Read more
The naturalness of the synthesized speech depends on many factors. The more complete the synthesizer considers all factors, the higher the naturalness of result. In this work are considered such components of language and speech as alphabet, grammar, system of phonemes, articulator basis, acoustic characteristics and statistical characteristics of texts. These components are presented as the parts of one whole, as there is in reality. The formal structure of the experimental synthesizer is described. It knows how to consider all components and how to change other participants with change in one of them. At the second part is described the method of the introduction of the measure, metric and order relation on the set of objects of arbitrary nature through their properties. For the statistical text analysis it is used contextual matrices. On the basis of these concepts is determined the allophonic base for the synthesizer. Are introduced many other useful concepts. The content of the book will interest all, who deal with synthesis or recognition of speech. It will interest students of the specialty of language and speech processing. Both the theorists and the practitioners.
About the author
Doctor of Technical Sciences, the member of the academic council of the Center for Georgian Language Technologies, Georgian technical university. He completed the Tbilisi State University and obtained master degree in mathematics. His dissertation was successfully defended in 2006 at Georgian Technical University.
