MLP trained to separate problem speakers provides improved features for speaker identification

Abstract

In automatic speech recognition (ASR) the non-linear data prolection provided by a one hidden layer multilayer perceptron (MLP), trained to recognise phonemes, has previously been shown to provide feature enhancement which can substantially increase ASR performance, especially in noise. Previous attempts to apply an analogous approach to speaker identification have not succeeded in improving performance, except by combining MLP processed features with other features. We present test resultsfor the TIMIT database which show that the advantage of MJLP preprocessing for open set speaker identification increases with the number of speakers used to train the MiEP and that improved identification is obtained as this number increases beyond sixty. We also present a methodfor selecting the speakers used for MLEP training which further improves identification performance.