> I suspect naively it shouldn't be too hard to decode an existing recording of music into the notes played. What you expect not to be too hard to do, is actua

> I suspect naively it shouldn't be too hard to decode an existing recording of music into the notes played. What you expect not to be too hard to do, is actually the 'holy grail' of the field - Algorithmic Melody Transcription is the kind of thing people write entire papers on[0]. [0]: https://www.ee.columbia.edu/~dpwe/pubs/PolEFGSO07-meleval.pdf https://www.ee.columbia.edu/~dpwe/pubs/PolEFGSO07-meleval.pd... There are a few toy examples if you search "algorithmic music transcription github," one of the better-known ones is [1] (but it's limited to piano music). They're all pretty nascent. It's a real hard problem :) [1]: https://github.com/rachhshruti/automatic-music-transcription https://github.com/rachhshruti/automatic-music-transcription