Sixth International Conference on Spoken Language Processing
(ICSLP 2000)

Beijing, China
October 16-20, 2000

A Unified Approach for Speech Synthesis and Speech Recognition Using Stochastic Markov Graphs

Matthias Eichner, Matthias Wolff, Rüdiger Hoffmann

Laboratory of Acoustics and Speech Communication, Dresden University of Technology, Germany

With the progress of speech synthesis towards the development of complete TTS systems, the databases of speech synthesizers obtain more and more similarity with databases of speech recognizers. This offers new possibilities in combining systems for speech synthesis and recognition. In a new project, we are developing a speech dialogue system with the synthesis and recognition components using unified databases. In this paper, we describe the aim of the project, the framework of the system and first results at the acoustic and the word level.


Full Paper

Bibliographic reference.  Eichner, Matthias / Wolff, Matthias / Hoffmann, Rüdiger (2000): "A unified approach for speech synthesis and speech recognition using stochastic Markov graphs", In ICSLP-2000, vol.1, 701-704.