TABLE OF CONTENTS The concept of a talking machine is completely embedded in a communication framework. A speech synthesis system provides an oral message that conveys the information that is expected by a receiver, most often a human being.. Under the assumption that the listener is co-operative, this section describes the linguistic and acoustic structures required for constructing a spoken message. It is essential for the system to provide an acoustic message that can be understood by the listener in the same way as the speaker intended it. In other words, the various steps in constructing the speech signal must minimize ambiguities in the interpretation of the spoken message between the transmitter and the receiver. The largest factor of complexity in a speech synthesis system is that the underlying mechanisms generating a spoken message are inherent to the cognitive process that governs the whole sequence of operations, from the mental representation to the control of the phonation organs.
A leading paradigm of the field considers the input of a text-to-speech synthesis system as a text which is constrained by the linguistic rules of a given languae [2].
The pronunciation of a text utterance starts with a transcription step. Indeed, we write by combining elements from a set of signs (the orthographic alphabet) and similarly, we speak by combining elementary acoustic signs. The description of this universe of elementary sounds by abstract signs is the phonemic alphabet. Phonemes are descriptive...
