Class 28 includes a brief presentation by Kiki Gutierrez, a computer science professor on his research “Searching and Similarity:”
Class 28 Video: Feature Extraction and Machine Learning (II)
Readings:
- The original generalized toolkit for music feature extraction (Audio = JMir, Scores = JSymbolic): Cory McKay, Automatic Music Classification with jMIR (Ph.D. Dissertation: McGill University, 2010) (PDF). Look especially at chapter 4.
- Michael Cuthbert, Christopher Ariza, and Lisa Friedland, “Feature Extraction and Machine Learning on Symbolic Music using the music21 Toolkit,” (PDF) Proceedings of the International Symposium on Music Information Retrieval (2011), pp. 387–92. Note that this article was designed for music21 v.1, Python 2, and Orange v.2, so the syntax of some things there will no longer work, but the general ideas are still sound.
- Work with three UROPs that resulted in a publication on better feature extractors with MusicXML etc. that do not work on MIDI: Michael Cuthbert, Christopher Ariza, Jose Cabal-Ugaz, Beth Hadley, and Neena Parikh, “Hidden Beyond MIDI’s Reach: Feature Extraction and Machine Learning with Rich Symbolic Formats in music21” (PDF) Proceedings of the Neural Information Processing Systems Conference (Music and Machine Learning, Workshop 4) (2011).
The main ML library used in the video is Orange: https://orangedatamining.com/. Another library often used today is Scikit-Learn: https://scikit-learn.org/stable/.
File: Blue Danube Waltz (XML)
