Development of Techniques for the Computational Modelling of Harmony

Abstract

Abstract. This research is concerned with the development of representational and modelling techniques employed in the construction of statistical models of four-part harmony. Multiple viewpoint systems have been chosen to represent both surface and underlying musical structure, and it is this framework, along with Prediction by Partial Match (PPM), which will be developed during this work. Two versions of the framework are described, starting with the strictest possible application of multiple viewpoints and PPM, and then extending and generalising a little. Some implementation details are reported, as are some preliminary results. Key words: Harmony, statistical models, machine learning, evaluation 1