F0 Contour Modeling for Arabic Text-to-Speech Synthesis Using Fujisaki Parameters and Neural Networks

Speech synthesis quality depends on its naturalness and intelligibility. Theseabstract concepts are the concern of phonology. In terms of phonetics, they aretransmitted by prosodic components, mainly the fundamental frequency (F0)contour. F0 contour modeling is performed either by setting rules or byinvestigating databases, with or without parameters and following a timelysequential path or a parallel and super-positional scheme. In this study, we optedto model the F0 contour for Arabic using the Fujisaki parameters to be trained byneural networks. Statistical evaluation was carried out to measure the predictedparameters accuracy and the synthesized F0 contour closeness to the naturalone. Findings concerning the adoption of Fujisaki parameters to Arabic F0contour modeling for text-to-speech synthesis were discussed.