Interspeech'2005 - Eurospeech
Aiming at F0 control for communicative speech synthesis, the relationship between word attributes and F0 characteristics was analyzed. By analyzing one-phrase utterances in conversational situations, we studied correlations between word attributes defined by their impressions and prosodic control characteristics. For word attribute description, we used three dimensions in perceptual impressions, confident-doubtful, allowable-unacceptable and positive-negative obtained from our previous studies on one syllable utterances of "n". The result showed that F0 height, F0 dynamic patterns and duration could be consistently controlled by the word attributes. The positive-(negative) can be controlled by F0 height, while confident-doubtful, allowable-unacceptable were reflected in F0 dynamic patterns and duration. These results confirmed the usefulness of word attributes in communicative speech synthesis.
Bibliographic reference. Greenberg, Yoko / Tsuzaki, Minoru / Kato, Hiroaki / Sagisaka, Yoshinori (2005): "Communicative speech synthesis using constituent word attributes", In INTERSPEECH-2005, 517-520.