A METHOD FOR EXTRACTING A MUSICAL UNIT TO PHRASE MUSIC DATA IN THE COMPRESSED DOMAIN OF TWINVQ AUDIO COMPRESSION (WedPmOR5)

Author(s) : Motohiro Nakanishi (University of Electro-Communications, Japan) Michihiro Kobayakawa (University of Electro-Communications, Japan) Mamoru Hoshi (University of Electro-Communications, Japan) Tadashi Ohmori (University of Electro-Communications, Japan)

Abstract : A method for phrasing music data into musically meaningful pieces ({\it e.g.}, bar, phrase and so on) is an important function to analyze music data. To realize this function, we propose a method for extracting a musical unit of music data {\it in the compressed domain of TwinVQ audio compression (MPEG-4 audio)}. Our key idea is to extract a musical unit of music data from the sequence of the autocorrelation coefficients computed {\it in the encoding step of TwinVQ audio compression}. We call the sequence of the autocorrelation coefficients the ``autocorrelation sequence $\mbox{\boldmath$r$}$''. We use the $k$-th autocorrelation sequence \mbox{\boldmath$r$}_k$ ($k=1, \ldots, 20$) of music dat a for extracting a musical unit of music data. Firstly, we calculate the $j_{k}$-th autocorrelation coefficient $a_{k}î{j_k}$ of the $k$-th autocorrelation sequence $\mbox{\boldmath$r$}_k$ ($j_k=38,\ldots,208$; $k=1, \ldots, 20$). Secondary, for detecting peak of the sequence $(a_{k}î{38}, \ldots, a_{k}î{208})$, Laplacian filter is applied to the sequence, and then we obtain the order $p_k$ for which maximum differential coefficient is attained. Finally, we compute the musical unit of music data using the order $p_k$. To evaluate the performance of extracting the musical unit by our method, we collected $64$ music data and got the autocorrelation sequences by applying the TwinVQ encoder to each. And then we applied our extraction algorithm to each autocorrelation sequence. The results of experiments on $64$ music data show very good performance of extracting a musical unit for phrasing music data.

Menu