Boltzmann machine neural network for Arabic speech recognition

Abstract EN

Boltzmann machine neural network has been used to recognizc the Arabic specch.

Fast Fourier transformation algorithm has been used to extract spectral features from an a coustic signal.

The spectral feature size is reduced by scries of operations in order to make it salable as input for a neural network which is used as a recogni/.er by Boltzmann Machine Neural network which has been used as a recognizer for phonemes .

٨ training set consist ofa number of ،Arabic phoneme repesentations, is used to train the neural network.

The neural network rccognized Arabic.

After Boltzmann Machine Neural network training the system with few selected Arabic phonemes, the results came out to be very encouraging .