Evaluation of Distance Measures Between Gaussian Mixture Models of MFCCs

Jesper Højvang Jensen, Dan P. W. Ellis, Mads Græsbøll Christensen, Søren Holdt Jensen

Research output: Contribution to book/anthology/report/conference proceedingArticle in proceedingResearchpeer-review

39 Citations (Scopus)
1193 Downloads (Pure)

Abstract

In music similarity and in the related task of genre classification, a distance measure between Gaussian mixture models is frequently needed. We present a comparison of the Kullback-Leibler distance, the earth movers distance and the normalized L2 distance for this application. Although the normalized L2 distance was slightly inferior to the Kullback-Leibler distance with respect to classification performance, it has the advantage of obeying the triangle inequality, which allows for efficient searching.
Original languageEnglish
Title of host publicationProceedings of the 8th International Conference on Music Information Retrieval
Number of pages2
PublisherAustrian Computer Society
Publication date2007
Pages107-108
ISBN (Print)978-3-85403-218-2
Publication statusPublished - 2007
EventInternational Conference on Music Information Retrieval - Vienna, Austria
Duration: 23 Sep 200727 Sep 2007
Conference number: 8

Conference

ConferenceInternational Conference on Music Information Retrieval
Number8
CountryAustria
CityVienna
Period23/09/200727/09/2007

    Fingerprint

Cite this

Jensen, J. H., Ellis, D. P. W., Christensen, M. G., & Jensen, S. H. (2007). Evaluation of Distance Measures Between Gaussian Mixture Models of MFCCs. In Proceedings of the 8th International Conference on Music Information Retrieval (pp. 107-108). Austrian Computer Society.