pseeth

This definitely needs to be fixed, still!

the py3-networks branch of nussl is all Python 3. Can you verify that this bug disappears when using the Python 3 version of nussl?

The model path isn’t right anymore anyway. DeepClustering is in the middle of an overhaul to make it more general and better. That overhaul just needs documentation but it is...

Looking off the bat, it seems that there are things changing in effects.py here that aren't relevant to the rest of the PR? I'll CR the rest when ready but...

This seems logical, but our OverlapAdd/SegmentAndSeparate implementation actually had some issues when I tried it last. In the latest version of nussl I decided not to do it. But totally...

For actual implementation, we can just use `scipy.signal.convolve`, I think? We could do a GPU implementation via torch as well but I don't know if that's worth it.

Cool, I can get this going. @abugler, any thoughts?

One more note, I think this could go as a function that takes two AudioSignal objects in `mixing.py` perhaps? And the `AudioSignal.convolve` can call that function with `self` and `other`....

I think I'd like it to be in`mixing.py` with a hook in `audio_signal.py`.

The repo I'm working on is private, but Texify should have access to all the repositories my organization has access to, private or public, right?