How to recommend customized music through analyzing playlists of users

摘要

PURPOSE: A customized music recommending method through the play list analysis of users for recognizing the taste of a user through contents-based music analysis is provided to supply a recommendation catalog to the user by making a recommendation list. CONSTITUTION: An MFCC(Mel Frequency Cepstral Coefficients) is extracted from musical data. Music models is constructed by using an HMM(Hidden Markov Models) method. Similarity between music models is calculated. The similar music is grouped based on music reproduction list which a user listened to in the past. The conformity with music groups is evaluated. The recommendation list is supplied to the user based on the evaluated result.