Evaluation of set-class similarity measures for tonal analysis

TitleEvaluation of set-class similarity measures for tonal analysis
Publication TypeMaster Thesis
Year of Publication2014
AuthorsHarley, N.
AbstractThis work explores the limitations of exiting approaches to computational modelling and description of tonality. Set class theory is presented as alternative or complement to existing approaches. Set class similarity is presented as a tool for the representation of set class information for structural analysis. A survey of set class similarity measures from the literature is conducted as well as a rendering of traditional musicological terminology in the language of set class theory. Six of the most suitable measures are chosen for further evaluation. An analysis methodology is outlined which emphasises systematicity and perceptual relevance. This methodology consists of a number of computational techniques and is used to analyse specific musical examples. The analytical potential of the similarity measures is evidenced through the reconstruction of basic music intuition and analysis using the proposed methodology.