Effects of speech codecs on a remote speaker recognition system using noval SAD

Abstract

This paper deals with the effects of speech codecs on remote texit-independent speaker recognition performance in VoIPapplications, considering three types of speech codec: PCM, DPCM and ADPCM conforming to International Telecommunications Union-Telecoms (ITU-T) recommendation used in telephony and VoIP (Voice over Internet Protocol). In order to improve the speaker recognition performance in noisy environment we propose a new robust speech activity detection algorithm (SAD) using "Adaptive Threshold", which can be simulated with speech wave files of TIMIT database submerged in an additive noise. Moreover, the speaker recognition system is based on Vector Quantization and Mel-Frequency Cepstral features extraction algorithm. Where, the feature extraction proceed after (for testing phase) and before (for training phase ) the speech is sending over communication channel. Therefore, the digital channels can introduce several types of degradation. To overcome the degradation of channel, a convolutional code is used as error-control coding with the AWGN channel, Finely, the best overall performance of speech codecs was observed for the PCM code in terùs of recognition rate accuracy and execution time.