Speaker verification using sequence discriminant support vector machines
- 22 February 2005
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Speech and Audio Processing
- Vol. 13 (2) , 203-210
- https://doi.org/10.1109/tsa.2004.841042
Abstract
This paper presents a text-independent speaker verification system using support vector machines (SVMs) with score-space kernels. Score-space kernels generalize Fisher kernels and are based on underlying generative models such as Gaussian mixture models (GMMs). This approach provides direct discrimination between whole sequences, in contrast with the frame-level approaches at the heart of most current systems. The resultant SVMs have a very high dimensionality since it is related to the number of parameters in the underlying generative model. To address problems that arise in the resultant optimization we introduce a technique called spherical normalization that preconditions the Hessian matrix. We have performed speaker verification experiments using the PolyVar database. The SVM system presented here reduces the relative error rates by 34% compared to a GMM likelihood ratio system.Keywords
This publication has 11 references indexed in Scilit:
- Speaker identification via support vector classifiersPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- Testing with the YOHO CD-ROM voice verification corpusPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- A hybrid GMM/SVM approach to speaker identificationPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- Learning the decision function for speaker verificationPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- Evaluation of kernel methods for speaker verification and identificationPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- Extractive summarization of voicemail using lexical and prosodic feature subset selectionPublished by International Speech Communication Association ,2001
- The NIST speaker recognition evaluation – Overview, methodology, systems, results, perspectiveSpeech Communication, 2000
- A Tutorial on Support Vector Machines for Pattern RecognitionData Mining and Knowledge Discovery, 1998
- Speaker identification and verification using Gaussian mixture speaker modelsSpeech Communication, 1995
- Mixture Densities, Maximum Likelihood and the EM AlgorithmSIAM Review, 1984