Arabic Text-To-Speech Synthesis: A Preliminary Evaluation
Areej Al-Wabil, Centre for Human-Computer Interaction Design, City University, London, United Kingdom ; Hend Al-Khalifa, Southampton University, United Kingdom ; Wafa Al-Saleh, College of Education, Saudi Arabia
EdMedia + Innovate Learning, in Vancouver, Canada ISBN 978-1-880094-62-4 Publisher: Association for the Advancement of Computing in Education (AACE), Waynesville, NC
Text-To-Speech synthesis is used in software applications that enable access to text for visually-impaired and individuals with specific learning difficulties. In order to successfully incorporate Arabic synthetic speech into educational software, it is important to measure how well synthetic speech can be understood by end users. This paper evaluates Arabic text-to-speech synthesizers employed in five commercial Arabic screen readers. Two evaluators listened to vowelized Arabic words and determined whether the words were correctly pronounced by the synthesizers. The level of correctness for the synthesizers is in the range 89% to 94% of the vowelized words. The evaluators also listened to the same synthesizers speaking phrases in which there were non-vowelized words and determined whether they were correctly pronounced or not. The level of correctness for the synthesizers ranged from 87% to 93% of the non-vowelized words.
Al-Wabil, A., Al-Khalifa, H. & Al-Saleh, W. (2007). Arabic Text-To-Speech Synthesis: A Preliminary Evaluation. In C. Montgomerie & J. Seale (Eds.), Proceedings of ED-MEDIA 2007--World Conference on Educational Multimedia, Hypermedia & Telecommunications (pp. 4423-4430). Vancouver, Canada: Association for the Advancement of Computing in Education (AACE).
© 2007 Association for the Advancement of Computing in Education (AACE)