Development of an automatic phonetization system for Arabic text-to-speech synthesis

Abstract

The goal of this paper is to solve the phonetization task of the Arabic text dedicated to text-to-speech synthesis. In this paper, we will describe the steps we followed in order to develop an automatic phonetization system, and the various established rules of phonemic and phonetic transcription. Our text phonetization system is entirely based on a representation according to the Unicode standard, and is divided into two parts. The first part concerns the graphemephoneme transcription that combines use of a lexicon of exceptions, and a base of phonemic transcription rules. The second part concerns the generation of allophones by applying phonetic transcription rules to the phonemes obtained from the grapheme-phoneme transcription. The whole system performs an automatic phonetization for Arabic text-tospeech synthesis.