A novel text-independent speaker verification system using ant colony optimization algorithm

Abstract

Automatic speaker verification (ASV) has become increasingly desirable in recent years. This system in general, requires 20 to 40 features as input for satisfactory verification. In this paper, features size is reduced by Ant Colony Optimization (ACO) technique to increase the ASV performance. After feature reduction phase, feature vectors are applied to a Gaussian Mixture Model (GMM) which is a text-independent speaker verification Model. Experiments are conducted on a subset of TIMIT corpora. The results indicate that with the optimized feature set, the performance of the ASV system is improved. Moreover, the speed of verification is significantly increased because number of features is reduced over 73% which consequently decrease the complexity of our ASV system. © 2008 Springer-Verlag.