Word spotting in a variable noise level environment

Details Word spotting in a variable noise level environment Word spotting in a variable noise level environment

1997-02-03

1998-08-11

Knepper, David D.

Data processing: speech signal processing, linguistics, language

Speech signal processing

Recognition

704253, G10L 506, G10L 708

Patent

active

057941947

ABSTRACT:
Low- and high-dimensional feature parameters are obtained by analyzing a speech pattern to be recognized. A word spotting section extracts a plurality of approximate word feature vector candidates representing word features from the low-dimensional feature parameter without fixing word boundaries. Start and end points are detected for each of the word feature vector candidates. Detail word feature vector candidates are extracted from the high-dimensional feature parameter in accordance with the detected start and end points. A recognition dictionary stores reference patterns with which the detail word feature vector candidates are matched. A pattern matching section calculates similarity values between each of the detail word feature vector candidates and the reference patterns stored in the recognition dictionary. A recognition result selects reference patterns stored in the recognition dictionary when its similarity value is greater than a prescribed threshold value.

REFERENCES:
patent: 4625287 (1986-11-01), Matsuura et al.
patent: 4720802 (1988-01-01), Damoulakis et al.
patent: 4783802 (1988-11-01), Takebayashi et al.
patent: 4856067 (1989-08-01), Yamada et al.
patent: 4882755 (1989-11-01), Yamada et al.
patent: 4937871 (1990-06-01), Hattori
patent: 4985924 (1991-01-01), Matsuura et al.
patent: 5023911 (1991-06-01), Gerson
Keyword-Spotting in Noisy Continuous Speech Using Word Pattern Vector Subabstrction and Noise Immunity Learning, Yoichi Takebayashi, et al. ICASSP, 1992, pp. 85-88.
A Consideration of Learning in Speech Recognition From the Viewpoint of A1 Class-Description Learning, pp. 705-713, Yoichi Takebayashi Proceedings of the 21st Hawaii International Conference on System Sciences, Jan. 5-8, 1988.
A Speaker Independent Recognition Algorithm for Connected Word Using Word Boundary Hypothesizer, Teruhiko Ukita, et al. ICASSP Proceedings 1986, Apr. 7-11, 1986, pp. 21.4.1-21.4.4.
Speaker-Adaptive Connected Syllable Recognition Based on the Multiple Similarity Method, Hiroyuki Tsuboi, et al. ICASSP 1986 Proceedings, Apr. 7-11, 1986, pp. 49.8.1-49.8.4.
Telephone Speech Recognition Using a Hybrid Method, Yoichi Takebayashi, et al. IEEE 7th International Conference on Pattern Recognition Proceedings Jul. 30-Aug. 2, 1984, pp. 1232-1235.

Affiliated with

Also associated with

Rating

Word spotting in a variable noise level environment does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Word spotting in a variable noise level environment, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Word spotting in a variable noise level environment will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-403573