Generalised fuzzy hidden markov models for speech recognition

Abstract

A generalised fuzzy approach to statistical modelling techniques for speech recognition is proposed in this paper. Fuzzy C-means (FCM) and fuzzy entropy (FE) techniques are combined into a generalised fuzzy technique and applied to hidden Markov models (HMMs). A more robust version of the above fuzzy technique based on the noise clustering (NC) method is also proposed. Experimental results were performed on the TI46 speech data corpus. A significant result for isolatedword recognition performed on a highly confusable vocabulary consisting of the nine English E-set words is that, a 33.8% recognition error rate for the HMM-based system was reduced to 30.5%, 29.9%, 29.8% and 27.8%, respectively, by using the FCM-HMM, the FE-HMM, the NC-FE-HMM, and the NC-FCM-HMM-based systems.