Acoustic echo cancellation for microphone array with dynamically changing beam forming

Patent metadata

Field	Value
Publication number	US-9549079-B2
Application number	US-201314018635-A
Country	US
Kind code	B2
Filing date	Sep 5, 2013
Priority date	Sep 5, 2013
Publication date	Jan 17, 2017
Grant date	Jan 17, 2017

Abstract

Acoustic echo cancellation is improved by receiving a speaker signal that is used to produce audio in a room, and receiving audio signals that capture audio from an array of microphones in the room, including an acoustic echo from the speakers. To cancel the acoustic echo, one adaptive filter is associated with a corresponding subspace in the room. Each of the audio signals is assigned to at least one of the adaptive filters, and a set of coefficients is iteratively determined for each of the adaptive filters. The coefficients for an adaptive filter are determined by selecting each of the audio signals assigned to that adaptive filter and adapting the filter to remove an acoustic echo from each of the selected audio signals. At each iteration, a different audio signal is selected from the audio signals assigned to the adaptive filter in order to determine the set of coefficients.

First claim

What is claimed is:

1. A method comprising:
receiving a speaker signal that is used to produce audio in a room;
obtaining a plurality of audio signals derived from an array of microphones in the room;
associating a plurality of adaptive filters to a corresponding plurality of subspaces in the room, wherein each of the adaptive filters is associated with one subspace;
assigning each of the plurality of audio signals to at least one of the associated adaptive filters;
iteratively determining a set of coefficients for a first adaptive filter associated with a first subspace by adapting the first adaptive filter to remove an acoustic echo from each of the audio signals assigned to the first adaptive filter;
selecting a first audio signal from the plurality of audio signals assigned to the first adaptive filter, and a first coefficient from the set of coefficients for the first adaptive filter;
producing an echo cancelled microphone signal using the first audio signal and the first coefficient of the first adaptive filter;
monitoring the performance of at least one other audio signal with the first coefficient, wherein the at least one other audio signal is assigned to a second adaptive filter associated with a second subspace; and
based on the monitored performance, re-assigning the at least one other audio signal to the first adaptive filter.

2. The method of claim 1 , wherein the second subspace is physically adjacent to the first subspace.

3. The method of claim 1 , wherein each coefficient in the set of coefficients comprises a plurality of weighted coefficients.

4. The method of claim 1 , wherein the plurality of audio signals comprises a plurality of audio beams and a plurality of individual microphone signals.

5. The method of claim 4 , wherein the plurality of audio beams are derived by summing signals from at least two microphones from the array of microphones with at least one relative delay between the at least two microphones from the array of microphones.

6. The method of claim 1 , further comprising selecting, in at least one further iteration, a second audio signal, wherein the second audio signal was not used to determine the first coefficient; and
producing the echo cancelled microphone signal using the second audio signal and the first coefficient of the first adaptive filter.

7. An apparatus comprising:
at least one speaker configured to produce audio in a room from a speaker signal;
a microphone array comprising a plurality of microphones configured to capture audio, the captured audio comprising an echo of the audio produced by the at least one speaker; and
a processor coupled to the at least one speaker and microphone array, and configured to:
receive the speaker signal;
obtain a plurality of audio signals derived from the microphone array;
associate a plurality of adaptive filters to a corresponding plurality of subspaces in the room, wherein each of the adaptive filters is associated with one subspace;
assign each of the plurality of audio signals to at least one of the associated adaptive filters;
iteratively determine a set of coefficients for a first adaptive filter associated with a first subspace by adapting the first adaptive filter to remove an acoustic echo from each of the audio signals assigned to the first adaptive filter;
select a first audio signal from the plurality of audio signals assigned to the first adaptive filter, and a first coefficient from the set of coefficients for the first adaptive filter;
produce an echo cancelled microphone signal using the first audio signal and the first coefficient of the first adaptive filter;

monitor the performance of at least one other audio signal with the first coefficient, wherein the at least one other audio signal is assigned to a second adaptive filter associated with a second subspace; and based on the monitored performance, re-assign the at least one other audio signal to the first adaptive filter.

8. The apparatus of claim 7 , wherein the second subspace is physically adjacent to the first subspace.

9. The apparatus of claim 7 , wherein each coefficient in the set of coefficients comprises a plurality of weighted coefficients.

10. The apparatus of claim 7 , wherein the plurality of audio signals comprises a plurality of audio beams and a plurality of individual microphone signals.

11. The apparatus of claim 10 , wherein the processor is configured to derive the plurality of audio beams by summing signals from at least two microphones from the plurality of microphones with at least one relative delay between the at least two microphones from the plurality of microphones.

12. The apparatus of claim 7 , wherein the processor is further configured to:
select, in at least one further iteration, a second audio signal, wherein the second audio signal was not used to determine the first coefficient; and
produce the echo cancelled microphone signal using the second audio signal and the first coefficient of the first adaptive filter.

13. One or more non-transitory computer readable storage media encoded with software comprising computer executable instructions and when the software is executed operable to cause a processor to:
receive a speaker signal that is used to produce audio in a room;
obtain a plurality of audio signals derived from a plurality of microphones in the room;
associate a plurality of adaptive filters to a corresponding plurality of subspaces in the room, wherein each of the adaptive filters is associated with one subspace;
assign each of the plurality of audio signals to at least one of the associated adaptive filters;
iteratively determine a set of coefficients for a first adaptive filter associated with a first subspace by adapting the first adaptive filter to remove an acoustic echo from each of the audio signals assigned to the first adaptive filter;
select a first audio signal from the plurality of audio signals assigned to the first adaptive filter, and a first coefficient from the set of coefficients for the first adaptive filter;
produce an echo cancelled microphone signal using the first audio signal and the first coefficient of the first adaptive filter;
monitor the performance of at least one other audio signal with the first coefficient, wherein the at least one other audio signal is assigned to a second adaptive filter associated with a second subspace; and based on the monitored performance, re-assign the at least one other audio signal to the first adaptive filter.

14. The computer readable storage media of claim 13 , wherein the second subspace is physically adjacent to the first subspace.

15. The computer readable storage media of claim 13 , wherein each coefficient in the set of coefficients comprises a plurality of weighted coefficients.

16. The computer readable storage media of claim 13 , wherein the plurality of audio signals comprises a plurality of audio beams and a plurality of individual microphone signals.

17. The computer readable storage media of claim 16 , wherein the computer executable instructions cause the processor to derive the plurality of audio beams by summing signals from at least two microphones from the plurality of microphones with at least one relative delay between the at least two microphones from the plurality of microphones.

18. The computer readable storage media of claim 13 , further comprising instructions that cause the processor to:
select in at least on further iteration a second audio signal, wherein the second audio signal was not used to determine the first coefficient; a

Assignees

• Cisco Tech Inc

Inventors

• Bao Feng
• Kunapuli Subrahmanyam Venkata
• Yang Fei
• Bao Xiangyu
• Sundsbarm Tor A

Classifications

• H04M9/082 Primary

  using echo cancellers (echo cancellers per se H04B3/23) · CPC title

• G10K11/346

  using phase variation · CPC title