/Digital_Life_Server

Yet another voice assistant, but alive.

Primary LanguagePythonMIT LicenseMIT

Digital Life Server

这是「数字生命」服务部分代码。包括与前端通信,语音识别,chatGPT接入和语音合成。
For other part of the project, please refer to:
Launcher 启动此服务器的图形界面。
UE Client 用于渲染人物动画,录音,和播放声音的前端部分。

Getting stuffs ready to roll:

Clone this repo

git clone https://github.com/zixiiu/Digital_Life_Server.git --recursive

Install prerequisites

  1. install pytorch

    pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
  2. install other requirements

    pip install -r requirements.txt
  3. Build monotonic_align
    This may not work that well but you know what that suppose to mean.

    cd "TTS/vits/monotonic_align"
    mkdir monotonic_align
    python setup.py build_ext --inplace
    cp monotonic_align/*.pyd .
  4. Download models
    百度网盘
    ASR Model:
    to /ASR/resources/models
    Sentiment Model:
    to /SentimentEngine/models
    TTS Model:
    to /TTS/models

Start the server

run-gpt3.5-api.bat