Efficient HMM evaluation for recognition of nonverbal speech

Author

Deller, J.R. ; Snider, R.K.

Author_Institution

Dept. of Electr. Eng., Michigan State Univ., East Lansing, MI, USA

fYear

1989

Firstpage

657

Abstract

A procedure is described for evaluating the likelihood of a hidden Markov model (HMM) using only O(N) flops per observation, where N is the number of states. This improved computational complexity is necessary for near-real-time recognition algorithms for nonverbal speech to be run on ordinary personal computers. In preliminary experiments these measures have significantly reduced the computational load. Further significant reductions are anticipated with larger databases

Keywords

Markov processes; computerised pattern recognition; medical computing; physiological models; speech recognition; computational complexity; computational load; databases; efficient hidden Markov model evaluation; flops; near-real-time recognition algorithms; nonverbal speech recognition; ordinary personal computers; Computational complexity; Computer architecture; Digital signal processing; Equations; Hidden Markov models; Laboratories; Microcomputers; Signal processing algorithms; Speech analysis; Speech recognition;

fLanguage

English

Publisher

ieee

Conference_Titel

Engineering in Medicine and Biology Society, 1989. Images of the Twenty-First Century., Proceedings of the Annual International Conference of the IEEE Engineering in

Conference_Location

Seattle, WA

Type

conf

DOI

10.1109/IEMBS.1989.95723

Filename

95723