Talk at Imperial College London on generating structured music with local search optimisation and machine learning
I'll be giving a talk on generating structured music with local search optimisation and machine learning. The seminar is organised by the Department of Computing, at Imperial College London and will take place on the 14th of October at 15:30.
Department of Computing, Imperial College London
Generating structured music with local search optimisation and machine learning
Dorien Herremans, PhD (MSCA Fellow) Queen Mary University of London
Date/time: 14 Oct 2015, 15:30 - 16:30
Venue: Huxley LT 144, South Kensington Campus
Abstract: Many state of the art music generation/improvisation systems generate music that sounds good on a note-to-note level. However, these compositions often lack long term structure or coherence. By looking at generating music as an optimisation problem, this research overcomes this problem and generates music that has a larger structure. A powerful variable neighbourhood search algorithm (VNS) was developed, which is able to generate a range of musical styles based on its objective function, whilst constraining the music to a structural template. In the first stage of the project, an objective function based on rules from music theory was used to generate counterpoint.
In this research, a machine learning approach is combined with the VNS in order to generate structured music for the bagana, an Ethiopian lyre. Different ways are explored in which a Markov model can be used to construct quality metrics that represent how well a fragment fits the chosen style (e.g. music for bagana). Current research that aims to extend the objective function with models such as recursive neural networks is also briefly discussed. The approach followed in this research allows us to combine the power of machine learning methods with optimisation algorithms.