Comparing Audio Features and Playlist Statistics for Music Classification
Résumé
In recent years, a number of approaches have been developed for theautomatic recognition of music genres, but also more specific categories (styles,moods, personal preferences, etc.). Among the different sources for building clas-sification models, features extracted from the audio signal play an important rolein the literature. Although such features can be extracted from any digitized musicpiece independently of the availability of other information sources, their extractioncan require considerable computational costs and the audio alone does not alwayscontain enough information for the identification of the distinctive properties of amusical category.In this work we consider playlists that are created and shared by music listenersas another interesting source for feature extraction and music categorisation. Themain idea is that the tracks of a playlist are often from the same artist or belong tothe same category, e.g., they have the same genre or style, which allows us to exploittheir co-occurrences for classification tasks. In the paper, we evaluate strategies forbetter genre and style classification based on the analysis of larger collections ofuser-provided playlists and compare them to a recent classification technique fromthe literature. Our first results indicate that an already comparably simple playlist-based classifiers can in some cases outperform an advanced audio-based classification technique.