Brian McFee
Brian McFee
I ran into a funny bug just now, and wanted to get some opinions on whether this should be considered an mir_eval problem or not. [This check](https://github.com/craffel/mir_eval/blob/master/mir_eval/chord.py#L1387) in the directional...
In implementing #242 , I noticed that the chord submodule defines a derived exception class, but it's the only place where we do this. Everywhere else, primitive exception types (valueerror,...
Looks like we're hitting numerical precision issues again, sometimes, in the source separation tests on OSX. See https://github.com/craffel/mir_eval/actions/runs/8426164125/job/23106646394?pr=374 The tests passed previously (eg in #370), and since we're now seeing...
This was discussed in #370, but we agreed to migrate this repo to a dedicated organization. (`mir_eval`?) There are a couple of other things that will need to be squared...
This came up in #370, but the source separation metrics are presently implemented using the old / legacy scipy.fftpack API. As the modernization PR bumps the minimum scipy version requirement...
:wave: Seems like we're overdue for a new release here, if for no reason other than #352 (mir_eval currently doesn't work on the most recent numpy). I've gone ahead and...
Any interest in setting mir_eval up on rtd? It would be nice to have automatic doc builds for PRs, if nothing else.
This PR implements extension-reduction in the chord metrics. It's currently WIP, as it has uncovered some undocumented quirks in chord eval that I'd like to get settled before we merge...
**Warning, possibly bad ideas to follow.** mir_eval is generally pretty defensive about its inputs, and (i think?) all of the evaluators include at least some minimal validation of inputs for...
Most of the hierarchy measures are defined in terms of a *meet matrix*. This is a square matrix where each row and column corresponds to a frame (default, sampled at...