DocumentCode :
705343
Title :
Sample iterative likelihood maximization for speaker verification systems
Author :
Garcia, Guillermo ; Eriksson, Thomas
Author_Institution :
Dept. of Signals & Syst, Chalmers Univ. of Technol., Goteborg, Sweden
fYear :
2010
fDate :
23-27 Aug. 2010
Firstpage :
596
Lastpage :
900
Abstract :
Gaussian Mixture Models (GMMs) have been the dominant technique used for modeling in speaker recognition systems. Traditionally, the GMMs are trained using the Expectation Maximization (EM) algorithm and a large set of training samples. However, the convergence of the EM algorithm to a global maximum is conditioned on proper parameter initialization, a large enough training sample set, and several iterations over this training set. In this work, a Sample Iterative Likelihood Maximization (SILM) algorithm based on a stochastic descent gradient method is proposed. Simulation results showed that our algorithm can attain high log-likelihoods with fewer iterations in comparison to the EM algorithm. A maximum of eight times faster convergence rate can be achieved in comparison with the EM algorithm.
Keywords :
Gaussian processes; convergence; expectation-maximisation algorithm; gradient methods; mixture models; sampling methods; speaker recognition; EM algorithm; GMM; Gaussian mixture model; convergence rate; expectation maximization algorithm; high log-likelihood; sample iterative likelihood maximization; speaker recognition system; speaker verification system; stochastic descent gradient method; training sample set; Clustering algorithms; Convergence; Databases; Signal processing algorithms; Speaker recognition; Speech; Training;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Signal Processing Conference, 2010 18th European
Conference_Location :
Aalborg
ISSN :
2219-5491
Type :
conf
Filename :
7096616
Link To Document :
بازگشت