Comparative Analysis of Performance Expression using Similarity Metrics based on Statistical Model and Musical Score Information

Abstract

There is a complex phenomenon in the performance expression of Western music. It is difficult even for performers themselves to describe the causality of “score directions” and “instrument operation” clearly and objectively. In this paper, we propose a method that can help to elucidate the factors contributing to performance expression. It is based on a technique to obtain the systematized structure of the statistical models. The model associates the feature quantities of performance expression and the combination of simple musical directions that can be surely retrieved from the score. In this method, degrees to which those directions contribute to the performance expression are also defined. It allows the application of standard multivariate analysis techniques for comparative analysis of the expressions between performances. Evaluation showed the utility of the proposed method in contributing to an objective understanding of the factors that performers emphasized for control of expression in their performances.