Unit Selection Model in Arabic Speech Synthesis

In text-to-speechsynthesis the speech unit is segmented and extracted from natural speech, coded, classified, labelled and stored in the inventory of units. To generate a smooth synthesized speech the unit selection plays an important and crucial role. A multiple-instances approach of speech unit is suggested. A set of speech units is prepared for each speech unit type, taking into consideration target unit structure, the preceding and the succeeding syllables, and position in the utterance. In this paper, a diacritic Arabic Text-To-Speech system is described. The goal of this paper is to get intelligible, high quality speech synthesis based on unit selection using a syllables model.