+    "^transformer_blocks.11.ff.ff.0.0.weight$": "transformer_blocks.11.ff.project_in.weight",
+    "^transformer_blocks.11.ff.ff.0.0.bias$": "transformer_blocks.11.ff.project_in.bias",
+    "^transformer_blocks.11.ff.ff.2.weight$": "transformer_blocks.11.ff.ff.weight",
+    "^transformer_blocks.11.ff.ff.2.bias$": "transformer_blocks.11.ff.ff.bias",
+    "^transformer_blocks.12.ff.ff.0.0.weight$": "transformer_blocks.12.ff.project_in.weight",
+    "^transformer_blocks.12.ff.ff.0.0.bias$": "transformer_blocks.12.ff.project_in.bias",
+    "^transformer_blocks.12.ff.ff.2.weight$": "transformer_blocks.12.ff.ff.weight",
+    "^transformer_blocks.12.ff.ff.2.bias$": "transformer_blocks.12.ff.ff.bias",
+    "^transformer_blocks.13.ff.ff.0.0.weight$": "transformer_blocks.13.ff.project_in.weight",
+    "^transformer_blocks.13.ff.ff.0.0.bias$": "transformer_blocks.13.ff.project_in.bias",
+    "^transformer_blocks.13.ff.ff.2.weight$": "transformer_blocks.13.ff.ff.weight",
+    "^transformer_blocks.13.ff.ff.2.bias$": "transformer_blocks.13.ff.ff.bias",
+    "^transformer_blocks.14.ff.ff.0.0.weight$": "transformer_blocks.14.ff.project_in.weight",
+    "^transformer_blocks.14.ff.ff.0.0.bias$": "transformer_blocks.14.ff.project_in.bias",
+    "^transformer_blocks.14.ff.ff.2.weight$": "transformer_blocks.14.ff.ff.weight",
+    "^transformer_blocks.14.ff.ff.2.bias$": "transformer_blocks.14.ff.ff.bias",
+    "^transformer_blocks.15.ff.ff.0.0.weight$": "transformer_blocks.15.ff.project_in.weight",
+    "^transformer_blocks.15.ff.ff.0.0.bias$": "transformer_blocks.15.ff.project_in.bias",
+    "^transformer_blocks.15.ff.ff.2.weight$": "transformer_blocks.15.ff.ff.weight",
+    "^transformer_blocks.15.ff.ff.2.bias$": "transformer_blocks.15.ff.ff.bias",
+    "^transformer_blocks.16.ff.ff.0.0.weight$": "transformer_blocks.16.ff.project_in.weight",
+    "^transformer_blocks.16.ff.ff.0.0.bias$": "transformer_blocks.16.ff.project_in.bias",
+    "^transformer_blocks.16.ff.ff.2.weight$": "transformer_blocks.16.ff.ff.weight",
+    "^transformer_blocks.16.ff.ff.2.bias$": "transformer_blocks.16.ff.ff.bias",
+    "^transformer_blocks.17.ff.ff.0.0.weight$": "transformer_blocks.17.ff.project_in.weight",
+    "^transformer_blocks.17.ff.ff.0.0.bias$": "transformer_blocks.17.ff.project_in.bias",
+    "^transformer_blocks.17.ff.ff.2.weight$": "transformer_blocks.17.ff.ff.weight",
+    "^transformer_blocks.17.ff.ff.2.bias$": "transformer_blocks.17.ff.ff.bias",
+    "^transformer_blocks.18.ff.ff.0.0.weight$": "transformer_blocks.18.ff.project_in.weight",
+    "^transformer_blocks.18.ff.ff.0.0.bias$": "transformer_blocks.18.ff.project_in.bias",
+    "^transformer_blocks.18.ff.ff.2.weight$": "transformer_blocks.18.ff.ff.weight",
+    "^transformer_blocks.18.ff.ff.2.bias$": "transformer_blocks.18.ff.ff.bias",
+    "^transformer_blocks.19.ff.ff.0.0.weight$": "transformer_blocks.19.ff.project_in.weight",
+    "^transformer_blocks.19.ff.ff.0.0.bias$": "transformer_blocks.19.ff.project_in.bias",
+    "^transformer_blocks.19.ff.ff.2.weight$": "transformer_blocks.19.ff.ff.weight",
+    "^transformer_blocks.19.ff.ff.2.bias$": "transformer_blocks.19.ff.ff.bias",
+    "^transformer_blocks.20.ff.ff.0.0.weight$": "transformer_blocks.20.ff.project_in.weight",
+    "^transformer_blocks.20.ff.ff.0.0.bias$": "transformer_blocks.20.ff.project_in.bias",
+    "^transformer_blocks.20.ff.ff.2.weight$": "transformer_blocks.20.ff.ff.weight",
+    "^transformer_blocks.20.ff.ff.2.bias$": "transformer_blocks.20.ff.ff.bias",
+    "^transformer_blocks.21.ff.ff.0.0.weight$": "transformer_blocks.21.ff.project_in.weight",
+    "^transformer_blocks.21.ff.ff.0.0.bias$": "transformer_blocks.21.ff.project_in.bias",
+    "^transformer_blocks.21.ff.ff.2.weight$": "transformer_blocks.21.ff.ff.weight",
+    "^transformer_blocks.21.ff.ff.2.bias$": "transformer_blocks.21.ff.ff.bias",
+}
+
+
+def parse_arguments():
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--model_name",
+        type=str,
+        default="F5TTS_Base",
+        choices=[
+            "F5TTS_Base",
+        ],
+    )  # TODO: support F5TTS_v1_Base
+    parser.add_argument("--timm_ckpt", type=str, default="./ckpts/model_1200000.pt")
+    parser.add_argument(
+        "--output_dir", type=str, default="./tllm_checkpoint", help="The path to save the TensorRT-LLM checkpoint"
+    )
+    parser.add_argument("--hidden_size", type=int, default=1024, help="The hidden size of DiT")
+    parser.add_argument("--depth", type=int, default=22, help="The number of DiTBlock layers")
+    parser.add_argument("--num_heads", type=int, default=16, help="The number of heads of attention module")
+    parser.add_argument("--cfg_scale", type=float, default=4.0)
+    parser.add_argument("--tp_size", type=int, default=1, help="N-way tensor parallelism size")
+    parser.add_argument("--cp_size", type=int, default=1, help="Context parallelism size")
+    parser.add_argument("--pp_size", type=int, default=1, help="N-way pipeline parallelism size")
+    parser.add_argument("--dtype", type=str, default="float16", choices=["float32", "bfloat16", "float16"])
+    parser.add_argument("--fp8_linear", action="store_true", help="Whether use FP8 for linear layers")
+    parser.add_argument(
+        "--workers", type=int, default=1, help="The number of workers for converting checkpoint in parallel"
+    )
+    args = parser.parse_args()
+    return args
+
+
+def convert_timm_dit(args, mapping, dtype="float32"):
+    weights = {}
+    tik = time.time()
+    torch_dtype = str_dtype_to_torch(dtype)
+    tensor_parallel = mapping.tp_size
+
+    model_params = dict(torch.load(args.timm_ckpt))
+    model_params = {
+        k: v for k, v in model_params["ema_model_state_dict"].items() if k.startswith("ema_model.transformer")
+    }
+    prefix = "ema_model.transformer."
+    model_params = {key[len(prefix) :] if key.startswith(prefix) else key: value for key, value in model_params.items()}
+
+    timm_to_trtllm_name = FACEBOOK_DIT_NAME_MAPPING
+
+    def get_trtllm_name(timm_name):
+        for k, v in timm_to_trtllm_name.items():
+            m = re.match(k, timm_name)
+            if m is not None:
+                if "*" in v:
+                    v = v.replace("*", m.groups()[0])
+                return v
+        return timm_name
+
+    weights = dict()
+    for name, param in model_params.items():
+        if name == "input_embed.conv_pos_embed.conv1d.0.weight" or name == "input_embed.conv_pos_embed.conv1d.2.weight":
+            weights[get_trtllm_name(name)] = param.contiguous().to(torch_dtype).unsqueeze(-1)
+        else:
+            weights[get_trtllm_name(name)] = param.contiguous().to(torch_dtype)
+
+    assert len(weights) == len(model_params)
+
+    # new_prefix = 'f5_transformer.'
+    new_prefix = ""
+    weights = {new_prefix + key: value for key, value in weights.items()}
+    import math
+
+    scale_factor = math.pow(64, -0.25)
+    for k, v in weights.items():
+        if re.match("^transformer_blocks.*.attn.to_k.weight$", k):
+            weights[k] *= scale_factor
+            weights[k] = split_q_tp(v, args.num_heads, args.hidden_size, tensor_parallel, mapping.tp_rank)
+
+        elif re.match("^transformer_blocks.*.attn.to_k.bias$", k):
+            weights[k] *= scale_factor
+            weights[k] = split_q_bias_tp(v, args.num_heads, args.hidden_size, tensor_parallel, mapping.tp_rank)
+
+        elif re.match("^transformer_blocks.*.attn.to_q.weight$", k):
+            weights[k] = split_q_tp(v, args.num_heads, args.hidden_size, tensor_parallel, mapping.tp_rank)
+            weights[k] *= scale_factor
+
+        elif re.match("^transformer_blocks.*.attn.to_q.bias$", k):
+            weights[k] = split_q_bias_tp(v, args.num_heads, args.hidden_size, tensor_parallel, mapping.tp_rank)
+            weights[k] *= scale_factor
+
+        elif re.match("^transformer_blocks.*.attn.to_v.weight$", k):
+            weights[k] = split_q_tp(v, args.num_heads, args.hidden_size, tensor_parallel, mapping.tp_rank)
+
+        elif re.match("^transformer_blocks.*.attn.to_v.bias$", k):
+            weights[k] = split_q_bias_tp(v, args.num_heads, args.hidden_size, tensor_parallel, mapping.tp_rank)
+
+        elif re.match("^transformer_blocks.*.attn.to_out.weight$", k):
+            weights[k] = split_matrix_tp(v, tensor_parallel, mapping.tp_rank, dim=1)
+
+    tok = time.time()
+    t = time.strftime("%H:%M:%S", time.gmtime(tok - tik))
+    print(f"Weights loaded. Total time: {t}")
+    return weights
+
+
+def save_config(args):
+    if not os.path.exists(args.output_dir):
+        os.makedirs(args.output_dir)
+    config = {
+        "architecture": "F5TTS",
+        "dtype": args.dtype,
+        "hidden_size": 1024,
+        "num_hidden_layers": 22,
+        "num_attention_heads": 16,
+        "dim_head": 64,
+        "dropout": 0.1,
+        "ff_mult": 2,
+        "mel_dim": 100,
+        "text_num_embeds": 256,
+        "text_dim": 512,
+        "conv_layers": 4,
+        "long_skip_connection": False,
+        "mapping": {
+            "world_size": args.cp_size * args.tp_size * args.pp_size,
+            "cp_size": args.cp_size,
+            "tp_size": args.tp_size,
+            "pp_size": args.pp_size,
+        },
+    }
+    if args.fp8_linear:
+        config["quantization"] = {
+            "quant_algo": "FP8",
+            # TODO: add support for exclude modules.
+            # 'exclude_modules': "*final_layer*",
+        }
+
+    with open(os.path.join(args.output_dir, "config.json"), "w") as f:
+        json.dump(config, f, indent=4)
+
+
+def covert_and_save(args, rank):
+    if rank == 0:
+        save_config(args)
+
+    mapping = Mapping(
+        world_size=args.cp_size * args.tp_size * args.pp_size,
+        rank=rank,
+        cp_size=args.cp_size,
+        tp_size=args.tp_size,
+        pp_size=args.pp_size,
+    )
+
+    weights = convert_timm_dit(args, mapping, dtype=args.dtype)
+
+    safetensors.torch.save_file(weights, os.path.join(args.output_dir, f"rank{rank}.safetensors"))
+
+
+def execute(workers, func, args):
+    if workers == 1:
+        for rank, f in enumerate(func):
+            f(args, rank)
+    else:
+        with ThreadPoolExecutor(max_workers=workers) as p:
+            futures = [p.submit(f, args, rank) for rank, f in enumerate(func)]
+            exceptions = []
+            for future in as_completed(futures):
+                try:
+                    future.result()
+                except Exception as e:
+                    traceback.print_exc()
+                    exceptions.append(e)
+            assert len(exceptions) == 0, "Checkpoint conversion failed, please check error log."
+
+
+def main():
+    args = parse_arguments()
+    world_size = args.cp_size * args.tp_size * args.pp_size
+
+    assert args.pp_size == 1, "PP is not supported yet."
+
+    tik = time.time()
+    if args.timm_ckpt is None:
+        return
+    print("start execute")
+    execute(args.workers, [covert_and_save] * world_size, args)
+
+    tok = time.time()
+    t = time.strftime("%H:%M:%S", time.gmtime(tok - tik))
+    print(f"Total time of converting checkpoints: {t}")
+
+
+if __name__ == "__main__":
+    main()

src/f5_tts/runtime/triton_trtllm/scripts/export_vocoder_to_onnx.py

@@ -0,0 +1,138 @@
+# Copyright (c) 2024, NVIDIA CORPORATION.  All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import argparse
+
+import torch
+import torch.nn as nn
+from conv_stft import STFT
+from huggingface_hub import hf_hub_download
+from vocos import Vocos
+
+
+opset_version = 17
+
+
+def get_args():
+    parser = argparse.ArgumentParser(formatter_class=argparse.ArgumentDefaultsHelpFormatter)
+    parser.add_argument(
+        "--vocoder",
+        type=str,
+        default="vocos",
+        choices=["vocos", "bigvgan"],
+        help="Vocoder to export",
+    )
+    parser.add_argument(
+        "--output-path",
+        type=str,
+        default="./vocos_vocoder.onnx",
+        help="Output path",
+    )
+    return parser.parse_args()
+
+
+class ISTFTHead(nn.Module):
+    def __init__(self, n_fft: int, hop_length: int):
+        super().__init__()
+        self.out = None
+        self.stft = STFT(fft_len=n_fft, win_hop=hop_length, win_len=n_fft)
+
+    def forward(self, x: torch.Tensor):
+        x = self.out(x).transpose(1, 2)
+        mag, p = x.chunk(2, dim=1)
+        mag = torch.exp(mag)
+        mag = torch.clip(mag, max=1e2)
+        real = mag * torch.cos(p)
+        imag = mag * torch.sin(p)
+        audio = self.stft.inverse(input1=real, input2=imag, input_type="realimag")
+        return audio
+
+
+class VocosVocoder(nn.Module):
+    def __init__(self, vocos_vocoder):
+        super(VocosVocoder, self).__init__()
+        self.vocos_vocoder = vocos_vocoder
+        istft_head_out = self.vocos_vocoder.head.out
+        n_fft = self.vocos_vocoder.head.istft.n_fft
+        hop_length = self.vocos_vocoder.head.istft.hop_length
+        istft_head_for_export = ISTFTHead(n_fft, hop_length)
+        istft_head_for_export.out = istft_head_out
+        self.vocos_vocoder.head = istft_head_for_export
+
+    def forward(self, mel):
+        waveform = self.vocos_vocoder.decode(mel)
+        return waveform
+
+
+def export_VocosVocoder(vocos_vocoder, output_path, verbose):
+    vocos_vocoder = VocosVocoder(vocos_vocoder).cuda()
+    vocos_vocoder.eval()
+
+    dummy_batch_size = 8
+    dummy_input_length = 500
+
+    dummy_mel = torch.randn(dummy_batch_size, 100, dummy_input_length).cuda()
+
+    with torch.no_grad():
+        dummy_waveform = vocos_vocoder(mel=dummy_mel)
+        print(dummy_waveform.shape)
+
+    dummy_input = dummy_mel
+
+    torch.onnx.export(
+        vocos_vocoder,
+        dummy_input,
+        output_path,
+        opset_version=opset_version,
+        do_constant_folding=True,
+        input_names=["mel"],
+        output_names=["waveform"],
+        dynamic_axes={
+            "mel": {0: "batch_size", 2: "input_length"},
+            "waveform": {0: "batch_size", 1: "output_length"},
+        },
+        verbose=verbose,
+    )
+
+    print("Exported to {}".format(output_path))
+
+
+def load_vocoder(vocoder_name="vocos", is_local=False, local_path="", device="cpu", hf_cache_dir=None):
+    if vocoder_name == "vocos":
+        # vocoder = Vocos.from_pretrained("charactr/vocos-mel-24khz").to(device)
+        if is_local:
+            print(f"Load vocos from local path {local_path}")
+            config_path = f"{local_path}/config.yaml"
+            model_path = f"{local_path}/pytorch_model.bin"
+        else:
+            print("Download Vocos from huggingface charactr/vocos-mel-24khz")
+            repo_id = "charactr/vocos-mel-24khz"
+            config_path = hf_hub_download(repo_id=repo_id, cache_dir=hf_cache_dir, filename="config.yaml")
+            model_path = hf_hub_download(repo_id=repo_id, cache_dir=hf_cache_dir, filename="pytorch_model.bin")
+        vocoder = Vocos.from_hparams(config_path)
+        state_dict = torch.load(model_path, map_location="cpu", weights_only=True)
+        vocoder.load_state_dict(state_dict)
+        vocoder = vocoder.eval().to(device)
+    elif vocoder_name == "bigvgan":
+        raise NotImplementedError("BigVGAN is not supported yet")
+        vocoder.remove_weight_norm()
+        vocoder = vocoder.eval().to(device)
+    return vocoder
+
+
+if __name__ == "__main__":
+    args = get_args()
+    vocoder = load_vocoder(vocoder_name=args.vocoder, device="cpu", hf_cache_dir=None)
+    if args.vocoder == "vocos":
+        export_VocosVocoder(vocoder, args.output_path, verbose=False)

src/f5_tts/runtime/triton_trtllm/scripts/export_vocos_trt.sh

@@ -0,0 +1,43 @@
+#!/bin/bash
+# Copyright (c) 2025, NVIDIA CORPORATION.  All rights reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+TRTEXEC="/usr/src/tensorrt/bin/trtexec"
+
+ONNX_PATH=$1
+ENGINE_PATH=$2
+echo "ONNX_PATH: $ONNX_PATH"
+echo "ENGINE_PATH: $ENGINE_PATH"
+PRECISION="fp32"
+
+
+MIN_BATCH_SIZE=1
+OPT_BATCH_SIZE=1
+MAX_BATCH_SIZE=8
+
+MIN_INPUT_LENGTH=1
+OPT_INPUT_LENGTH=1000
+MAX_INPUT_LENGTH=3000
+
+MEL_MIN_SHAPE="${MIN_BATCH_SIZE}x100x${MIN_INPUT_LENGTH}"
+MEL_OPT_SHAPE="${OPT_BATCH_SIZE}x100x${OPT_INPUT_LENGTH}"
+MEL_MAX_SHAPE="${MAX_BATCH_SIZE}x100x${MAX_INPUT_LENGTH}"
+
+${TRTEXEC} \
+    --minShapes="mel:${MEL_MIN_SHAPE}" \
+    --optShapes="mel:${MEL_OPT_SHAPE}" \
+    --maxShapes="mel:${MEL_MAX_SHAPE}" \
+    --onnx=${ONNX_PATH} \
+    --saveEngine=${ENGINE_PATH}
+

src/f5_tts/runtime/triton_trtllm/scripts/fill_template.py

@@ -0,0 +1,36 @@
+#! /usr/bin/env python3
+from argparse import ArgumentParser
+from string import Template
+
+
+def main(file_path, substitutions, in_place, participant_ids):
+    with open(file_path) as f:
+        pbtxt = Template(f.read())
+
+    sub_dict = {"max_queue_size": 0}
+    sub_dict["participant_ids"] = participant_ids
+    for sub in substitutions.split(","):
+        key, value = sub.split(":")
+        sub_dict[key] = value
+
+    pbtxt = pbtxt.safe_substitute(sub_dict)
+
+    if in_place:
+        with open(file_path, "w") as f:
+            f.write(pbtxt)
+    else:
+        print(pbtxt)
+
+
+if __name__ == "__main__":
+    parser = ArgumentParser()
+    parser.add_argument("file_path", help="path of the .pbtxt to modify")
+    parser.add_argument(
+        "substitutions",
+        help="substitutions to perform, in the format variable_name_1:value_1,variable_name_2:value_2...",
+    )
+    parser.add_argument("--in_place", "-i", action="store_true", help="do the operation in-place")
+    parser.add_argument("--participant_ids", help="Participant IDs for the model", default="")
+    args = parser.parse_args()
+
+    main(**vars(args))

src/f5_tts/scripts/count_params_gflops.py

@@ -1,12 +1,13 @@
-import sys
 import os
+import sys
+
 
 sys.path.append(os.getcwd())
 
-from f5_tts.model import CFM, DiT
-
-import torch
 import thop
+import torch
+
+from f5_tts.model import CFM, DiT
 
 
 """ ~155M """

src/f5_tts/socket_client.py

@@ -1,10 +1,12 @@
-import socket
 import asyncio
-import pyaudio
-import numpy as np
 import logging
+import socket
 import time
 
+import numpy as np
+import pyaudio
+
+
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
 

src/f5_tts/socket_server.py

@@ -1,7 +1,6 @@
 import argparse
 import gc
 import logging
-import numpy as np
 import queue
 import socket
 import struct
@@ -10,6 +9,7 @@ import traceback
 import wave
 from importlib.resources import files
 
+import numpy as np
 import torch
 import torchaudio
 from huggingface_hub import hf_hub_download
@@ -18,12 +18,13 @@ from omegaconf import OmegaConf
 
 from f5_tts.infer.utils_infer import (
     chunk_text,
-    preprocess_ref_audio_text,
-    load_vocoder,
-    load_model,
     infer_batch_process,
+    load_model,
+    load_vocoder,
+    preprocess_ref_audio_text,
 )
 
+
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
 

src/f5_tts/train/README.md

@@ -1,5 +1,11 @@
 # Training
 
+Check your FFmpeg installation:
+```bash
+ffmpeg -version
+```
+If not found, install it first (or skip assuming you know of other backends available).
+
 ## Prepare Dataset
 
 Example data processing scripts, and you may tailor your own one along with a Dataset class in `src/f5_tts/model/dataset.py`.

src/f5_tts/train/datasets/prepare_csv_wavs.py

@@ -1,12 +1,13 @@
-import os
-import sys
-import signal
-import subprocess  # For invoking ffprobe
-import shutil
 import concurrent.futures
 import multiprocessing
+import os
+import shutil
+import signal
+import subprocess  # For invoking ffprobe
+import sys
 from contextlib import contextmanager
 
+
 sys.path.append(os.getcwd())
 
 import argparse
@@ -16,12 +17,10 @@ from importlib.resources import files
 from pathlib import Path
 
 import torchaudio
-from tqdm import tqdm
 from datasets.arrow_writer import ArrowWriter
+from tqdm import tqdm
 
-from f5_tts.model.utils import (
-    convert_char_to_pinyin,
-)
+from f5_tts.model.utils import convert_char_to_pinyin
 
 
 PRETRAINED_VOCAB_PATH = files("f5_tts").joinpath("../../data/Emilia_ZH_EN_pinyin/vocab.txt")
@@ -209,11 +208,11 @@ def save_prepped_dataset(out_dir, result, duration_list, text_vocab_set, is_fine
     out_dir.mkdir(exist_ok=True, parents=True)
     print(f"\nSaving to {out_dir} ...")
 
-    # Save dataset with improved batch size for better I/O performance
     raw_arrow_path = out_dir / "raw.arrow"
-    with ArrowWriter(path=raw_arrow_path.as_posix(), writer_batch_size=100) as writer:
+    with ArrowWriter(path=raw_arrow_path.as_posix()) as writer:
         for line in tqdm(result, desc="Writing to raw.arrow ..."):
             writer.write(line)
+        writer.finalize()
 
     # Save durations to JSON
     dur_json_path = out_dir / "duration.json"

src/f5_tts/train/datasets/prepare_emilia.py

@@ -7,20 +7,18 @@
 import os
 import sys
 
+
 sys.path.append(os.getcwd())
 
 import json
 from concurrent.futures import ProcessPoolExecutor
 from importlib.resources import files
 from pathlib import Path
-from tqdm import tqdm
 
 from datasets.arrow_writer import ArrowWriter
+from tqdm import tqdm
 
-from f5_tts.model.utils import (
-    repetition_found,
-    convert_char_to_pinyin,
-)
+from f5_tts.model.utils import convert_char_to_pinyin, repetition_found
 
 
 out_zh = {
@@ -183,6 +181,7 @@ def main():
     with ArrowWriter(path=f"{save_dir}/raw.arrow") as writer:
         for line in tqdm(result, desc="Writing to raw.arrow ..."):
             writer.write(line)
+        writer.finalize()
 
     # dup a json separately saving duration in case for DynamicBatchSampler ease
     with open(f"{save_dir}/duration.json", "w", encoding="utf-8") as f:

src/f5_tts/train/datasets/prepare_emilia_v2.py

@@ -0,0 +1,95 @@
+# put in src/f5_tts/train/datasets/prepare_emilia_v2.py
+# prepares Emilia dataset with the new format w/ Emilia-YODAS
+
+import json
+import os
+from concurrent.futures import ProcessPoolExecutor
+from importlib.resources import files
+from pathlib import Path
+
+from datasets.arrow_writer import ArrowWriter
+from tqdm import tqdm
+
+from f5_tts.model.utils import repetition_found
+
+
+# Define filters for exclusion
+out_en = set()
+en_filters = ["ا", "い", "て"]
+
+
+def process_audio_directory(audio_dir):
+    sub_result, durations, vocab_set = [], [], set()
+    bad_case_en = 0
+
+    for file in audio_dir.iterdir():
+        if file.suffix == ".json":
+            with open(file, "r") as f:
+                obj = json.load(f)
+                text = obj["text"]
+                if any(f in text for f in en_filters) or repetition_found(text, length=4):
+                    bad_case_en += 1
+                    continue
+
+                duration = obj["duration"]
+                audio_file = file.with_suffix(".mp3")
+                if audio_file.exists():
+                    sub_result.append({"audio_path": str(audio_file), "text": text, "duration": duration})
+                    durations.append(duration)
+                    vocab_set.update(list(text))
+
+    return sub_result, durations, vocab_set, bad_case_en
+
+
+def main():
+    assert tokenizer in ["pinyin", "char"]
+    result, duration_list, text_vocab_set = [], [], set()
+    total_bad_case_en = 0
+
+    executor = ProcessPoolExecutor(max_workers=max_workers)
+    futures = []
+    dataset_path = Path(dataset_dir)
+    for sub_dir in dataset_path.iterdir():
+        if sub_dir.is_dir():
+            futures.append(executor.submit(process_audio_directory, sub_dir))
+
+    for future in tqdm(futures, total=len(futures)):
+        sub_result, durations, vocab_set, bad_case_en = future.result()
+        result.extend(sub_result)
+        duration_list.extend(durations)
+        text_vocab_set.update(vocab_set)
+        total_bad_case_en += bad_case_en
+
+    executor.shutdown()
+
+    if not os.path.exists(f"{save_dir}"):
+        os.makedirs(f"{save_dir}")
+
+    with ArrowWriter(path=f"{save_dir}/raw.arrow") as writer:
+        for line in tqdm(result, desc="Writing to raw.arrow ..."):
+            writer.write(line)
+        writer.finalize()
+
+    with open(f"{save_dir}/duration.json", "w", encoding="utf-8") as f:
+        json.dump({"duration": duration_list}, f, ensure_ascii=False)
+
+    with open(f"{save_dir}/vocab.txt", "w") as f:
+        for vocab in sorted(text_vocab_set):
+            f.write(vocab + "\n")
+
+    print(f"For {dataset_name}, sample count: {len(result)}")
+    print(f"For {dataset_name}, vocab size is: {len(text_vocab_set)}")
+    print(f"For {dataset_name}, total {sum(duration_list) / 3600:.2f} hours")
+    print(f"Bad en transcription case: {total_bad_case_en}\n")
+
+
+if __name__ == "__main__":
+    max_workers = 32
+    tokenizer = "char"
+    dataset_dir = "/home/ubuntu/emilia-dataset/Emilia-YODAS/EN"
+    dataset_name = f"Emilia_EN_{tokenizer}"
+    # save_dir = os.path.expanduser(f"~/F5-TTS/data/{dataset_name}")
+    save_dir = str(files("f5_tts").joinpath("../../")) + f"/data/{dataset_name}"
+
+    print(f"Prepare for {dataset_name}, will save to {save_dir}\n")
+    main()

src/f5_tts/train/datasets/prepare_libritts.py

@@ -1,15 +1,17 @@
 import os
 import sys
 
+
 sys.path.append(os.getcwd())
 
 import json
 from concurrent.futures import ProcessPoolExecutor
 from importlib.resources import files
 from pathlib import Path
-from tqdm import tqdm
+
 import soundfile as sf
 from datasets.arrow_writer import ArrowWriter
+from tqdm import tqdm
 
 
 def deal_with_audio_dir(audio_dir):
@@ -60,6 +62,7 @@ def main():
     with ArrowWriter(path=f"{save_dir}/raw.arrow") as writer:
         for line in tqdm(result, desc="Writing to raw.arrow ..."):
             writer.write(line)
+        writer.finalize()
 
     # dup a json separately saving duration in case for DynamicBatchSampler ease
     with open(f"{save_dir}/duration.json", "w", encoding="utf-8") as f:

src/f5_tts/train/datasets/prepare_ljspeech.py

@@ -1,14 +1,16 @@
 import os
 import sys
 
+
 sys.path.append(os.getcwd())
 
 import json
 from importlib.resources import files
 from pathlib import Path
-from tqdm import tqdm
+
 import soundfile as sf
 from datasets.arrow_writer import ArrowWriter
+from tqdm import tqdm
 
 
 def main():
@@ -37,6 +39,7 @@ def main():
     with ArrowWriter(path=f"{save_dir}/raw.arrow") as writer:
         for line in tqdm(result, desc="Writing to raw.arrow ..."):
             writer.write(line)
+        writer.finalize()
 
     # dup a json separately saving duration in case for DynamicBatchSampler ease
     with open(f"{save_dir}/duration.json", "w", encoding="utf-8") as f:

src/f5_tts/train/datasets/prepare_wenetspeech4tts.py

@@ -4,15 +4,16 @@
 import os
 import sys
 
+
 sys.path.append(os.getcwd())
 
 import json
 from concurrent.futures import ProcessPoolExecutor
 from importlib.resources import files
-from tqdm import tqdm
 
 import torchaudio
 from datasets import Dataset
+from tqdm import tqdm
 
 from f5_tts.model.utils import convert_char_to_pinyin
 

src/f5_tts/train/finetune_cli.py

@@ -5,9 +5,9 @@ from importlib.resources import files
 
 from cached_path import cached_path
 
-from f5_tts.model import CFM, UNetT, DiT, Trainer
-from f5_tts.model.utils import get_tokenizer
+from f5_tts.model import CFM, DiT, Trainer, UNetT
 from f5_tts.model.dataset import load_dataset
+from f5_tts.model.utils import get_tokenizer
 
 
 # -------------------------- Dataset Settings --------------------------- #

src/f5_tts/train/finetune_gradio.py

@@ -1,14 +1,12 @@
 import gc
 import json
-import numpy as np
 import os
 import platform
-import psutil
 import queue
 import random
 import re
-import signal
 import shutil
+import signal
 import subprocess
 import sys
 import tempfile
@@ -16,21 +14,23 @@ import threading
 import time
 from glob import glob
 from importlib.resources import files
-from scipy.io import wavfile
 
 import click
 import gradio as gr
 import librosa
+import numpy as np
+import psutil
 import torch
 import torchaudio
 from cached_path import cached_path
 from datasets import Dataset as Dataset_
 from datasets.arrow_writer import ArrowWriter
 from safetensors.torch import load_file, save_file
+from scipy.io import wavfile
 
 from f5_tts.api import F5TTS
-from f5_tts.model.utils import convert_char_to_pinyin
 from f5_tts.infer.utils_infer import transcribe
+from f5_tts.model.utils import convert_char_to_pinyin
 
 
 training_process = None
@@ -138,6 +138,8 @@ def load_settings(project_name):
         "logger": "none",
         "bnb_optimizer": False,
     }
+    if device == "mps":
+        default_settings["mixed_precision"] = "none"
 
     # Load settings from file if it exists
     if os.path.isfile(file_setting):
@@ -176,50 +178,12 @@ def get_audio_duration(audio_path):
     return audio.shape[1] / sample_rate
 
 
-def clear_text(text):
-    """Clean and prepare text by lowering the case and stripping whitespace."""
-    return text.lower().strip()
-
-
-def get_rms(
-    y,
-    frame_length=2048,
-    hop_length=512,
-    pad_mode="constant",
-):  # https://github.com/RVC-Boss/GPT-SoVITS/blob/main/tools/slicer2.py
-    padding = (int(frame_length // 2), int(frame_length // 2))
-    y = np.pad(y, padding, mode=pad_mode)
-
-    axis = -1
-    # put our new within-frame axis at the end for now
-    out_strides = y.strides + tuple([y.strides[axis]])
-    # Reduce the shape on the framing axis
-    x_shape_trimmed = list(y.shape)
-    x_shape_trimmed[axis] -= frame_length - 1
-    out_shape = tuple(x_shape_trimmed) + tuple([frame_length])
-    xw = np.lib.stride_tricks.as_strided(y, shape=out_shape, strides=out_strides)
-    if axis < 0:
-        target_axis = axis - 1
-    else:
-        target_axis = axis + 1
-    xw = np.moveaxis(xw, -1, target_axis)
-    # Downsample along the target axis
-    slices = [slice(None)] * xw.ndim
-    slices[axis] = slice(0, None, hop_length)
-    x = xw[tuple(slices)]
-
-    # Calculate power
-    power = np.mean(np.abs(x) ** 2, axis=-2, keepdims=True)
-
-    return np.sqrt(power)
-
-
 class Slicer:  # https://github.com/RVC-Boss/GPT-SoVITS/blob/main/tools/slicer2.py
     def __init__(
         self,
         sr: int,
         threshold: float = -40.0,
-        min_length: int = 2000,
+        min_length: int = 20000,  # 20 seconds
         min_interval: int = 300,
         hop_size: int = 20,
         max_sil_kept: int = 2000,
@@ -250,7 +214,7 @@ class Slicer:  # https://github.com/RVC-Boss/GPT-SoVITS/blob/main/tools/slicer2.
             samples = waveform
         if samples.shape[0] <= self.min_length:
             return [waveform]
-        rms_list = get_rms(y=samples, frame_length=self.win_size, hop_length=self.hop_size).squeeze(0)
+        rms_list = librosa.feature.rms(y=samples, frame_length=self.win_size, hop_length=self.hop_size).squeeze(0)
         sil_tags = []
         silence_start = None
         clip_start = 0
@@ -304,8 +268,7 @@ class Slicer:  # https://github.com/RVC-Boss/GPT-SoVITS/blob/main/tools/slicer2.
             silence_end = min(total_frames, silence_start + self.max_sil_kept)
             pos = rms_list[silence_start : silence_end + 1].argmin() + silence_start
             sil_tags.append((pos, total_frames + 1))
-        # Apply and return slices.
-        ####音频+起始时间+终止时间
+        # Apply and return slices: [chunk, start, end]
         if len(sil_tags) == 0:
             return [[waveform, 0, int(total_frames * self.hop_size)]]
         else:
@@ -432,7 +395,7 @@ def start_training(
         fp16 = ""
 
     cmd = (
-        f"accelerate launch {fp16} {file_train} --exp_name {exp_name}"
+        f'accelerate launch {fp16} "{file_train}" --exp_name {exp_name}'
         f" --learning_rate {learning_rate}"
         f" --batch_size_per_gpu {batch_size_per_gpu}"
         f" --batch_size_type {batch_size_type}"
@@ -451,7 +414,7 @@ def start_training(
         cmd += " --finetune"
 
     if file_checkpoint_train != "":
-        cmd += f" --pretrain {file_checkpoint_train}"
+        cmd += f' --pretrain "{file_checkpoint_train}"'
 
     if tokenizer_file != "":
         cmd += f" --tokenizer_path {tokenizer_file}"
@@ -705,7 +668,7 @@ def transcribe_all(name_project, audio_files, language, user=False, progress=gr.
 
             try:
                 text = transcribe(file_segment, language)
-                text = text.lower().strip().replace('"', "")
+                text = text.strip()
 
                 data += f"{name_segment}|{text}\n"
 
@@ -814,7 +777,7 @@ def create_metadata(name_project, ch_tokenizer, progress=gr.Progress()):
             error_files.append([file_audio, "very short text length 3"])
             continue
 
-        text = clear_text(text)
+        text = text.strip()
         text = convert_char_to_pinyin([text], polyphone=True)[0]
 
         audio_path_list.append(file_audio)
@@ -833,9 +796,10 @@ def create_metadata(name_project, ch_tokenizer, progress=gr.Progress()):
     min_second = round(min(duration_list), 2)
     max_second = round(max(duration_list), 2)
 
-    with ArrowWriter(path=file_raw, writer_batch_size=1) as writer:
+    with ArrowWriter(path=file_raw) as writer:
         for line in progress.tqdm(result, total=len(result), desc="prepare data"):
             writer.write(line)
+        writer.finalize()
 
     with open(file_duration, "w") as f:
         json.dump({"duration": duration_list}, f, ensure_ascii=False)
@@ -1097,7 +1061,7 @@ def vocab_extend(project_name, symbols, model_type):
     return f"vocab old size : {size_vocab}\nvocab new size : {size}\nvocab add : {vocab_size_new}\nnew symbols :\n{vocab_new}"
 
 
-def vocab_check(project_name):
+def vocab_check(project_name, tokenizer_type):
     name_project = project_name
     path_project = os.path.join(path_data, name_project)
 
@@ -1125,7 +1089,9 @@ def vocab_check(project_name):
         if len(sp) != 2:
             continue
 
-        text = sp[1].lower().strip()
+        text = sp[1].strip()
+        if tokenizer_type == "pinyin":
+            text = convert_char_to_pinyin([text], polyphone=True)[0]
 
         for t in text:
             if t not in vocab and t not in miss_symbols_keep:
@@ -1230,8 +1196,8 @@ def infer(
     with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as f:
         tts_api.infer(
             ref_file=ref_audio,
-            ref_text=ref_text.lower().strip(),
-            gen_text=gen_text.lower().strip(),
+            ref_text=ref_text.strip(),
+            gen_text=gen_text.strip(),
             nfe_step=nfe_step,
             speed=speed,
             remove_silence=remove_silence,
@@ -1496,7 +1462,9 @@ Using the extended model, you can finetune to a new language that is missing sym
             txt_info_extend = gr.Textbox(label="Info", value="")
 
             txt_extend.change(vocab_count, inputs=[txt_extend], outputs=[txt_count_symbol])
-            check_button.click(fn=vocab_check, inputs=[cm_project], outputs=[txt_info_check, txt_extend])
+            check_button.click(
+                fn=vocab_check, inputs=[cm_project, tokenizer_type], outputs=[txt_info_check, txt_extend]
+            )
             extend_button.click(
                 fn=vocab_extend, inputs=[cm_project, txt_extend, exp_name_extend], outputs=[txt_info_extend]
             )

src/f5_tts/train/train.py

@@ -10,6 +10,7 @@ from f5_tts.model import CFM, Trainer
 from f5_tts.model.dataset import load_dataset
 from f5_tts.model.utils import get_tokenizer
 
+
 os.chdir(str(files("f5_tts").joinpath("../..")))  # change working directory to root of project (local editable)