Neural speech synthesis system for Arabic language using CELP algorithm

Speech is the most natural and widespread form of human communication. That is why speech synthesis has interested researchers for decades. It turns out that developing an unlimited text-to-speech system is an enormous task. The traditional methods (synthesis by rule and synthesis by concatenation of pre-recorded sounds) used for this have not given good results. In such a situation, neural networks (NNs) have the potential to give better results thanks to their property of interpolation and their capacity of generalisation. We present a synthesis system for the Arabic language. The choice of parameters which will be used to drive the NN is very important and have an effective influence on the quality of produced speech. Work has been done to evaluate different methods based on linear predictive coding; the resulting speech was machine-like and not intelligible. We suggest to use CELP to drive the NN, which provides high quality speech.