Sixth International Conference on Spoken Language Processing (ICSLP 2000)

Beijing, China
October 16-20, 2000

A Portable Development Tool for Spoken Dialogue Systems

Satoru Kogure, Seiichi Nakagawa

Dept. of Information and Computer Sciences, Toyohashi University of Technology, Aichi, Japan

Speech recognition and language processing technologies have recently been improved, and speech recognition systems and dialogue systems are now in practical use. Nevertheless, not only the fundamental techniques but also the techniques for improving portability and expansibility of the systems need to be developed further. Whereas we need many user utterances to construct N-gram based language models for speech recognition. It is difficult to construct the language model for a new task/domain because many utterances cannot be collected easily. So, we propose an efficient method that is able to coustruct a language model, preparing at least only several hundreds utterances, We designed a domain-independent platform for developing any spoken dialogue systems for retrieving information from a database, and used the platform to build a literature retrieval system. When a system is implemented, the various items in the dictionary for understanding can be semi-automatically created by preparing the databases, dialogue samples and so forth.

Full Paper

Bibliographic reference.  Kogure, Satoru / Nakagawa, Seiichi (2000): "A portable development tool for spoken dialogue systems", In ICSLP-2000, vol.1, 238-241.