From music theory to machine learning for evaluating generated music - Talk at UCSD

While travelling through Southern California, I stopped at the University of California, San Diego (UCSD), to give a talk on generating structured music. Professor Shlomo Dubnov organized a talk for a summer school currently going on at UCSD on July 15th.

Title: From music theory to machine learning for evaluating generated music

Abstract:
Automatically composing music can be seen as a combinatorial optimisation problem. We just have to find the right combination of notes that makes a musical piece sound "good". But what makes music sound good?

In the first part of our research, we use optimisation techniques to generate counterpoint pieces. Counterpoint is a formally defined style that can be evaluated using strict rules, which are defined in music theory. In the second part of the research, we break free from the restriction of a formally defined style and we explore how styles can be "learned" from existing pieces. Combining machine learning with optimisation has the added advantage that we can impose structural constraints and thus enforce a theme and long term coherence to a piece.