Rhythm Detection in Recorded Music

Abstract

My research concerns rhythm detection in recorded music. Using the matrix mathematics system Matlab, we analyzed the rhythmic structure of several musical selections and compared the results to our musical analysis by ear. Rhythm is a low frequency component of music, in contrast to pitch, which is a high frequency component. Our first task was to filter out the high frequencies in order to study rhythm. Our data reduction algorithm separates recorded music into frequency bands, of roughly half an octave each, and extracts the energy in each band. Once this was done, the Discrete Fourier Transform can identify the primary rhythmic content of each band. We found that the time signatures of the pieces we analyzed were characterized by the rhythmic relationship between low and high frequency bands. We have written Matlab algorithms that implement these ideas.

13 Figures and Tables

Cite this paper

@inproceedings{FlannickRhythmDI, title={Rhythm Detection in Recorded Music}, author={Joseph E. Flannick and Adlai Waksman} }