Two stage speaker verification using self organising map and multilayer perceptron neural network

Abstract

In this paper a two stage speaker verification system is presented. The first stage contains a modified Self Organising Map (SOM) that filters speech data using cluster information extracted from three selected vowels for a claimed speaker. Filtered frames from the first stage are then fed into the second stage which consists of three Multi Layer Perceptron (MLP) networks; these networks acting as individual claimed speaker vowel verifiers. Sixty four Discrete Fourier Transform (DFT) spectrum components are used as the input feature vectors. The system provides a verification performance of 94.54% when evaluated using 50 speakers from the Centre for Spoken Language Understanding (CSLU2002) speaker verification database. © Springer-Verlag London Limited 2011.