Improvements of Audio-Based Music Similarity and Genre Classificaton

Audio-based music similarity measures can be used to automatically generate playlists or recommendations. In this paper the similarity measure that won the ISMIR’04 genre classification contest is reviewed. In addition, further improvements are presented. In particular, two new descriptors are presented and combined with two previously published similarity measures. The performance is evaluated in a series of experiments on four music collections. The evaluations are based on genre classification, assuming that very similar tracks belong to the same genre. On two collections the improvements lead to a substantial performance increase.