Audio Segmentation by Singular Value Clustering

This paper presents a statistical approach to sound texture modeling based on a singular value analysis of spectral features or an eigenvector analysis of their similarity matrix. Using dimension reduction techniques we perform grouping of the signal into similar sounding audio segments that are recurrent in time. The method allows an automatic segmentation of audio signal into larger groups of similar sounding audio objects and can be used for visualization purposes, audio texture synthesis and creative audio manipulations. We present a principled approach that brings methods such as audio similarity analysis and spectral audio basis representations into one framework.