ISCA Archive Interspeech 2005
ISCA Archive Interspeech 2005

A flexible and integrated interface between speech recognition, speech interpretation and dialog management

Robert Lieb, Matthias Thomae, Günther Ruske, Daniel Bobbert, Frank Althoff

This paper presents an integrated interface between speech recognition, speech interpretation and dialog control intended for spoken dialog systems coping with natural speech input. During the system design phase the interface co-ordinates corpus acquisition and annotation, grammar development and the construction of stochastic hierarchical language models. During system runtime, it links together speech recognition and interpretation by efficient one-stage decoding of semantic trees, from which semantic content can easily be extracted. To gain robustness, the interface provides a way to interpret semantic confidences estimated during the decoding process. Furthermore, the dialog control can manage dynamic vocabulary and language model parts depending on the dialog context. The suggested interface helps the developer to build up and maintain the speech understanding part of a spoken dialog system in a consistent and flexible way. In addition, the reported experimental results show that information extraction performance can be increased by the presented methods.


doi: 10.21437/Interspeech.2005-309

Cite as: Lieb, R., Thomae, M., Ruske, G., Bobbert, D., Althoff, F. (2005) A flexible and integrated interface between speech recognition, speech interpretation and dialog management. Proc. Interspeech 2005, 3445-3448, doi: 10.21437/Interspeech.2005-309

@inproceedings{lieb05_interspeech,
  author={Robert Lieb and Matthias Thomae and Günther Ruske and Daniel Bobbert and Frank Althoff},
  title={{A flexible and integrated interface between speech recognition, speech interpretation and dialog management}},
  year=2005,
  booktitle={Proc. Interspeech 2005},
  pages={3445--3448},
  doi={10.21437/Interspeech.2005-309}
}