AN HMM ADAPTATION METHOD FOR NOISE AND DISTORTION BY MAXIMIZING LIKELIHOOD

This paper describes a new HMM synthesis method in which HMM adapts to additive noise and multiplicative distortion. The conventional HMM sy nthesis method can only be applied to additive noise. In the method de scribed here, the likelihood of the synthesized HMM to adapt to speech is maximized, so that multiplicative distortion is eliminated when th e method is applied. Within the framework of this method, adaptation t o variations in the SN ratio, considered a problem in conventional HMM synthesis, can be formulated as part of the adaptation to multiplicat ive distortion. As a result of evaluating speech recognition rates usi ng our method, we have confirmed that the method is effective for impr oving the recognition rate of speech that contains additive noise and multiplicative distortion. (C) 1998 Scripta Technica.