- Added transcription, key, tempo, and hierarchical segmentation evaluation
- Improved documentation and the sonify submodule
- Made mir_eval Python 3-compatible
- Added continuous integration and near-100% coverage
You can grab the latest version from pip or from our GitHub repository. Or, if you are allergic to Python, our mir_eval web service has been updated to 0.3 too, so you can run all of your eval without installing anything or writing any code.
We'd like to thank new contributors Nils Werner, Rachel Bittner, Stefan Balke, and Fabian-Robert Stöter. On that note, and while I have your attention, I'd like to emphasize our goal of making mir_eval a community endeavor: If you disagree with how a metric has been implemented or think you have found a bug, please open an issue, and if you have a new feature to add, pull requests are welcome!
See you all in NYC in August!
(Based on official announcement sent to the ISMIR community mailing list)