MPS detected, but output doesn't separate

tsurumeso / vocal-remover

Vocal Remover using Deep Neural Networks

MIT License

1.47k stars 215 forks source link

MPS detected, but output doesn't separate #173

Open duringleaves opened 4 months ago

duringleaves commented 4 months ago

I'm running it successfully on the CPU, but when I enable GPU (Mac M1, MPS - validated it detects torch.backends.mps) the output isn't split. The mixed file is spit out to the instrumentals while the vocals file is silent. Not sure where to start troubleshooting.

AndrewYatzkan commented 1 week ago

Same issue here, bumping!