Similarity Measures for Rhythmic Sequences

João M. Martins
Marcelo Gimenes

Computer Music Research, Faculty of Technology University of Plymouth,
Plymouth, Devon PL4 8AA United Kingdom.

Jônatas Manzolli
Adolfo Maia Jr.

Departamento de Matemática Aplicada, IMECC
Núcleo Interdisciplinar de Comunicação Sonora (NICS), UNICAMP
13.081-970, Campinas, SP, Brazil.


This paper presents a new model for measuring similarity in a general Rhythm Space. Similarity is measured by establishing a comparison between subsequences of a given rhythm. We introduce the hierarchical subdivision of rhythm sequences in several levels, and compute a Distance Matrixfor each level using the "block distance". The information about the similarity of the rhythmic substructures is retrieved from the matrices and coded into a Similarity Coefficient Vector (SCV). Wealso present possibilities for the reduction to single values of similarity derived from the SCV. In addition, two applications of the formal model are presented, showing the potential for development using this approach.

Download paper: [pdf] 117860 bytes.