Isolated word recognition using weighted state probabilities (WSP), a new approach for recognition in noise

Author

Vaich, T. ; Cohen, A.

Author_Institution

Dept. of Electr. Eng. & Comput. Eng., Ben-Gurion Univ. of the Negev, Beer-Sheva, Israel

fYear

1996

Firstpage

98

Lastpage

101

Abstract

Recognition of speech in extreme noisy environments is a difficult task. A new approach is suggested to enhance the performance of recognition in very low SNRs. The weighted state probabilities (WSP) method considers the heuristic states pattern recognition based on the left to right HMM configuration and the standard probability of getting the given observation sequence. On a ten digits (Hebrew) recognition task, with SNR of 10 dB, the WSP has improved recognition results from 0% to 50%. It is suggested to apply the method, in conjunction with parallel model combination (PMC) enhancement algorithm, to very low SNR word spotting systems

Keywords

hidden Markov models; noise; pattern recognition; probability; speech enhancement; speech recognition; 10 dB; HMM configuration; Hebrew recognition task; SNR; digit recognition; heuristic states pattern recognition; isolated word recognition; noisy environments; observation sequence; parallel model combination enhancement algorithm; recognition results; speech recognition performance; standard probability; weighted state probabilities; word spotting systems; Acoustic noise; Additive noise; Background noise; Hidden Markov models; Mobile handsets; Signal to noise ratio; Speech enhancement; Speech recognition; Testing; Working environment noise;

fLanguage

English

Publisher

ieee

Conference_Titel

Electrical and Electronics Engineers in Israel, 1996., Nineteenth Convention of

Conference_Location

Jerusalem

Print_ISBN

0-7803-3330-6

Type

conf

DOI

10.1109/EEIS.1996.566902

Filename

566902