Closed jorgeauh closed 1 year ago
IIRC there is a 6-stem model with support for piano and guitar which you can access if you run the Demucs CLI directly instead of using StemRoller: https://github.com/facebookresearch/demucs Unfortunately this model isn't included with StemRoller, and won't be until I'm able to implement model download/management within StemRoller (so people can change which model is used for splitting). Also, the documentation states that piano extraction does not usually work well... I don't think Demucs comes with pre-trained models for synths unfortunately, so can't help you there.
Thanks for sharing this amazing project, only for curiosity Is it possible that we can get the isolation of guitars, synths and pianos individually instead of get mixed in a single audio file? Thank you in advanced