Content-based Metrics and Environments for Music Search Engines

Searching for appropriate music content with actual music search engines may be difficult and imprecise. Existing solutions adopt whether trivial interfaces, which only allow queries by title, author, etc. or very rich user interfaces, that require a lot of musical knowledge and hence are not suitable for the common user. Moreover, they usually rely on methods like string edit distances or functional approaches (as for the symbolic level) or simple features extraction methods (as for the audio level) that cannot provide structural similarity informations. Our recent theoretic results in metrics for music information retrieval rely on graph theory and provide new tools for extract characterizing structural quantities from musical data to be implemented within an efficient query environment.