Data processing: structural design – modeling – simulation – and em – Simulating nonelectrical device or system – Mechanical
Reexamination Certificate
1999-03-08
2001-11-27
Dorvil, Richemond (Department: 2641)
Data processing: structural design, modeling, simulation, and em
Simulating nonelectrical device or system
Mechanical
C704S235000
Reexamination Certificate
active
06324499
ABSTRACT:
CROSS REFERENCE TO RELATED APPLICATIONS
(Not Applicable)
STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT
(Not Applicable)
BACKGROUND OF THE INVENTION
1. Technical Field
This invention relates to the field of speech recognition software and more particularly to a system for eliminating mis-recognitions of spoken words resulting from randomly occurring background noises in an acoustic environment.
2. Description of the Related Art
In recent years software developers have devised various application programs to enable computers to recognize spoken words in various languages. A common problem when using such voice recognition software in a home or office setting is mis-recognitions of spoken words when randomly occurring background noises occur. As used herein, the phrase “randomly occurring background noise” refers to those noises which are random with respect to their occurency in an acoustic environment such as an office. Examples of randomly occurring background noises include telephone ringers, file drawers opening and closing, sneezing, coughing, breathing noises and the like. These randomly occurring noises are to be distinguished from those noises which are merely random in their noise characteristics, such as white noise. In any case, the problem of randomly occurring noises is particularly acute when the user has a far field microphone.
Conventional speech recognition software applications do not distinguish or respond appropriately to randomly occurring noises. Thus, speech recognition systems using far-field desktop or monitor microphones that do not have noise-canceling capability tend to have low recognition accuracy because they cannot distinguish between randomly occurring environmental noise and speech. Conversely, near field microphones, such as those used in headsets, are less prone to problems arising from randomly occurring environmental noise. However, such microphones are more prone to reception of randomly occurring noise in the form of inadvertent personal noises, such as breathing, coughing or sneezing. Accordingly, there exists a need for voice recognition systems to have a more sophisticated ability to intelligently handle randomly occurring noises including environmental noises and personal noises.
SUMMARY OF THE INVENTION
Certain events can be anticipated and associated with randomly occurring noises. For example, when a user is making use of a computer voice recognition application and a nearby telephone rings, the user may wish to answer that telephone. Under these circumstances, it will normally be desirable for the voice recognition process to be suspended. Alternatively, for certain other types of randomly occurring noises, e.g. a telephone ringing in an adjacent cubicle, it may be more appropriate for the voice recognition software to ignore the noise. In either case, the randomly occurring noise should not be interpreted as a spoken word.
With the present system, frequent randomly occurring noises such as background office noise or personal noises are mapped to perform one of four actions. In particular, such noises can be mapped to various computer voice recognition functions so that they are either (1) ignored by the voice recognition application, (2) cause operation of the speech recognition system to be suspended, (3) cause a microphone input to be turned off, or (4) execute a user defined macro. Different randomly occurring noises may require a different response. For example, when the operation of the speech recognition system is suspended, it is possible to resume use of the speech system only with a special voice command such as “wake up”. This is possible since the microphone remains on even though the speech recognition function is suspended. Conversely, if the microphone is simply turned off, the speech system cannot be caused to resume operation by means of a speech command. One of the foregoing alternatives may be more appropriate than the other in certain circumstances and it is therefore important to allow the user flexibility to control the specific response selected for a particular randomly occurring noise.
A dialog allows the user to specify the name of a noise and the action to be performed when the noise occurs. Additionally, each noise can be recorded as it sounds through the microphone. If the system cannot recognize the particular noise sufficiently to be able to distinguish it again, the system notifies the user.
Thus, the invention concerns a method and system for responding to randomly occurring background noise. The system receives an audio signal representative of sound in an audio environment and processes the audio signal to identify certain non-speech sounds. A pre-defined action is performed in response to the non-speech sound which has been identified. The pre-defined action is selected from one of the group consisting of disabling a microphone source of the audio signal, suspending further processing of the audio signal by the speech recognition system, and executing a user-defined macro.
The system may perform additional steps including recording a sound which is to be identified as a non-speech sound and assigning one of the pre-defined actions to be performed in response when the non-speech sound has been identified. In one instance, the non-speech sound may be a telephone ringer. In another instance, the non-speech sound can be a personal sound.
The system permits speech recognition to occur more smoothly in an open-plan office layout and in other noisy environments. For noise canceling microphones, noise recognition will eliminate some mis-recognitions caused by nearby noises, such as telephone and keyboard noises. For microphones without the noise canceling feature, noise recognition will significantly enhance accuracy by preventing randomly occurring environmental noises from being recognized as speech. The invention also allows users who have exceptional difficulties with breath noises to avoid interpretation of these randomly occurring personal noises as speech and thereby enjoy the benefit of speech recognition software.
REFERENCES:
patent: 5764852 (1998-06-01), Williams
patent: 5905971 (1999-05-01), Hovel
patent: 5970446 (1999-10-01), Goldberg et al.
patent: 6067514 (2000-05-01), Chen
patent: 6076059 (2000-06-01), Glickman et al.
ICASSP 91. 1991 International Conference on Acoustics, Speech and Signal Processing. Rose et al., “Robust speaker identification in noisy environments using noise adaptive speaker models” pp. 401-404. May 1991.
Ballard Barbara
Lewis James R.
Akerman & Senterfitt
Dorvil Richemond
International Business Machines Corp.
LandOfFree
Noise recognizer for speech recognition systems does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Noise recognizer for speech recognition systems, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Noise recognizer for speech recognition systems will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2583024