Residual separation #606
-
|
Is there a way to get the residual audio from an inference? |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments
-
|
Do you use your own pretrained model? You would need the model to return the unseparated audio from the model but a simple class wrapper with a forward like # Pseudo code
def forward(self, x):
estimate = self.model(x)
remaining = x - estimate
return torch.cat([estimate, remaining])would do the trick I guess. |
Beta Was this translation helpful? Give feedback.
-
|
I used a DCUNET model and trained with my own dataset. |
Beta Was this translation helpful? Give feedback.
Do you use your own pretrained model?
You would need the model to return the unseparated audio from the model but a simple class wrapper with a forward like
would do the trick I guess.