SIR Mathematics Investigation Abstract

RECOGNITION OF MUSICAL STRUCTURES USING RECURRENCE QUANTIFICATION ANALYSIS

Presenter:

Sean Clarke, Illinois Mathematics and Science Academy

Advisor:

Dr. Charles Webber, Loyola University Medical Center

Abstract:

Automatic recognition of music structure has become increasingly important with the advent of the MP3 music collection. A recent release of Winamp, a popular digital music player, came with such a tool claiming to generate play lists similar to a selected song by analyzing its structure. Many different methods of music structure recognition have been developed, each having different strength and weaknesses. Despite a lot of research into the area, a computational understanding of music remains elusive.

Recurrence Quantification Analysis is a mathematical tool that characterizes the recurrent features of a dynamic system. RQA has several advantages over many other types of signal analysis which often do not perform well for these applications (eg. the Fourier Transform). RQA can be used to create a time series which serves as a meta-data that marks significant music sections of songs. It is hoped that this method of analysis will lead to a useful method for analyzing the structure of the song.