search

nomadkaraoke / python-audio-separator

Easy to use vocal separation from CLI or as a python package, using a variety of amazing pre-trained models (primarily from UVR)
MIT License
390 stars 64 forks source link

Request to add Band-split RNN model! #75

Open piwawa opened 2 months ago

piwawa commented 2 months ago

image

This model is the SOTA model to split vocals on MUSDB18. In my daily use case, vocal separation is most commonly used.

Here are some inference projects.

https://paperswithcode.com/paper/music-source-separation-with-band-split-rnn#code

beveradb commented 2 months ago

PRs are of course welcome!

Have you actually tried any of the code you linked to? eg. https://github.com/amanteur/BandSplitRNN-Pytorch

piwawa commented 2 months ago

PRs are of course welcome!

Have you actually tried any of the code you linked to? eg. https://github.com/amanteur/BandSplitRNN-Pytorch

Here is colab test script i created, you can easily try it with your own audio!

https://colab.research.google.com/github/piwawa/BandSplitRNN-PyTorch/blob/main/BandSplitRNN_colab.ipynb

The test results are good for me.

https://github.com/piwawa/BandSplitRNN-PyTorch/tree/main/example