Unit Selection Approach for the Czech TTS System

摘要

The present paper deals with the beginning of the use of unit selection-based concatenative speech synthesis in the Czech Republic. The same approach which is employed in the Festival TTS was adopted, and Festival was used to do most of the work presented here. However, in contrast to Festival, not only phones but also triphones were used in the role of units. Moreover, our very modern and high quality Czech TTS system ARTIC was used for the concatenation and modification of the signal of selected candidates and the production of resulting speech.