Eur. Phys. J. Special Topics 222, 473-485 (2013)
https://doi.org/10.1140/epjst/e2013-01853-8

Regular Article

Quantitative characterisation of audio data by ordinal symbolic dynamics

¹ Max-Planck-Institut für extraterrestrische Physik, Giessenbachstr. 1, 85748 Garching, Germany
² Centro de Investigación Operativa, Universidad Miguel Hernandez, Avda. de la Universidad s/n, 03202 Elche, Spain

^a e-mail: tsa@mpe.mpg.de

Received: 27 March 2013
Revised: 25 April 2013
Published online: 25 June 2013

Abstract

Ordinal symbolic dynamics has developed into a valuable method to describe complex systems. Recently, using the concept of transcripts, the coupling behaviour of systems was assessed, combining the properties of the symmetric group with information theoretic ideas. In this contribution, methods from the field of ordinal symbolic dynamics are applied to the characterisation of audio data. Coupling complexity between frequency bands of solo violin music, as a fingerprint of the instrument, is used for classification purposes within a support vector machine scheme. Our results suggest that coupling complexity is able to capture essential characteristics, sufficient to distinguish among different violins.