Trajectory of Fifths in Music Data Mining

Abstract

This article presents a new concept of a music piece’s content representation using the so-called trajectory of fifths. The herein proposed approach is an enhancement to the music data mining techniques based on the analysis of the music signature. Locations of consecutive points of the trajectory of fifths reflect the variability of the music signature over time. This is because their coordinates are defined by the resultant vectors representing consecutive music signatures. If properly quantified, the trajectories of fifths can be a source of valuable information about the variability of harmonic structure of a given piece of music. This article shows the potential applicability of the trajectory of fifths in the processes of music tonality (major/minor key) evaluation as well as music genre classification. The experiments which have been carried out, backed by statistical analysis, confirm the usability of the proposed approach.