A faster training algorithm and genetic algorithm to recognize some of Arabic phonemes

الملخص EN

Neural network and Genetic algorithm are used in this work to recognize some of Arabic phonemes recorded by more than person in various age.

This work implemented in two basic stage, first one include speech signal process such as segment, analyses and determined the best samples which represent the basic feature to sound signal.

The second stage contain recognize operation which done by Liebenberg-Marquardt (LM) Algorithm which is one of the faster training algorithm in the Neural network used to determine number of hidden neuron with Genetic Algorithm.

A research tool has been implemented, using the Mat lab 6.5, sound forge5 and programming language pascal7.