Chris Schmandt, Barry Arons

Abstract

This paper describes several key components of a speech based dialog system. This system was developed to explore opportunities for speech recognition in an integrated office environment. In order to achieve adequate performance despite speech recognition errors, it was necessary to develop a robust parser and a dialog generator which uses speech synthesis to converse with the user. Although designed for a specific application and recognition hardware, the underlying approach of the parser and dialog generator may be generalized to facilitate human-computer voice interaction in other environments.