Brian McFee

Quick followup: it turns out that we can't rely on `ifft(..., n=max_size)` to handle the array allocation for us. The phrasing of the documentation on this point is a little...

I dig it.

Do you think it makes sense to move this stuff out to a `librosa.filters` module? I can imagine a few things going here: mel, chroma, cqt, dct. `librosa.features` would mainly...

I've been thinking about this one again, now that the filters module is pretty stable. The main difficulty I see here is providing a clean interface. We now have the...

No, I was thinking the filter shape would be a pluggable parameter, eg: `f = filters.chroma(..., shape=filters.triangle)`. But for that to work, each of the filter shape functions needs to...

I'm not sure how what you're describing differs from what I said. At the end of the day, we'll still need a function to build a chroma filterbank. This will...

@lostanlen this all sounds great. The one thing I'd like to sort out is how to deal with parametrizing windows in linear or log space. I think it should be...

@lostanlen I was thinking about this one again, and since we merged #414 quite a while ago, it seems like the code's in good shape to add custom/asymmetric window functions...

Note: this issue is in a half-done state at the moment. #1405 provides the generic filter positioning, but does not (yet) implement a function to apply the transform. That'll be...

Coming back to this one after another zillion years: is there still any interest in expanding our collection of window functions, eg to include gammatones and asymmetric triangles? I'm envisioning...