Transcribe music

[thewh1teagle]

Describe the feature

It will be great to be able to transcribe music/radio with Vibe.
This way we can easily transcribe songs or videos that have music.

we need to integrate vocal separation model (separate music and speech).
Currently we can open https://vocalremover.org -> upload the audio -> download the Vocal -> transcribe in Vibe.

[thewh1teagle]

https://github.com/Anjok07/ultimatevocalremovergui

This is the best open source vocal separator. but written in pytorch.
In theory it's possible to rewrite it with onnxruntime so it will be fast, efficient, and lightweight. but require some work.

[altunenes]

https://github.com/modelscope/ClearerVoice-Studio

could be useful. and it seems not too complicated compared the previous methods