Home // ICDT 2011, The Sixth International Conference on Digital Telecommunications // View article

New Adaptation Method Using Two-dimensional PCA for Speaker Verification

Authors:
Chunyan Liang
Xiang Zhang
Lin Yang
Li Lu
Yonghong Yan

Keywords: speaker recognition; 2DPCA; eigenvoice; SVM

Abstract:
In this paper, a new adaptation method based on two-dimensional principal component analysis is introduced into speaker recognition. In the method, mixture and dimension of mean vectors based on the Gaussian Mixture Models (GMMs) are differentiated, and the covariance matrix is computed dimension-wisely. The experiments are carried out on the core conditions of NIST 2008 speaker recognition evaluation data. The experimental results indicate that the 2DPCA-based method can achieve comparable performance to the conventional eigenvoice approach. Besides, the fusion of the two different systems can make significant performance improvement compared to the eigenvoice system alone, achieving relative reduction on EER between 7% and 25% for different test conditions.

Pages: 46 to 50

Copyright: Copyright (c) IARIA, 2011

Publication date: April 17, 2011

Published in: conference

ISSN: 2308-3964

ISBN: 978-1-61208-127-4

Location: Budapest, Hungary

Dates: from April 17, 2011 to April 22, 2011