Applying mathematical morphology for content-based music retrieval

Mikko Karvonen, Mika Laitinen, Kjell Lemström, Juho Vikman

Tutkimustuotos: Artikkeli kirjassa/raportissa/konferenssijulkaisussaKonferenssiartikkeliTieteellinenvertaisarvioitu


In this paper, we introduce a new methodology for contentbased
music retrieval purposes. We show how to apply the framework
of mathematical morphology for error tolerant approximate retrieval of
polyphonic, symbolically encoded music. To this end, we have developed
two algorithms based on the framework and carried out some experiments
to compare their performance against some well-known algorithms developed
earlier for related problems. In a typical query setting, our new
algorithms do not perform as fast as the rivaling algorithms in the experiment. However, the advantage of the novel algorithms is in the ease of
adjusting the error tolerance. In addition, as shown by our experiments,
in certain settings the new algorithms become faster than their existing
counterpart algorithms.
Otsikko7th International Symposium on Computer Music Modeling and Retrieval : Book of Papers
Julkaisupäivätammikuuta 2010
ISBN (elektroninen)978-84-614-0853-5
TilaJulkaistu - tammikuuta 2010
OKM-julkaisutyyppiA4 Artikkeli konferenssijulkaisuussa
TapahtumaInternational Symposium on Computer Music Modeling and Retrieval - Malaga, Espanja
Kesto: 21 kesäkuuta 201024 kesäkuuta 2010
Konferenssinumero: 7th


  • 113 Tietojenkäsittely- ja informaatiotieteet

Siteeraa tätä