Closed lminer closed 3 years ago
Where exactly? Or you mean to provide resampling?
I mean provide for resampling, i.e. if you want to evaluate tracks that are 16000 instead of 44100.
Right. Got it. But maybe this should be handled in https://github.com/sigsep/sigsep-mus-eval
Could be. Although dealing with it here might provide a little more flexibility. For example if you want to run inference on something in 16000, you can do it off the bat.
Good point. Will add it here then. I will pick resampy for quality reasons. I would recommend to use tf or pytorch native resampling though for training...
closed by #64
It would be nice to be able to set different sample rates