mir-dataset-loaders
Add fma keys dataset
This adds support for the FMA Keys dataset, which is a new dataset for the evaluation of key detection containing 340 hours (5489 songs) of song-level key and mode annotations, spread across 17 genres.
@stellaywong thanks for contributing to mirdata! Just FYI, we're about to make a release for ISMIR in a couple of weeks. Do you have time to address this changes before? It would be great to include this loader in the release! If you don't have the time don't worry, but please let us know
Thank you for the reviews! I've made the changes and pushed the updates.
It looks like the PR build is stuck waiting for status to be reported? Do you know how we can kick start the build?
Hey @stellaywong, thanks for the update :) Sorry that I was afk for a few days. OK, that looks quite good, you are missing a few things that I am gonna list in my next review, but the loader looks good! Hope that you will have some time to do the final updates... thanks for the patience!
Hi @magdalenafuentes , @genisplaja
Apologies for the delay, this hung around for too long from our side. I've addressed the comments, would it be possible to re-open this PR so I can put up the changes?
I'm happy to send up a new PR with the changes if that works better too?
Hey @gandalfhz, thanks for the notice! Since we pushed up a pre-release for mirdata before the ISMIR conference, I cannot re-open this PR now. I would therefore kindly ask you to open a new PR with the FMA Keys dataloader and the requested changes from the review I left, and we can move on with that first, and then we will go for PR #682. Thanks, I'll keep an eye so we can have all of that merged before ISMIR in Korea :)