Multiple-stage adaptive filtering of audio signals

• US 9,685,171 B1
• Filed: 11/20/2012
• Issued: 06/20/2017
• Est. Priority Date: 11/20/2012
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Yang, Jun
• Primary Examiner(s): Godbold, Douglas
• Application Number:
  US13/682,362
• Time in Patent Office: 1,673 Days
• Field of Search: 704224-226
• CPC Class Codes: G10L 2021/02165 Two microphones, one receiving mainly the noise signal and the other one mainly the speech signal
• CPC Class Codes: G10L 21/0208 Noise filtering

Abstract

The systems, devices, and processes described herein may include a first microphone that detects a target voice of a user within an environment and a second microphone that detects other noise within the environment. A target voice estimate and/or a noise estimate may be generated based at least in part on one or more adaptive filters. Based at least in part on the voice estimate and/or the noise estimate, an enhanced target voice and an enhanced interference, respectively, may be determined. One or more words that correspond to the target voice may be determined based at least in part on the enhanced target voice and/or the enhanced interference. In some instances, the one or more words may be determined by suppressing or canceling the detected noise.

Claims

1. A system comprising:

memory;

one or more processors; and

one or more computer-executable instructions stored in the memory and executable by the one or more processors to;

cause a first microphone to detect a target voice associated with a user within an environment and to cause a second microphone to detect noise within the environment;

implement a delay with respect to a first audio signal that represents the noise and refrain from delaying a second audio signal that represents the target voice;

terminate the delay based at least in part on detecting the noise;

process, by a first adaptive filter, the target voice to generate a target voice estimate, the target voice estimate representing a first estimate of the target voice of the user;

process, by the first adaptive filter, the noise to generate a noise estimate, the noise estimate representing a second estimate of the noise within the environment; and

generate, by a second adaptive filter different from the first adaptive filter, an enhanced target voice based at least in part on the target voice estimate and the noise estimate, and based at least in part on a suppression of the noise.

2. The system as recited in claim 1, wherein the delay starts at a first time at which the first microphone detects the noise and ends at a second time at which the second microphone detects the noise, the delay being implemented with respect to a synchronization between the first microphone and the second microphone.

3. The system as recited in claim 1, wherein the one or more computer-executable instructions are further executable by the one or more processors to:

determine one or more words that correspond to the target voice based at least in part on the enhanced target voice and the suppression of the noise; and

cause an operation to be performed within the environment based at least in part on the one or more words.

4. The system as recited in claim 1, wherein the first adaptive filter implements the delay utilizing one or more algorithms.

5. A system comprising:

a first microphone to detect a first sound;

a second microphone to detect a second sound;

memory;

one or more processors; and

one or more computer-executable instructions stored in the memory and executable by the one or more processors to perform operations comprising;

determining that the first sound is representative of at least a portion of a target voice;

determining that the second sound is representative of at least a portion of noise;

implementing a delay with respect to a first audio signal that represents the noise and refraining from delaying a second audio signal that represents the target voice;

terminating the delay based at least in part on detecting the noise;

processing, by a first adaptive filter, the target voice to generate a target voice estimate, the target voice estimate representing a first estimate of the target voice of a user associated with the first sound;

processing, by the first adaptive filter, the noise to generate a noise estimate, the noise estimate representing a second estimate of the noise within an environment associated with the user; and

generating, by a second adaptive filter different from the first adaptive filter, an enhanced target voice based at least in part on the target voice estimate and the noise estimate.

6. The system as recited in claim 5, wherein the operations further comprise determining one or more words that correspond to the target voice based at least in part on the enhanced target voice.

7. The system as recited in claim 6, wherein the operations further comprise causing an operation to be performed within an environment based at least in part on the one or more words.

8. The system as recited in claim 5, wherein the operations further comprise:

determining that the target voice is associated with the user within the environment; and

determining that the noise is different from the target voice.

9. The system as recited in claim 5, wherein the delay is associated with a first time at which the first microphone detects the second sound and a second time at which the second microphone detects the second sound, and wherein the operations further comprise:

implementing the delay with respect to a synchronization between the first microphone and the second microphone.

10. The system as recited in claim 9, wherein an amount of the delay is based on a length of the first adaptive filter, and wherein the operations further comprise adjusting the amount of the delay based at least in part on at least one of the target voice estimate or the noise estimate.

11. The system as recited in claim 5, wherein the operations further comprise determining the enhanced target voice based at least in part on a suppression of the noise.

12. A method comprising:

determining that a first sound captured by a first microphone is representative of at least a portion of a target voice;

determining that a second sound captured by a second microphone is representative of at least a portion of noise;

implementing a delay with respect to a first audio signal that represents the noise and refraining from delaying a second audio signal that represents the target voice;

terminating the delay based at least in part on detecting the noise;

processing, by a first adaptive filter, the target voice to generate a target voice estimate, the target voice estimate representing a first estimate of the target voice of a user associated with the first sound;

processing, by the first adaptive filter, the noise to generate a noise estimate, the noise estimate representing a second estimate of the noise within an environment associated with the user; and

generating, by a second adaptive filter different from the first adaptive filter, an enhanced target voice based at least in part on at least one of the target voice estimate or the noise estimate.

13. The method as recited in claim 12, wherein the delay is associated with a first time at which the first microphone captured the second sound and a second time at which the second microphone captured the second sound, the delay corresponding to a synchronization between the first microphone and the second microphone, and further comprising:

determining an amount of the delay based at least partly on a length of the first adaptive filter.

14. The method as recited in claim 13, further comprising adjusting the amount of the delay based at least in part on at least one of the target voice estimate or the noise estimate.

15. The method as recited in claim 12, further comprising:

suppressing at least a portion of the noise; and

determining the enhanced target voice based at least in part on the suppressing of the at least the portion of the noise.

16. A method comprising:

detecting a first sound representative of a target voice and a second sound representative of noise, the first sound being captured by a first microphone and the second sound being captured by a second microphone;

implementing a delay with respect to a first audio signal that represents the noise and refraining from delaying a second audio signal that represents the target voice;

terminating the delay based at least in part on detecting the noise;

processing, by a first adaptive filter, the target voice to generate a target voice estimate, the target voice estimate representing a first estimate of the target voice of a user associated with the first sound;

processing, by the first adaptive filter, the noise to generate a noise estimate, the noise estimate representing a second estimate of the noise within an environment associated with the user; and

generating, by a second adaptive filter different from the first adaptive filter, an enhanced target voice based at least in part on at least one of the target voice estimate or the noise estimate.

17. The method as recited in claim 16, wherein the delay being is with a first time at which the first microphone detects the second sound and a second time at which the second microphone detects the second sound, and further comprising:

determining the delay based at least in part on a synchronization between the first microphone and the second microphone.

18. The method as recited in claim 17, further comprising adjusting the amount of the delay based at least in part on at least one of the target voice estimate or the noise estimate.

19. The method as recited in claim 16, further comprising determining the enhanced target voice based at least in part on a suppression of the noise.

20. The method as recited in claim 16, further comprising:

determining one or more words that correspond to the target voice based at least in part on the enhanced target voice; and

causing an operation to be performed within an environment based at least in part on the one or more words.