A Heuristic Approach for Music Segmentation

Music segmentation is an important and essential step for successful development of music information retrieval system, particularly for polyphonic music database. In the past, most researchers assume music is monophonic. If the music is polyphonic, there is not satisfyingly solved by existing known algorithms. In this paper, we base on the LBDM and propose a heuristic method to extract the representative fragments of melody from different tracks. The extracted represented fragments are sentence-like phrases that can server as indexes of music. Experiments show the satisfying results in the proposed method.