We present a machine learning approach to the problem of identifying performers from their interpretative styles. In particular, we investigate how violinists express their view of the musical content in audio recordings and feed this information to a number of machine learning techniques in order to induce classifiers capable of identifying the interpreters. We apply sound analysis techniques based on spectral models for extracting expressive features such as pitch, timing, and amplitude representing both note characteristics and the musical context in which they appear. Our results indicate that the features extracted contain sufficient information to distinguish the considered performers, and the explored machine learning methods are capable of learning the expressive patterns that characterize each of the interpreters.

This work was supported by the Spanish Ministry of Science and Innovation under grant TIN2009-14247- C02-01 DRIMS Project.