Recent developments in distributed system processing have opened the doors to the running of highly complex systems over a number of networked computers. This enables the complexity of a system to be hidden behind a small, lightweight user interface - for example a downloadable web page. The Maya system makes use of such interfaces to combine the functionality of speech recognition, synthesis, robust parsing, text generation and dialogue management into a highly flexible multimodal architecture, working in real time. This paper describes the development of the architecture and interfaces to each system component. The configuration of the system to particular tasks is discussed, making use of an email secretary task as an example. Once configured, the system is able to adopt all the functionality of a conventional email system and extend these capabilities by allowing complex queries to be made about mail messages.
Cite as: Downey, S., Breen, A.P., Fernández, M., Kaneen, E. (1998) Overview of the maya spoken language system. Proc. 5th International Conference on Spoken Language Processing (ICSLP 1998), paper 0391, doi: 10.21437/ICSLP.1998-262
@inproceedings{downey98_icslp,
author={Simon Downey and Andrew P. Breen and Maria Fernández and Edward Kaneen},
title={{Overview of the maya spoken language system}},
year=1998,
booktitle={Proc. 5th International Conference on Spoken Language Processing (ICSLP 1998)},
pages={paper 0391},
doi={10.21437/ICSLP.1998-262}
}