ChatTTS-完整使用指南

免費且超強大的 AI TTS，文字轉語音模型+工具，有許多語氣的控制，也可以很精準的寫程式控制效果，是RD眼中好用的Local端開源的TTS

內容目錄

特色說明

1.大規模的數據：10萬小時的訓練資料，現在開源的是4小時的版本

2.專用設計：專門對於對話情境、視頻介紹的情境所設計的模型

3.開源特性：可以很簡單的整合到你的WEB中

4.支持語氣：oral, laugh, break

安裝前準備

python 3.10

CUDA

GIT

gradio

安裝說明

github 複製

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

Shell
​x
 
git clone https://github.com/2noise/ChatTTScd ChatTTS

安裝依賴

pip install --upgrade -r requirements.txt

Shell
 
pip install --upgrade -r requirements.txt

執行 webui

python examples/web/webui.py

Shell
 
python examples/web/webui.py

利用 CLI

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

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

要整合在 python 程式碼中，可以安裝 PyPI

pip install ChatTTS
pip install git+https://github.com/2noise/ChatTTS
pip install -e .

Shell
 
pip install ChatTTSpip install git+https://github.com/2noise/ChatTTSpip install -e .

整合程式碼

###################################
# Sample a speaker from Gaussian.

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
)

###################################
# For sentence level manual control.

# use oral_(0-9), laugh_(0-2), break_(0-7) 
# to generate special token in text to synthesize.
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,
)

###################################
# For word level manual control.

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)
"""
In some versions of torchaudio, the first line works but in other versions, so does the second line.
"""
try:
    torchaudio.save("word_level_output.wav", torch.from_numpy(wavs[0]).unsqueeze(0), 24000)
except:
    torchaudio.save("word_level_output.wav", torch.from_numpy(wavs[0]), 24000)

Python
 
#################################### Sample a speaker from Gaussian.​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)​#################################### For sentence level manual control.​# use oral_(0-9), laugh_(0-2), break_(0-7) # to generate special token in text to synthesize.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,)​#################################### For word level manual control.​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)"""In some versions of torchaudio, the first line works but in other versions, so does the second line."""try:    torchaudio.save("word_level_output.wav", torch.from_numpy(wavs[0]).unsqueeze(0), 24000)except:    torchaudio.save("word_level_output.wav", torch.from_numpy(wavs[0]), 24000)