Noise cancellation for voice activation

• US 9,552,825 B2
• Filed: 04/17/2013
• Issued: 01/24/2017
• Est. Priority Date: 04/17/2013
• Status: Active Grant

• Current Assignee: Ademco Inc. (Resideo Technologies, Inc.)
• Original Assignee: Honeywell International Inc.
• Inventors: Venkatesha, Sharath, Pham, Hai D.
• Primary Examiner(s): YEN, ERIC L
• Application Number:
  US13/864,950
• Publication Number:
  US 20140316778A1
• Time in Patent Office: 1,378 Days
• US Class Current: 1/1
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 21/0208 Noise filtering

Abstract

Devices, methods, and systems for noise cancellation for voice activation are described herein. For example, one or more embodiments include determining, with a computing device, whether a background noise received with the computing device is recognized, creating a stored noise cancellation filter based on the background noise when the background noise is not recognized, receiving a signal that includes the background noise and a voice command at the computing device, filtering out the background noise from the signal with the stored noise cancellation filter, and processing the voice command.

Claims

1. A method for noise cancellation for voice activation, comprising:

determining, with a computing device, whether a background noise received with the computing device is recognized;

creating a new stored noise cancellation filter based on the background noise when the background noise is not recognized;

receiving a signal that includes the background noise and a voice command at the computing device;

filtering out the background noise from the signal with the new stored noise cancellation filter;

processing the filtered signal at the computing device with a voice recognition engine;

determining a confidence level associated with the filtered signal;

applying a second stored noise cancellation filter to the filtered signal to filter out the background noise when the filtered signal does not meet a threshold confidence level; and

applying additional stored noise cancellation filters to the filtered signal to filter out the background noise until at least one of a defined time limit is exceeded and the processed signal meets the threshold confidence level.

2. The method of claim 1, wherein the method includes:

entering a sleep mode with the computing device upon a period of inactivity; and

receiving the background noise when the computing device is in the sleep mode.

3. The method of claim 2, wherein creating the stored noise cancellation filter based on the background noise when the background noise is not recognized includes creating the stored noise cancellation filter while the computing device is in the sleep mode.

4. The method of claim 1, wherein the method includes receiving a pre-existing stored noise cancellation filter with the computing device.

5. The method of claim 4, wherein the method includes filtering out the background noise with the pre-existing stored noise cancellation filter.

6. The method of claim 1, wherein the method includes uploading the new stored noise cancellation filter to a server and distributing the new stored noise cancellation filter to a second computing device for filtering out the background noise associated with the new stored noise cancellation filter.

7. The method of claim 6, wherein the method includes grouping a plurality of noise cancellation filters stored on the server based on a type of environment in which they were created;

and

download the grouped plurality of noise cancellation filters with the second computing device upon an installation of the second computing device.

8. The method of claim 1, wherein determining whether the background noise is recognized includes determining whether a correlation exists between the background noise and at least one of a plurality of stored noise cancellation filters.

9. The method of claim 1, wherein the method includes:

sending the background noise to a second computing device;

determining, with the second computing device, whether a correlation exists between the background noise and one of a plurality of stored noise cancellation filters.

10. The method of claim 1, wherein the method includes:

applying a plurality of stored noise cancellation filters to the background noise and the voice command to filter out the background noise based on a correlation between the background noise and the plurality of stored noise cancellation filters, wherein the plurality of stored noise cancellation filters have an overall first weight;

modifying the overall first weight associated with the plurality of stored noise cancellation filters if the determined confidence level does not meet the threshold confidence level; and

applying a plurality of stored noise cancellation filters with an overall second weight to the background noise and the voice command to filter out the background noise in response to the confidence level not meeting the threshold confidence level.

11. A machine-readable non-transitory medium storing instructions for noise cancellation for voice activation, executable by a machine to cause the machine to:

determine, with a computing device, whether a first background noise received by the computing device from an environment proximate to the computing device is recognized;

create a first stored noise cancellation filter based on the first background noise when the first background noise is not recognized;

upload the first stored noise cancellation filter to a second computing device for filtering out background noise associated with the first stored noise cancellation filter;

receive a second background noise and a voice command at the second computing device;

process the second background noise and the voice command at the second computing device with a voice recognition engine;

determine a confidence level associated with a processed signal that includes the second background noise and the voice command;

filter out the second background noise with the first stored noise cancellation filter at the second computing device when the processed signal does not meet the confidence level; and

process the filtered signal with the voice recognition engine;

determine a second confidence level upon processing the filtered signal with the voice recognition engine;

determine whether the processed filtered signal meets the confidence level; and

reprocess the filtered signal with a different noise cancellation filter until the processed filtered signal meets the confidence level.

12. The medium of claim 11, wherein the second computing device receives the second background noise from a second environment, wherein the second environment is proximate to the second computing device.

13. The medium of claim 11, further comprising instructions executable to upload the first stored noise cancellation filter to a server, wherein the server distributes the first stored noise cancellation filter to the second computing device.

14. The medium of claim 13, further comprising instructions executable to distribute a group of stored noise cancellation filters to the second computing device, wherein the group of stored noise cancellation filters is formed based on a type of environment that the noise cancellation filters were created.

15. The medium of claim 11, further comprising instructions executable to:

combine a plurality of stored noise cancellation filters; and

filter out the second background noise with the combined plurality of stored noise cancellation filters.

16. A system for noise cancellation for voice activation, the system comprising:

a processor;

a memory having instructions stored thereon which when executed by the processor cause the processor to;

determine, with a first computing device, whether a first background noise received by the first computing device is recognized;

create a first stored noise cancellation filter based on the first background noise when the first background noise is not recognized;

receive a second background noise and a voice command at the first computing device;

process the second background noise and the voice command at the first computing device with a voice recognition engine;

determine a confidence level associated with a processed signal that includes the second background noise and the voice command;

apply the first stored noise cancellation filter to the signal to filter out the second background noise when the processed signal does not meet the confidence level;

process the filtered signal with the voice recognition engine;

determine a second confidence level upon processing the filtered signal with the voice recognition engine;

determine whether the processed filtered signal meets the confidence level; and

reprocess the filtered signal with a different noise cancellation filter until the processed filtered signal meets the confidence level.

17. The system of claim 16, wherein a second stored noise cancellation filter is generated at a second computing device and downloaded by the first computing device from the second computing device in response to a selection received by the first computing device.

18. The system of claim 16, wherein a second stored noise cancellation filter is downloaded by the first computing device from a server in communication with the first computing device and a second computing device where the second stored noise cancellation filter is generated.