In beat tracking, a listener's experience of the tempo from a previous excerpt of a music piece is usually a good prediction of the tempo of the following excerpt in the same piece of music. Human beings have this ability to adaptively adjust his or her tap to synchronize with the tempo of music. In this paper, an adaptive learning approach based on maximum a posteriori (MAP) estimation is proposed to integrate the propagated knowledge from the previous excerpt and to infer the tempo. Our experiments on real musical signals show that: (1) the extracted tempo and beat using MAP are more robust and less sensitive to the window size of the analysis; and (2) the adaptive framework facilitates easy fusion using results and knowledge from different analysis schemes.
展开▼
