Modeling of intonation bearing emphasis for TTS-synthesis of Greek dialogues

TTS (text-to-speech) synthesis of neutral-style Greek with good intelligibility and quality was achieved some time ago. As a further step towards expanding the application domain of the TTS-system developed in our laboratory, the incorporation of emphasis into the speech used in man-machine dialogues according to their context has been studied recently. This paper reports the method applied to the analysis of intonation patterns, the results of this analysis, the algorithm established for the creation of desired intonation patterns and its implementation in the existing TTS-system for Greek.