SYSTEMS AND METHODS FOR INTELLIGENT CONTROL OF MICROPHONES FOR SPEECH RECOGNITION APPLICATIONS

• US 20080167868A1
• Filed: 01/04/2007
• Published: 07/10/2008
• Est. Priority Date: 01/04/2007
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Viswanathan, Mahesh, Kanevsky, Dimitri
• Granted Patent: US 8,140,325 B2
• Time in Patent Office: Days
• US Class Current: 704/233
• CPC Class Codes: G10L 15/20 Speech recognition techniques specially adapted for robustness in adverse environments, e.g. in noise, of stress induced speech G10L21/02 takes precedence
• CPC Class Codes: G10L 15/24 Speech recognition using non-acoustical features
• CPC Class Codes: G10L 15/28 Constructional details of speech recognition systems

Abstract

Systems and methods for intelligent control of microphones in speech processing applications, which allows the capturing, recording and preprocessing of speech data in the captured audio in a way that optimizes speech decoding accuracy.

Claims

1. An audio processing system, comprising:

a microphone control system to (i) control capturing of audio input to a microphone system and to (ii) selectively output speech segments within captured audio to a speech decoding system;

a speech activity detection system to detect timing of speech activity events, wherein the speech activity events are indicators of speech input to the microphone system, and wherein the timing of speech activity events are used to determine the speech segments within captured audio; and

an acoustic quality control system to detect conditions regarding capturing of audio input to tire microphone system, which can result in captured audio with acoustic characteristics that adversely affect a decoding accuracy of the captured audio.

2. The system of claim 1, wherein the acoustic quality control system, provides feedback to a user regarding a detected condition having an adverse affect on an acoustic quality of audio input.

3. The system of claim 1, wherein the acoustic quality control system automatically adjusts one or more system parameters to compensate for degraded acoustic characteristics of the captured audio.

4. The system of claim 1, wherein the acoustic quality control system detects a change in an environmental static noise level and automatically adjusts a microphone volume sensitivity parameter.

5. The system of claim 1, wherein the acoustic quality control system detects a distance between a microphone and a speaker'"'"'s mouth, and automatically adjusts a microphone volume parameter and microphone volume sensitivity parameter based on the detected distance.

6. The system, of claim 1, wherein the acoustic quality control system detects a condition in which a speaker uses a microphone that is different from a microphone used by the speaker for training the speech decoding system, and automatically modulates captured audio input to substantially match an audio signal of the training microphone.

7. The system of claim 1, wherein the speech activity events include microphone switch button actions.

8. The system of claim 1, wherein the speech activity events include speech intent events that indicate a user'"'"'s intention to speak into the microphone, which are derived from biometric feature data of the user.

9. The system, of claim 1, wherein the speech activity events include speech intent events that indicate a user'"'"'s intention to speak into the microphone, which are derived from data acquired from tracking behavioral characteristics of the user.

18. The system of claim 1, wherein the one or more detected speech activity events include detection of user intention to speak into the microphone and detection of speech in captured audio.

10. An audio processing system, comprising

a microphone system, comprising a microphone for capturing audio input to the microphone system;

a microphone manager to (i) control capturing of audio input to the microphone system, (ii) control recording of captured audio, (iii) detect speech activity events, (iv) determine speech segments of captured audio comprising speech data based on timing of one or more detected speech activity events, (v) selectively output speech segments within captured, audio to a speech decoding system; and

to (vi) detect an environmental condition regarding capturing of audio, which can result in captured audio with acoustic characteristics that adversely affect a decoding accuracy of the captured audio;

a sensor system comprising one or more sensor devices to generate sensor data which is processed by the microphone manager to detect speech activity events and environmental conditions.

11. The audio system of claim 10, wherein the microphone manager automatically adjusts one or more microphone parameters and speech decoding parameters to compensate the degraded acoustic characteristics of the captured audio.

12. The system of claim 11, wherein the microphone manager maintains and periodically updates a sample of an environmental static noise level and a stored voice sample of a speaker, and automatically adjusts microphone volume sensitivity for the speaker based on a current sample of the environmental static noise level and the voice sample of the speaker.

13. The system of claim 12, wherein the sensor system, comprises a sensor device which generates sensor data which is used to determine a distance between a microphone and a speaker'"'"'s mouth, wherein the microphone manager automatically adjusts microphone volume level based on a current distance between the microphone and the speaker'"'"'s month, and automatically adjusts microphone volume sensitivity by transforming a volume parameter of the stored voice sample of a speaker based on the current distance, and transforming the speaker'"'"'s speech input to a microphone to adapt the microphone volume sensitivity based on the stored voice sample of the speaker and a scored sample of the current environmental static noise level.

14. The system of claim 10, wherein the microphone manager detects an environmental condition where a speaker uses a microphone in the microphone system which is different from a microphone used by the speaker for training the speech decoding system.

15. The system of claim 14, wherein the microphone manager selectively activates one of a plurality different types of microphones of the microphone system, which is the same or similar to the training microphone.

16. The system of claim 14, wherein the microphone manager modulates an audio input to the microphone to substantially catch an audio signal of the training microphone.

17. The system of claim 14, wherein the one or more detected speech activity events include detection of microphone switch button actions and detection of speech in captured audio.