search

Plachtaa / VITS-fast-fine-tuning

This repo is a pipeline of VITS finetuning for fast speaker adaptation TTS, and many-to-many voice conversion
Apache License 2.0
4.65k stars 698 forks source link

纯推理的运行需求 #552

Closed prometheus-alien closed 7 months ago

prometheus-alien commented 7 months ago

可不用考虑cuda版本，我本地cuda11.1推理没问题。

克隆库

git clone git@github.com:Plachtaa/VITS-fast-fine-tuning.git

运行requirement

cd VITS-fast-fine-tuning pip install -r requirements.txt

Build monotonic align

cd monotonic_align mkdir monotonic_align python setup.py build_ext --inplace cd ..

NumPy需要1.22以上

如果已安装1.22以下版本先卸载或者用新的虚拟环境

pip uninstall numpy pip install numpy==1.22

或

pip install numpy==1.23 `

命令行运行

python cmd_inference.py -m /home/Download/vits_model/Lisi/G_latest.pth -c /home/Download/vits_model/Lisi/config.json -o /home/Download/vits_result -l "简体中文" -t "轻轻的我走了， 正如我轻轻的来； 我轻轻的招手， 作别西天的云彩。" -s "Lisi" -ls 0.9 -on "李四_测试_语速0.9_2023-12-14"

LiaoYujia commented 4 months ago

你好，想问问cuda11.1 能够训练不？

prometheus-alien commented 4 months ago

你好，想问问cuda11.1 能够训练不？

不记得了 这个项目挺久没用了