Information theory concepts applied to the analysis of rhythm in recorded music with recurrent rhythmic patterns

Abstract

Repeating patterns are essential in music understanding and data compression. This paper applies information theory concepts to rhythmic analysis of recorded percussion music. Downbeat detection is addressed via lossy coding of an accentuation feature under rate-distortion criteria, assuming the correct alignment produces the simplest explanation for the data. The resulting description is suitable to related tasks, e.g. assessing performances’ complexity and estimating the number of different rhythmic patterns played.