Speech Synthesis

Abstract

Speech synthesis is the process of producing an acoustic signal by controlling a model of speech production with a set of parameters. If the model and parameters are sufficiently accurate then the production of intelligible synthetic speech should be possible. There are two basic approaches in modelling the speech production process. One is a direct approach which attempts to model the system in detail. This is commonly referred to as articulatory speech synthesis and attempts to directly model the motion of the speech articulators as well as the generation and propagation of sound inside the vocal tract. This approach is still the subject of research and although it seems to have the potential for producing the most natural-sounding speech in the long term, it has not as yet been as successful as approaches that attempt to simply copy the frequency response characteristic of the vocal tract. Such approaches are based on the source/filter model developed in chapter 1 and are collectively known as terminal-analogue synthesisers since they use a system which is an analogue of the speech production mechanism from a terminal point of view.