A Music Similarity Function Based On Time-Frequency Analysis and Pyramid Kernel

In this paper, content-based music similarityfunction from acoustic music signal in anefficient way is considered. This function is todetermine similarities among songs,particularly, a piece of input music signalcompared with storage music song’s signal intothe database and retrieve a whole music songaccording to the input query. Representing themusic signal having sparse nature isaccomplished by Matching Pursuit with timefrequency dictionaries. In order to match acandidate segment with the query segment, themusic signal similarity measure is performed bySpatial Pyramid Matching. Evaluation results onmusic similarity illustrate that our musicsimilarity function is better than such previousapproache