Machine Learning with Music Data
Using music data (such as MusicXML or some time-series variant of this) in Machine Learning raises all kinds of interesting challenges. It seems easy to uncover things that might have theoretical interest, but its another thing entirely to find something that might have musical meaning. This is where I keep data sets to explore these problems.

Keith Jarrett Piano Transcriptions

Musescore/MusicXML versions of these solos can be downloaded from here.
Some brief notes on ways to learn and practice these solos can be found here.



Recordings and compositions

This list is pretty imcomplete, will try and find time to organise my stuff and post here.


Mahler Symphony Midi Files

These files (sourced from here) can be opened in anything that can work with MIDI (such as MuseScore, Sibelius, Logic Pro, Music21 etc.) and easily converted to MusicXML. This format makes them really useful to explore different questions related to music analysis. Also note that I am currently (thru 2022) in the process of creating a database of MusicXML files of Nikolai Kapustin's works which I will post here soon.