Lyudovyk T. Information Technology for Individualized Text to Speech Synthesis

Thesis for Candidate degree in Technical Sciences: speciality 05.13.23 - systems and means of Artificial Intelligence. - International Research and Training Centre for Information Technologies and Systems, Kyjiv, 2006. The thesis is devoted to the development and implementation of information technology for Ukrainian text-to-speech synthesis with preservation of speakers voice and pronunciation peculiarities. The principles and steps of individual speech databases development are described. The acoustic-phonetic study of speech corpora is carried out and general regularities of Ukrainian speech at segmental and prosodic levels are discovered. General pronunciation models are deduced, namely transcribing, phrasing, pause and phoneme duration calculation, and intonation models. Automated means for individualizing general models are proposed and realised. Text-to-speech system architecture is established and corresponding software is developed. Intelligibility and naturalness of synthesized speech are tested and test results are provided.