Title :
Fuzzy smoothing of HMM parameters in speech recognition
Author :
Koo, J.M. ; Un, C.K.
Author_Institution :
Commun. Res. Lab., Korea Adv. Inst. of Sci. & Technol., Seoul, South Korea
fDate :
5/24/1990 12:00:00 AM
Abstract :
A technique to smooth hidden Markov model (HMM) parameters, based on the fuzzy mapping concept, is presented. The proposed algorithm is shown to result in a higher recognition rate than the co-occurrence smoothing method and requires neither an interpolation procedure nor the addition of more training data.
Keywords :
Markov processes; fuzzy set theory; speech analysis and processing; speech recognition; fuzzy mapping concept; fuzzy smoothing; hidden Markov model; model parameters; signal processing; speech recognition;
Journal_Title :
Electronics Letters
DOI :
10.1049/el:19900485
