Automatic Melody Segmentation

Abstract

The work presented in this dissertation investigates music segmentation. In the field of Musicology, segmentation refers to a score analysis technique, whereby notated pieces or passages of these pieces are divided into “units” referred to as sections, periods, phrases, and so on. Segmentation analysis is a widespread practice among musicians: performers use it to help them memorise pieces, music theorists and historians use it to compare works, music students use it to understand the compositional strategies of a given composer or genre. In the field of Music Psychology it is posited that a similar type of analysis is performed by our auditory system when constructing mental representations of music. In fact, most theories consider segmentation to be a core listening mechanism, fundamental to the way humans recognise, categorise, and memorise music. Digital music files often lack a segmentation analysis. Automatising segmentation has the potential of improving (or even enabling) situations where computers are used to search, browse, visualise, or summarise digital music collections. Moreover, investigation and modelling segmentation could lead to a better understanding of how music is perceived by humans. In this dissertation we investigate segmentation via computer simulation. We focus on the analysis of melody. We provide a conceptual model of melodic segmentation, and use it to introduce and test four different segmenters. The results obtained by our automated segmenters extends previous research in the area, and allows us to foresee a not-so-distant future where manual and automatic segmentations will be indistinguishable.