A natural language generation module for spoken dialogue systems has been developed that performs three steps: generating multiple versions of an utterance, choosing the best version by a set of criteria, and annotating the text using structural information accumulated during the generation process. The requirements of spoken output is catered for by several design decisions.
Cite as: Portele, T. (2000) Natural language generation for spoken dialogue. Proc. 6th International Conference on Spoken Language Processing (ICSLP 2000), vol. 3, 310-313
@inproceedings{portele00_icslp,
author={Thomas Portele},
title={{Natural language generation for spoken dialogue}},
year=2000,
booktitle={Proc. 6th International Conference on Spoken Language Processing (ICSLP 2000)},
pages={vol. 3, 310-313}
}