An analysis of note deviation manner in piano music

Abstract

Human music performance usually contains ‘‘out-of-score’’ features such as players’ physical characteristics and emotional expressions besides the on-score information (notes, rests, and various marks). Although those features are not written on scores, they have a very important role to add an artistic sense to the music. As one of the ‘‘out-of-score’’ features, the deviation manner of each note between the score and the human performance is focused on. In this study, among various genres and forms of music, F. Chopin’s mazurkas (a form of Polish folk dance), which have triple time and the accent on the second or third beat, are used. By limiting the music genres, it becomes simplified to compare the tendency of players. For the experimental dataset (three pianists, 51 mazurkas), after extracting the fluctuations (time differences) from the regular position of each note, the Fourier transform is applied to the sequence. As a result, a significant difference among the players and the playing modes is obtained, which show the temporal and the rhythm patterns. As a future work, by using the patterns, it is possible to create a virtual performer who has/imitates a certain player or a certain playing mode.