Data processing: speech signal processing – linguistics – language – Speech signal processing – Recognition
Reexamination Certificate
2000-02-10
2002-07-09
Banks-Harold, Marsha D. (Department: 2654)
Data processing: speech signal processing, linguistics, language
Speech signal processing
Recognition
C704S255000
Reexamination Certificate
active
06418411
ABSTRACT:
FIELD OF INVENTION
This invention relates to speech recognition and more particularly to speech recognition in a noisy environment.
BACKGROUND OF THE INVENTION
The use of cellular phones in a automobile or truck has grown rapidly over the recent few years. It is highly desirable for both the user and others around him for the operation of the cellular phone to be hands free so the user/driver can keep his or her hands on the wheel of the car or truck. In order to do that, the microphone is often placed remotely from the user near the upper left corner of the windshield. As the microphone is placed away from being near the mouth of the speaker, there is an increase in the mismatch conditions. This environment causes acoustic variations in the speech signal which do not carry linguistic information. variations are also caused by different tranducers, channels, speakers and noise backgrounds.
Hands-free speech recognition in automobile is always performed under mismatch condition. There is also speaker variability. A different speaker may use the phone or the speaker's voice changes somewhat. The environment is changing constantly as a function of engine status, road quality, window position and speaking position.
SUMMARY OF THE INVENTION
In accordance with one embodiment of the present invention, a one-time acoustic adaptation of Hidden Markov Models (HMM) in a low noise environment is performed for speaker and acoustic environment and an on-line compensation of HMMs to adjust the models to the utterance specific noise is performed for each incoming utterance.
REFERENCES:
patent: 5150449 (1992-09-01), Yoshida et al.
patent: 5778342 (1998-07-01), Erell et al.
patent: 5819223 (1998-10-01), Takagi
patent: 5970452 (1999-10-01), Aktas et al.
patent: 6151573 (2000-11-01), Gong
patent: 6253181 (2001-06-01), Junqua
patent: 6266636 (2001-07-01), Kosaka et al.
Banks-Harold Marsha D.
McFadden Susan
Troike Robert L.
LandOfFree
Method and system for adaptive speech recognition in a noisy... 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 and system for adaptive speech recognition in a noisy..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and system for adaptive speech recognition in a noisy... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2873697