Vowel recognition by a neural network using vocal tract area function

Abstract

A new system for vowel recognition by a neural network using the vocal tract area function is developed. A method by which the vocal tract area function is directly estimated from speech waves is presented. The effect of the adaptive inverse filter is given to autocorrelation coefficients directly, which reduces the computation time to about 30% of that necessary for the conventional method. A neural network algorithm based on the conjugate gradient method is presented. The recognition results are included to demonstrate the effectiveness of the method. >