We have developed a text-to-speech synthesis system called "Cralinet" that generates very naturalsounding voices that can be used by various voice response services. This article describes effective instances of voice response services at a contact center and explains a statistical technique for estimating the accents of personal names and an improved technique for generating correct intonations based on statistical intonation evaluation as fundamental techniques of speech synthesis for contact centers. Finally, the future prospects of speech synthesis technology are also discussed.
|ジャーナル||NTT Technical Review|
|出版ステータス||Published - 2007 1 1|
ASJC Scopus subject areas
- Computer Science Applications
- Computer Networks and Communications
- Electrical and Electronic Engineering