如何利用FunAudioLLM搭建语音对话系统

1.从GitHub上克隆项目

git clone --recursive https://github.com/FunAudioLLM/FunAudioLLM-APP.git

如果子模块克隆失败可以进入FunAudioLLM-APP目录运行以下命令

cd FunAudioLLM-APP
git submodule update --init --recursive

安装相关依赖项

pip install -r requirements.txt

2.配置cosyvoice环境并下载模型

cd cosyvoice
pip install -r requirements.txt

创建download,py文件，用于下载模型

# SDK模型下载
from modelscope import snapshot_download
snapshot_download('iic/CosyVoice-300M', local_dir='pretrained_models/CosyVoice-300M')
snapshot_download('iic/CosyVoice-300M-SFT', local_dir='pretrained_models/CosyVoice-300M-SFT')
snapshot_download('iic/CosyVoice-300M-Instruct', local_dir='pretrained_models/CosyVoice-300M-Instruct')
snapshot_download('iic/CosyVoice-ttsfrd', local_dir='pretrained_models/CosyVoice-ttsfrd')

运行download,py，下载文件

python download,py

3.配置sensevoice环境

pip install funasr

4.解决某些库版本冲突问题

pip install --upgrade pydantic
pip install --upgrade inflect

安装ffmpeg

sudo apt update
sudo apt install ffmpeg#Ubuntu系统

5.启动

cd voice_chat
sudo CUDA_VISIBLE_DEVICES="0" DS_API_TOKEN="YOUR-DS-API-TOKEN" python app.py >> ./log.txt

启动成功