Vincent Lostanlen
Vincent Lostanlen
Bonus points if this can be done without loading any metadata, but simply by parsing track_ids
@rabitt what do you mean, hardcoded? as constants? directly in the code of the module or in a `JSON` file? if the former, i'm a bit concerned about import time....
@rabitt i'm thinking about this a little bit more. My opinion is that this feature is overkill for small datasets like TinySOL (16 instruments), MSDB (8 instruments) or GTZAN Genre...
Helo @pseeth. Thank you so much for considering this. VocalSet would be a great addition to mirdata, indeed. We don't have a ton of freely downloadable vocal audio in mirdata...
I would prefer to make it in two stages: first, an API change, and then a feature enhancement otherwise this issue is never going to be closed, given how many...
As you prefer. It would be good to delimit the scope of this issue then: what belongs to v0.2, v0.3, and v1.0 This is a feature request so it cannot...
Yes. This message in #322 describes how I did it in Julia. It ended up being very little code per wavelet, and was highly extensible. But I had to implement...
@MuawizChaudhary there's Fortran contiguity and C contiguity. All of this can be inspected in `x.flags` for `x` a NumPy array.
customizing the algorithm is nice but i don't see why it should be stored as a mutable field of `self`?
``` from kymatio.torch import Scattering2D S = Scattering2D() scattering2d_breadthfirstsearch(S, x) ``` saved you one line ☺️