High quality Arabic text-to-speech synthesis using unit selection

Abstract

This work aims to develop a high quality Arabic concatenative speech synthesis system based on unit selection. The original unit selection algorithm was modified to integrate more phonological, linguistic and contextual features in order to improve the selection cost calculation from one side, and more prosodic parameters for more exact concatenation cost estimation from the other side. The subjective assessments based on MOS (Mean Opinion Score) tests and visual comparison of waveforms show satisfactory results.