Feature extracted from wavelet decomposition using biorthogonal Riesz basis for text-independent speaker recognition

摘要

A new speaker feature extracted from wavelet decomposition using biorthogonal Riesz bases is described. Biorthogonal Riesz bases can offer a significant computational advantage by reducing the dimensionality of the eigenvalue problem at a not square matrix. Our results have shown that these wavelet Riesz bases introduced better performance than the other wavelet transforms with respect to the percentages of recognition. (C) 2008 Elsevier Ltd. All rights reserved.