TY  - JOUR
T1  - Recognition of Arabic Pronunciation of Numbers Using Neural Networks
AU - , N. Boukezzoula AU - , F. Djahli AU - , Y. Bouterfa 
JO  - Asian Journal of Information Technology
VL  - 5
IS  - 1
SP  - 89
EP  - 95
PY  - 2006
DA  - 2001/08/19
SN  - 1682-3915
DO  - ajit.2006.89.95
UR  - https://makhillpublications.co/view-article.php?doi=ajit.2006.89.95
KW  - Formant extraction
KW  -learning phase
KW  -recognition phase
KW  -back-propagation
KW  -vocal number
AB  - In this study we present an extraction method of some parameters contained in the Arabic pronunciation of numbers from 0 to 9 (sifr to tessea), which are then used as an input vector of the proposed neural networks.Globally, the methodology is concentrated about three important steps. The first step is the extraction of characteristics as the formants and their respective bandwidths of some interesting particular woofs. A learning step using two cycles for adjusting, computing and saving the weights of the neural networks interconnections. Finally a recognition step or test verifying the credibility of the system.
ER  -