TEXT-TO-SPEECH SYNTHESIS ACCESS TO WEB-BASED DISTANCE EDUCATION RESOURCES: DESIGN AND IMPLEMENTATION

摘要

The thesis of this paper is that the combination of text mark-up language, such as eXtendible Mark-up Language (XML), Text-to-speech (TTS) synthesis, and Web technologies, would provide a more powerful technology for accessing Distance Education (DE) resource over the Internet. This work focuses on the design and implementation of the text markup language and TTS systems. We discuss the specification and design of computational mechanisms for realizing the various components of the TTS system using Yoruba as the case natural language. The design of the High Level (HLS) and Low Level (LLS) synthesis modules of the TTS system are discussed. An implementation based on Finite State Transducer (FST) Technology and Fuzzy logic technique is also demonstrated. Using subjective experiments, we are able to obtain 85% intelligibility and 70% naturalness accuracy for the TTS system. We are presently fine-tuning the design, using a prototype, and iteratively studying and improving its various components.