Robust numeric recognition in spoken language dialogue

This paper addresses the problem of automatic numeric recognition and understanding in spoken language dialogue. We show that accurate numeric understanding in fluent unconstrained speech demands maintaining robustness at several different levels of system design, including acoustic, language, understanding and dialogue. We describe a robust system for numeric recognition and present algorithms for feature extraction, acoustic and language modeling, discriminative training, utterance verification and numeric understanding and validation. Experimental results from a field-trial of a spoken dialogue system are presented that include customers' responses to credit card and telephone number requests. (C) 2001 Elsevier Science B.V. All rights reserved.