2025-07-17 –, Forum Hall
“MIDI, those cheesy sounds from the 90s?" is an actual question some guy asked me at an AI conference one day. What an inspiring question! This talk flips the outdated view of MIDI as retro noise, showcasing it as a powerful format for representing and analyzing music. While some think it’s obsolete, MIDI remains the backbone of modern music production and a very active research topic in machine learning. This talk unpacks MIDI’s structure and demonstrates how Python libraries like mido, pretty_midi, and MidiTok turn it into a tool for research and creativity. From visualization to music generation, practical examples reveal the modern applications of symbolic music.
The takeaway? Python makes it easier than ever to explore MIDI’s potential and apply it to a wide range of musical and analytical tasks.
Beginner
Mateusz Modrzejewski, PhD - Software engineer, researcher, conference speaker, author and co-author of papers on music information retrieval and audio AI. Assistant Professor at the Institute of Computer Science of Warsaw University of Technology, where he leads an Audio Intelligence Lab. Previously at Apple (Music Machine Learning team, Apple Music). Has also worked with research and engineering teams of other Fortune 500 companies, providing AI solutions and analytics.
Apart from his scientific and engineering work, he is also an experienced touring musician, having performed for audiences of up to 150,000 people and having toured in Poland, China, Vietnam, the UK, Germany, Ukraine, Lithuania and Estonia, among others. Some of the artists he has played with include The Dumplings, Grubson, Marek Dyjak, Chłopcy Kontra Basia, Maria Sadowska, Pablopavo i Ludziki, Majka Jeżowska, Michał Milczarek Trio.