2noise/ChatTTS

What it is

ChatTTS is a text-to-speech model designed specifically for dialogue scenarios such as LLM assistants. The repo contains the algorithm infrastructure and some simple examples. It supports multiple speakers and can predict and control fine-grained prosodic features including laughter, pauses, and interjections. The open-source model on HuggingFace is a 40,000-hour pre-trained model without SFT.

Who it's for

• Researchers and developers exploring dialogue-oriented text-to-speech for academic or educational purposes
• Developers building LLM assistant voice output who need English or Chinese speech synthesis
• Users who want fine-grained prosody control (laughter, pauses, interjections) in synthesized speech

Requirements

Setup

1. Clone the repo

   Clone the repository and enter the project directory.

   bashCopy

   git clone https://github.com/2noise/ChatTTS
   cd ChatTTS

2. Install requirements directly

   Install dependencies with pip.

   bashCopy

   pip install --upgrade -r requirements.txt

3. Or install from conda

   Create a conda environment and install the requirements.

   bashCopy

   conda create -n chattts python=3.11
   conda activate chattts
   pip install -r requirements.txt

4. Install from PyPI

   Install the stable version of the package.

   bashCopy

   pip install ChatTTS

Examples

python examples/web/webui.py

python examples/cmd/run.py "Your text 1." "Your text 2."

import ChatTTS
import torch
import torchaudio

chat = ChatTTS.Chat()
chat.load(compile=False) # Set to True for better performance

texts = ["PUT YOUR 1st TEXT HERE", "PUT YOUR 2nd TEXT HERE"]

wavs = chat.infer(texts)

for i in range(len(wavs)):
    try:
        torchaudio.save(f"basic_output{i}.wav", torch.from_numpy(wavs[i]).unsqueeze(0), 24000)
    except:
        torchaudio.save(f"basic_output{i}.wav", torch.from_numpy(wavs[i]), 24000)

rand_spk = chat.sample_random_speaker()
print(rand_spk) # save it for later timbre recovery

params_infer_code = ChatTTS.Chat.InferCodeParams(
    spk_emb = rand_spk, # add sampled speaker 
    temperature = .3,   # using custom temperature
    top_P = 0.7,        # top P decode
    top_K = 20,         # top K decode
)

params_refine_text = ChatTTS.Chat.RefineTextParams(
    prompt='[oral_2][laugh_0][break_6]',
)

wavs = chat.infer(
    texts,
    params_refine_text=params_refine_text,
    params_infer_code=params_infer_code,
)

text = 'What is [uv_break]your favorite english food?[laugh][lbreak]'
wavs = chat.infer(text, skip_refine_text=True, params_refine_text=params_refine_text,  params_infer_code=params_infer_code)

Pros & cons

Images