Music Retrieval by Content

MUSIR - A Retrieval Model for Music

Pekka Salosaari and Kalervo Järvelin

Department of Information Studies
University of Tampere
P.O.Box 607
FIN-33101 TAMPERE, Finland

REFERENCE

Salosaari, P. & Järvelin, K. (1998). MUSIR - A Retrieval Model for Music. University of Tampere, Department of Information Studies, manuscript 13 p.

Abstract

Traditionally, information retrieval in music has been based on surrogates of music, i.e., bibliographic descriptions of music documents. This does not provide access to the essence of music, whether it is defined as the the musical idea represented in the score, the gestures of the performer playing an instrument or the resulting auditive phenomenon - the sound. In this paper we develop a retrieval model for music content. We develop representations for music content and music queries, a matching method for the representations and show that the model has desirable properties for the retrieval of music content. Our model captures representative and memorable features of music in a simple representation, supports inexact retrieval, and ranks retrieved music documents. The MUSIR retrieval model is based on filtering the MIDI representation of music and n-gram matching.

Return to Kal's home page.
Return to Kal's publication list.
Paluu Kallen kotisivulle.
Paluu Kallen julkaisuluetteloon.