SO-VITS-SVC使用
https://zhuanlan.zhihu.com/p/630115251?utm_id=0
https://www.bilibili.com/read/cv22206231/
python版本,3.10
开个python虚拟环境
python -m venv
升级pip,否则会报错
用pip install -r requirements.txt按照依赖
打开webUI,python webUI.py
下载孙燕姿的模型,
模型文件,.pt
配置文件,.json
下载的配置文件有问题,会报错
Given groups=1, weight of size [xxx, 256, xxx], expected input[xxx, 768, xxx] to have 256 channels, but got 768 channels instead
需要把下面的改成768,
"gin_channels": 768,
"ssl_dim": 768,
加载模型,
会报没有dropout接口的错,
重新按照如下模块,
pip install --upgrade fastapi==0.84.0
pip install --upgrade gradio==3.41.2
pip install --upgrade pydantic==1.10.12
推理模型,
spleeter需要从新开个虚拟环境,会冲突
ffmpeg,别用brew安装,直接去下载可执行文件
推理报错,
torchaudio::sox_io_load_audio_file() expected a value of type 'str' for argument '_0' but instead found type 'posixpath'
修改代码,加一行