Deep Learning in Music Informatics

Deep architectures and hierarchical feature learning are burgeoning research topics with a variety of computer science and engineering application areas. Noting the specific challenges and opportunities faced in working to make machines more musically intelligent, MARL is actively exploring these promising methods in music informatics research (MIR). Here we provide the slides of a recent jointly organized presentation by deep learning practitioners in MIR, a walk-through programming tutorial tailored to the interests of MIR researchers, and point to a selection of some of our published work to date.

Deep Learning in Music Informatics – Demystifying the Dark Art

Given a growing interest within the MIR community, Erik M. Schmidt (Pandora), Philippe Hamel (Google), and Eric J. Humphrey (MARL) joined forces to present a thorough review of deep learning to the attendees of ISMIR2013 in Curitiba, PR, Brazil.

Python Tutorial

After covering the “why” and “what” of deep learning, it can also be helpful to see the “how” as well. To these ends, we’ve assembled some deep learning examples specific to music informatics, complete with source code (Python) and pre-processed data to get you up and running quickly.

Selected Publications