Estudio comparativo y nuevas propuestas de tecnicas de parametrizacion de la señal de voz para el reconocimiento del habla

Abstract

A correct choice of voice signal modeling method is essential to obtain good results in automatic speech recogniton. In this paper, a comparative study betwen two speech signal models, Linear Prediction Coeficients and mel-cepstrum filter bank, is presented. On the other hand, a new parameterization method, obtained as a hybrid from previous models (LPC-mel-cepstrum), is presented too.