Brian McFee
Brian McFee
I have a prototype of this working-ish in a local branch, but there are quite a few bugs to work out that I don't have bandwidth for at the moment....
Circling back on this: I think my initial design with a custom `dBNormalizer` class doesn't quite work. The class was designed to map raw amplitude/power data through decibel scaling to...
> Thanks. Yes, let's disable the tests too. Done. > museval might be a good fit as its an implementation of the bsseval `images` (unlike `_sources`) algorithm but it might...
Side note: failures on 3.12 builds are due to matplotlib removeing BrokenBarHCollection. Will fix that in a separate PR.
> Is there a way to tell codecov not to worry about the separation file's lines so it hates us less? I've added separation to the omit section, but it...
> Side note: failures on 3.12 builds are due to matplotlib removing BrokenBarHCollection. Will fix that in a separate PR. Actually just rolled it into this one - all that...
Ok, I think this is good to merge now. All matplotlib issues are resolved and tests are workign out of the box. @craffel ?
Yerp, realized this was already approved. :shipit:
One nitpick that I'm noticing just now: I'm not 100% satisfied with the hatching on sus4 chords. The final segment in our generated image comparison test: https://github.com/mir-evaluation/mir_eval/blob/4cac1405a7cf428e27fa9698c348348607672987/tests/baseline_images/test_display/test_display_chord_nolabels.png is a sus4,...
Bumping minimal matplotlib to 3.6 so as to handle colormap registration.