Title :
Improved acoustic modeling for speech recognition using 2D Markov random fields
Author_Institution :
ATR Interpreting Telecommun. Res. Labs., Kyoto, Japan
Abstract :
This paper argues that many HMM model inaccuracies are a direct consequence of the fact that the HMM is a one dimensional stochastic model applied to a two dimensional process. Thus we argue that a 2D stochastic process, known as a Markov random field (MRF) should perform better. We describe a training method for MRFs and analyze its convergence behavior
Keywords :
acoustic signal processing; convergence of numerical methods; hidden Markov models; random processes; speech processing; speech recognition; stochastic processes; 2D Markov random fields; 2D stochastic process; HMM; acoustic modeling; convergence; one dimensional stochastic model; pattern discrimination; speech recognition; training method; two dimensional process; Convergence; Electronic mail; Filters; Frequency; Frequency domain analysis; Hidden Markov models; Linear predictive coding; Markov random fields; Random variables; Read only memory; Speech recognition; Stochastic processes; Tiles; Tires;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1995. ICASSP-95., 1995 International Conference on
Conference_Location :
Detroit, MI
Print_ISBN :
0-7803-2431-5
DOI :
10.1109/ICASSP.1995.479648
