Integrated echo cancellation and noise suppression

• US 8,811,601 B2
• Filed: 04/04/2011
• Issued: 08/19/2014
• Est. Priority Date: 04/04/2011
• Status: Active Grant

• Current Assignee: Qualcomm, Inc.
• Original Assignee: Qualcomm, Inc.
• Inventors: Mohammad, Asif I., Chan, Kwokleung, Liu, Ian Ernan, Visser, Erik, Ramakrishnan, Dinesh
• Primary Examiner(s): Jamal, Alexander
• Application Number:
  US13/079,548
• Publication Number:
  US 20120250882A1
• Time in Patent Office: 1,233 Days
• Field of Search: 379/406.01
• US Class Current: 379/406.01
• CPC Class Codes: G10L 2021/02082 the noise being echo, reverberation of the speech
• CPC Class Codes: G10L 2021/02166 Microphone arrays; Beamforming
• CPC Class Codes: G10L 21/0208 Noise filtering
• CPC Class Codes: H04B 3/234 using double talk detection
• CPC Class Codes: H04M 9/082 using echo cancellers echo cancellers per se H04B3/23

Abstract

A method for echo cancellation and noise suppression is disclosed. Linear echo cancellation (LEC) is performed for a primary microphone channel on an entire frequency band or in a range of frequencies where echo is audible. LEC is performed on one or more secondary microphone channels only on a lower frequency range over which spatial processing is effective. The microphone channels are spatially processed over the lower frequency range after LEC. Non-linear noise suppression post-processing is performed on the entire frequency band. Echo post-processing is performed on the entire frequency band.

Claims

1. A method for echo cancellation and noise suppression, comprising:

performing linear echo cancellation (LEC) for a primary microphone channel on first and second frequency ranges, the second frequency range being higher than the first frequency range;

performing LEC for at least one secondary microphone channel on the first frequency range;

spatially processing the primary and secondary microphone channels over the first frequency range after LEC to form a channel; and

combining the channel after said spatial processing with the second frequency range of the primary microphone channel after LEC.

2. The method of claim 1, wherein the spatial processing comprises phase correlation-based masking.

32. A non-transitory computer-readable medium having instructions thereon, the instructions comprising code for causing a communication device to perform the method according to claim 1.

33. The method of claim 1, wherein the spatial processing is performed based on a probability that is high for negative values of echo return loss enhancement (ERLE) and low for positive values of ERLE.

34. The method of claim 1, said method comprising performing nonlinear post-processing on the channel after said combining, based on modeling harmonic distortions created due to a loudspeaker.

35. The method of claim 34, wherein said nonlinear post-processing is based on spectral subtraction.

39. The medium of claim 32, wherein the spatial processing is performed based on a probability that is high for negative values of echo return loss enhancement (ERLE) and low for positive values of ERLE.

40. The medium of claim 32, said method comprising performing nonlinear post-processing on the channel after said combining, based on modeling harmonic distortions created due to a loudspeaker.

3. A method for echo cancellation and noise suppression, comprising:

performing linear echo cancellation (LEC) for a primary microphone channel on at least a range of frequencies where echo is audible;

performing LEC for at least one secondary microphone channel only on a low frequency range over which spatial processing is effective; and

spatially processing the primary and secondary microphone channels over the low frequency range after LEC,

wherein the spatial processing is performed based on a doubletalk flag that is determined using a computed probability according to the equation

4. The method of claim 3, wherein no spatial processing is performed when the doubletalk flag indicates that far-end and near-end speech periods overlap or that echo attenuation provided by said performing LEC for the primary microphone channel is low.

29. The method of claim 3, wherein said doubletalk flag indicates when far-end speech and near-end speech periods overlap or when echo attenuation provided during linear echo cancellation is low.

5. A method for echo cancellation and noise suppression, comprising:

performing linear echo cancellation (LEC) for a primary microphone channel on at least a range of frequencies where echo is audible;

performing LEC for at least one secondary microphone channel only on a low frequency range over which spatial processing is effective;

spatially processing the primary and secondary microphone channels over the low frequency range after LEC to form a channel, wherein the spatial processing is performed based on a doubletalk flag that indicates when far-end speech and near-end speech periods overlap or when echo attenuation provided during linear echo cancellation is low; and

performing non-linear echo post-processing on the channel after said spatial processing, on at least the range of frequencies where echo is audible and based on the doubletalk flag, wherein the non-linear echo post-processing removes residual non-linear echo from the channel.

6. The method of claim 5, wherein said method comprises performing non-linear echo clipping based on a clipping flag that indicates high attenuation of at least a corresponding band of the primary microphone channel during linear echo cancellation, wherein the non-linear echo clipping removes residual echo from the channel after said post-processing,

wherein the clipping flag tracks echo attenuation during linear echo cancellation.

7. The method of claim 6, wherein the clipping flag is high if the echo attenuation during linear echo cancellation is above a predetermined threshold and low if the echo attenuation during linear echo cancellation is below a predetermined threshold.

30. The method of claim 5, wherein said doubletalk flag indicates when far-end speech and near-end speech periods overlap or when echo attenuation provided during linear echo cancellation is low.

31. The method of claim 5, said method comprising:

performing non-linear echo clipping on the channel after said spatial processing, based on clipping flags that indicate high attenuation of the primary microphone channel during linear echo cancellation, wherein the non-linear echo clipping removes residual echo from the channel; and

injecting comfort noise into the channel after clipping.

8. An apparatus for echo cancellation and noise suppression, comprising:

a linear echo canceller configured to perform linear echo cancellation (LEC) for a primary microphone channel on at least a range of frequencies where echo is audible and for at least one secondary microphone channel only on a low frequency range over which spatial processing is effective;

a spatial processor configured to spatially process the primary and secondary microphone channels over the low frequency range after LEC to form a channel; and

a post-processor configured to combine the channel after said spatial processing with the second frequency range of the primary microphone channel after LEC.

9. The apparatus of claim 8, wherein the spatial processor is configured to use phase correlation-based masking to spatially process the microphone channels.

10. The apparatus of claim 8, wherein the spatial processor is configured to operate based on a doubletalk flag that is determined using a computed probability according to the equation

11. The apparatus of claim 10, wherein the spatial processor is configured to not spatially process the microphone channels when the doubletalk flag indicates that far-end and near-end speech periods overlap or that echo attenuation provided by the linear echo canceller is low.

12. The apparatus of claim 8, said apparatus comprising a non-linear echo post-processor configured to perform non-linear echo post-processing on the channel after said spatial processing, on at least the range of frequencies where echo is audible and based on at least one doubletalk flag, wherein the non-linear echo post-processing removes residual non-linear echo from the channel.

13. The apparatus of claim 12, wherein said apparatus comprises a non-linear echo clipper configured to perform non-linear echo clipping based on a clipping flag that indicates high attenuation of at least a corresponding band of the primary microphone channel during linear echo cancellation, wherein the non-linear echo clipping removes residual echo from the channel after said post-processing, and

wherein the clipping flag tracks echo attenuation during linear echo cancellation.

14. The apparatus of claim 13, wherein the clipping flag is high if the echo attenuation during linear echo cancellation is above a predetermined threshold and low if the echo attenuation during linear echo cancellation is below a predetermined threshold.

36. The apparatus of claim 8, wherein the spatial processing is performed based on a probability that is high for negative values of echo return loss enhancement (ERLE) and low for positive values of ERLE.

37. The apparatus of claim 8, said method comprising performing nonlinear post-processing on the channel after said combining, based on modeling harmonic distortions created due to a loudspeaker.

38. The apparatus of claim 37, wherein said nonlinear post-processing is based on spectral subtraction.

15. An apparatus for echo cancellation and noise suppression, comprising:

means for performing linear echo cancellation (LEC) for a primary microphone channel on at least a range of frequencies where echo is audible;

means for performing LEC for at least one secondary microphone channel only on a low frequency range over which spatial processing is effective; and

means for spatially processing the primary and secondary microphone channels over the low frequency range after LEC to form a channel;

wherein the means for spatially processing is configured to operate based on a doubletalk flag that is determined using a computed probability according to the equation prob=

16. The apparatus of claim 15, wherein the means for spatially processing is configured to use phase correlation-based masking to spatially process the microphone channels.

17. The apparatus of claim 15, wherein said apparatus comprises means for combining the channel after said spatial processing with the second frequency range of the primary microphone channel after LEC.

18. The apparatus of claim 15, wherein the means for spatially processing is configured to not spatially process the microphone channels when the doubletalk flag indicates that far-end and near-end speech periods overlap or that echo attenuation provided by the means for performing LEC for the primary microphone channel is low.

19. The apparatus of claim 15, wherein the means for spatially processing is configured to operate based on the doubletalk flag, and

wherein said apparatus comprises means for performing non-linear echo post-processing on the channel after said spatial processing, on at least the range of frequencies where echo is audible and based on the doubletalk flag, wherein the non-linear echo post-processing removes residual non-linear echo from the channel.

20. The apparatus of claim 19, wherein said apparatus comprises means for performing non-linear echo clipping based on a clipping flag that indicates high attenuation of at least a corresponding band of the primary microphone channel during linear echo cancellation, wherein the non-linear echo clipping removes residual echo from the channel after said post-processing, and

wherein the clipping flag tracks echo attenuation during linear echo cancellation.

21. The apparatus of claim 20, wherein the clipping flag is high if the echo attenuation during linear echo cancellation is above a predetermined threshold and low if the echo attenuation during linear echo cancellation is below a predetermined threshold.

22. A computer-program product for echo cancellation and noise suppression, the computer-program product comprising a non-transitory computer-readable medium having instructions thereon, the instructions comprising:

code for causing a communication device to perform linear echo cancellation (LEC) for a primary microphone channel on at least a range of frequencies where echo is audible;

code for causing the communication device to perform LEC for at least one secondary microphone channel only on a low frequency range over which spatial processing is effective; and

code for causing the communication device to spatially process the primary and secondary microphone channels over the low frequency range after LEC to form a channel,

wherein the code for causing the communication device to spatially process is based on a doubletalk flag that is determined using a computed probability according to the equation

23. The computer-program product of claim 22, wherein the code for causing the communication device to spatially process comprises code for causing the communication device to perform phase correlation-based masking.

24. The computer-program product of claim 22, wherein said instructions comprise code for combining the channel after said spatial processing with the second frequency range of the primary microphone channel after LEC.

25. The computer-program product of claim 22, wherein the code for causing the communication device to spatially process comprises code for causing the communication device to not spatially process when the doubletalk flag indicates that far-end and near-end speech periods overlap or that echo attenuation provided by the communication device in response to said code for causing the communication device to perform LEC for a primary microphone channel is low.

26. The computer-program product of claim 22, wherein the code for causing the communication device to spatially process comprises code for causing the communication device to not spatially process when the doubletalk flag indicates that far-end and near-end speech periods overlap or that echo attenuation provided by the communication device in response to said code for causing the communication device to perform LEC for a primary microphone channel is low, and

wherein said instructions comprise code for causing the communication device to perform non-linear echo post-processing on the channel after said spatial processing, on at least the range of frequencies where echo is audible and based on the doubletalk flag, wherein the non-linear echo post-processing removes residual non-linear echo from the channel.

27. The computer-program product of claim 26, wherein said instructions comprise code for causing the communication device to perform non-linear echo clipping based on a clipping flag that indicates high attenuation of at least a corresponding band of the primary microphone channel during linear echo cancellation, wherein the non-linear echo clipping removes residual echo from the channel after said post-processing, and

wherein the clipping flag tracks echo attenuation during linear echo cancellation.

28. The computer-program product of claim 27, wherein the clipping flag is high if the echo attenuation during linear echo cancellation is above a predetermined threshold and low if the echo attenuation during linear echo cancellation is below a predetermined threshold.