A MRF-Based Parallel Processing for Speech Recognition Using Linear Predictive HMM

Hideki NODA

IEICE TRANSACTIONS on Information and Systems   Vol.E77-D    No.10    pp.1142-1147
Publication Date: 1994/10/25
Online ISSN: 
Print ISSN: 0916-8532
Type of Manuscript: PAPER
Category: Speech Processing
parallel processing,  speech recognition,  MRF model,  HMM,  ICM algorithm,  

Full Text: PDF>>
Buy this Article

Parallel processing in speech recognition is described, which is carried out at each frame on time axis. We have already proposed a parallel processing algorithm for HMM (Hidden Markov Model)-based speech recognition using Markov Random Fields (MRF). The parallel processing is realized by modeling the hidden state sequence by an MRF and using the Iterated Conditional Modes (ICM) algorithm to estimate the optimal state sequence given an observation sequence and model parameters. However this parallel processing with the ICM algorithm is applicable only to the standard HMM but not to the improved HMM like the linear predictive HMM which takes into account the correlations between nearby observation vectors. In this paper we propose a parallel processing algorithm applicable to the correlation-considered HMM, where a new deterministic relaxation algorithm called the Generalized ICM (GICM) algorithm is used instead of the ICM algorithm for estimation of the optimal state sequence. Speaker independent isolated word recognition experiments show the effectiveness of the proposed parallel processing using the GICM algorithm.