Techniques for high quality Arabic speech synthesis

The paper proposes a diphone/sub-syllable method for Arabic Text-to-speech systems. The proposed approach exploits the particular syllabic structure of the Arabic words. For good quality, the boundaries of the speech segments are chosen to occur only at the sustained portion of vowels. The speech segments consists of consonantshalf vowels, half vowel-consonants, half vowels, middle portion of vowels, and suffix consonants. The minimum set consists of about 310 segments for classical Arabic.