THE POTENTIAL ROLE OF SPEECH PRODUCTION MODELS IN AUTOMATIC SPEECH RECOGNITION

This paper investigates the issues that are associated with applying s peech production models to automatic speech recognition (ASR). Here th e applicability of articulatory representations to ASR is considered i ndependently of the role of articulatory representations in speech per ception. While the question of whether it is necessary or even possibl e for human listeners to recover the state of the articulators during the process of perceiving speech is an important one, it is not consid ered here. Hence, the authors refrain from posing completely new parad igms for ASR which more closely parallel the relationship between spee ch production and human speech understanding. Instead, work aimed at i ntegrating speech production models into existing ASR formalisms is de scribed. (C) 1996 Acoustical Society of America.