Application for TESPAR coding study and speaker recognition experiments [articol]

Abstract

TESPAR coding (Time Encoding Signal Processing and Recognition) represents an effectiveness alternative to the other common methods (Dynamic Time Warping, Vector Quantization, Hidden Markov Models, etc.) used for speech/speaker recognition. The important advantage of this method is the time processing of signal with a decrease of two orders of magnitude of the computational requirements. This work presents an application for TESPAR coding study allowing speaker recognition experiments using parallel neural networks architecture.