Zipf, neural networks and SVM for musical genre classification

Author

Dellandrea, Emmanuel ; Harb, Hadi ; Chen, Liming

Author_Institution

LIRIS, Ecole Centrale de Lyon, Ecully

fYear

2005

fDate

21-21 Dec. 2005

Firstpage

57

Lastpage

62

Abstract

We present in this paper audio classification schemes that we have experimented in order to perform musical genres classification. This type of classification is a part of a more general domain which is automatic semantic audio classification, the applications of which are more and more numerous in such fields as musical or multimedia databases indexing. Experimental results have shown that the feature set we have developed, based on Zipf laws, associated with a combination of classifiers organized hierarchically according to classes taxonomy allow an efficient classification

Keywords

audio databases; classification; database indexing; multimedia databases; music; neural nets; support vector machines; SVM; Zipf; audio classification; multimedia databases indexing; musical database; musical genre classification; neural networks; Algorithm design and analysis; Cepstral analysis; Cities and towns; Feature extraction; Frequency; Image analysis; Multimedia databases; Neural networks; Support vector machine classification; Support vector machines;

fLanguage

English

Publisher

ieee

Conference_Titel

Signal Processing and Information Technology, 2005. Proceedings of the Fifth IEEE International Symposium on

Conference_Location

Athens

Print_ISBN

0-7803-9313-9

Type

conf

DOI

10.1109/ISSPIT.2005.1577070

Filename

1577070