search

SociallyIneptWeeb / AICoverGen

A WebUI to create song covers with any RVC v2 trained AI voice from YouTube videos or audio files.
MIT License
1.05k stars 249 forks source link

Vocal/Instrumentation Separation Takes Too Much time on Colab Pro #93

Open Adwait10-prog opened 9 months ago

Adwait10-prog commented 9 months ago

When trying to separate even a small audio , the runtime is too much for the vocal separation and after that takes time for the next processes too, it was running fine till yesterday

Screenshot 2023-12-15 at 4 16 44 PM Screenshot 2023-12-15 at 4 22 28 PM
rhizomat commented 9 months ago

Same here. GPU doesn't seem to supported (anymore).

andrixpan commented 9 months ago

downgrade cuda version to 11.8, if you now have cuda 12.2 maybe sparate instrument file take a long, because render without gpu, but render with cpu. problem have incorrect cuda version, or onnx runtime doesnt support cuda 12.2 if you add the new command on colab !nvcc --version you will show cuda version on you use.. sorry my bad english no translate.

andrixpan commented 9 months ago

picture 3 picture 2 picture 1

onnxruntime only support CUDA 11.8 doesnt support CUDA 12.2

Porter-smith commented 9 months ago

I had some issue i saw that someone made a v2 collab of it on ai hub discord and fixed for me:

NO UI: https://colab.research.google.com/drive/1u1brjK8IZt647UsbZuGYfW29oFM2I4tk?usp=sharing W/ UI https://colab.research.google.com/github/hinabl/AICoverGen-Colab/blob/main/Hina_Mod_AICoverGen_colab.ipynb#scrollTo=NEglTq6Ya9d0