Data processing: speech signal processing – linguistics – language – Speech signal processing – Recognition
Reexamination Certificate
1998-08-28
2001-01-23
Dorvil, Richemond (Department: 2741)
Data processing: speech signal processing, linguistics, language
Speech signal processing
Recognition
C704S238000, C704S250000
Reexamination Certificate
active
06178401
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Technical Field
The present invention relates to speech recognition systems and, more particularly, to a method for reducing search complexity in a speech recognition system.
2. Discussion of Related Prior Art
A speech recognition system consists of the following three major components: an acoustic fast match; an acoustic detailed match; and a language model. The fast match is employed to construct a list of candidate words given the acoustic evidence and the language model is employed to similarly construct a list of candidate words given the word context. The list of candidate words are then ranked so that only the top ranking candidate words are processed by the more detailed, computationally expensive, and time consuming detailed match.
Since the fast match examines all the words in the vocabulary in order to construct the list of word candidates, the fast match represents one of the most computationally expensive components of the speech recognition system. Thus, it would be desirable and highly advantageous to provide a method for reducing the processing time of the fast match. Further, due to the more detailed evaluation performed by the detailed match, it would also be desirable and highly advantageous to provide a method for shortening the list of word candidates processed by the detailed match.
SUMMARY OF THE INVENTION
In one aspect of the invention, a method for reducing search complexity in a speech recognition system having a fast match, a detailed match, and a language model, comprises the steps of: optionally employing the fast match to generate candidate words and acoustic scores corresponding to the candidate words, based on at least one predetermined variable; employing the language model to generate language model scores; combining the acoustic scores with the language model scores and ranking the combined scores to determine top ranking candidate words to be later processed by the detailed match, when the fast match is employed; and employing the detailed match to generate detailed match scores for the top ranking candidate words.
These and other objects, features and advantages of the present invention will become apparent from the following detailed description of illustrative embodiments thereof, which is to be read in connection with the accompanying drawings.
REFERENCES:
patent: 5444617 (1995-08-01), Merialdo
patent: 5884259 (1999-03-01), Bahl et al.
patent: 5897616 (1999-04-01), Kanevsky et al.
patent: 5963905 (1999-10-01), Novak et al.
patent: 6067514 (2000-05-01), Chen
International Conference on Acoustics, Speech and Signal Processing. Dharanmipragada et al., “A fast vocabulary independent algorithm for spotting words in speech”. pp. 233-236 vol. 1, May 1998.
IBM Research Division. Bahl et al., “Matrix Fast Match: A fast method for identifying a short list of candidate words for decoding”, pp. 345-348, May 1998.
A Fast Approximate Acoustic Match for Large Vocabulary Speech Recognition, Bahl et al., IEEE Transactions on Speech and Audio Processing, vol. 1, No. 1, Jan. 1993, pp. 59-67.
A Tree Search Strategy for Large-Vocabulary Continuous Speech Recognition, Gopalakrishnan et al., IEEE, pp. 572-575.
Franz Martin
Novak Miroslav
Dorvil Richemond
F. Chau & Associates LLP
International Business Machines - Corporation
LandOfFree
Method for reducing search complexity in a speech... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for reducing search complexity in a speech..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for reducing search complexity in a speech... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2533079