Study to Build a Holy Quran Text-To-Speech System

Abstract

Building a Text-To-Speech (TTS) system needs the study and the preparation of several modules. In this paper we present some of these parts to build a speech synthesis system for the Quran recitation. Firstly, we will describe the database building process, because of its importance, especially in concatenative speech synthesis methods. Secondly, a Grapheme to Phoneme Transcription (GPT) module will be presented. This latter was developed by elaborating a set of Tajweed and Arabic reading rules. They were also organized in a new way to gain some computational time. Those developed modules are evaluated by verifying the correct transcription and the database's comprehensiveness, which gives satisfactory results.