Detecting double talk in acoustic echo cancellation using zero-crossing rate

1. A method for acoustic echo cancellation in a communication device, the method comprising:

receiving a first near-end audio signal in the communication device, wherein the first near-end audio signal comprises acoustic echo of a far-end audio signal reproduced by the communication device; and

performing echo cancellation on the first near-end audio signal to generate a second near-end audio signal with at least some of the acoustic echo removed, wherein the echo cancellation is performed responsive to presence or absence of double-talk (DT), and wherein a single zero-crossing rate (ZCR), computed at the output of the echo canceller, is used to detect the presence or absence of DT.

2. The method of claim 1 , wherein performing echo cancellation comprises:

applying an adaptive filter to the far-end audio signal to estimate the acoustic echo in the near-end audio signal, wherein current filter coefficients are used when the ZCR indicates the absence of DT and coefficients of the adaptive filter when current DT was initially detected are used when the ZCR indicates the presence of DT; and

subtracting the estimated acoustic echo from the first near-end audio signal to generate the second near-end audio signal.

3. The method of claim 1 , wherein the ZCR is computed over a window of samples of size M in an echo-cancelled near-end audio signal and is computed for every Kth sample in the echo-cancelled near-end audio signal, wherein M and K are integers, and wherein the echo-cancelled near-end audio signal is generated by applying an adaptive filter to the far-end audio signal to estimate the acoustic echo in the first near-end audio signal and subtracting the estimated acoustic echo from the first near-end audio signal, wherein coefficients used by the filter are updated each time a sample of the echo-cancelled near-end audio signal is generated.

4. The method of claim 3 , wherein M=1000 and K=1.

5. A method for acoustic echo cancellation in a communication device, the method comprising:

receiving a first near-end audio signal in the communication device, wherein the first near-end audio signal comprises acoustic echo of a far-end audio signal reproduced by the communication device;

computing a first estimate of the acoustic echo in a sample of the first near-end audio signal by applying an adaptive filter using first coefficients to a sample of the far-end audio signal;

subtracting the first estimate of the acoustic echo from the sample of the first near-end audio signal to generate a sample of a second near-end audio signal;

detecting double-talk based on a single zero-crossing rate (ZCR) of the second near-end audio signal, wherein the single zero-crossing rate is computed at the output of the echo canceller; and

outputting a sample of a third near-end audio signal responsive to the double-talk detection, wherein the sample of the third near-end audio signal is generated by

computing a second estimate of the acoustic echo in the sample of the first near-end audio signal by applying an adaptive filter using second coefficients to the sample of the far-end audio signal, wherein the second coefficients are coefficients of the adaptive filter immediately prior to when a current double-talk period began; and

subtracting the second estimate of the acoustic echo from the sample of the first near-end audio signal.

6. The method of claim 5 , further comprising:

computing the ZCR over a window of samples of size M in the second near-end audio signal, wherein M is an integer.

7. The method of claim 6 , wherein the ZCR is computed for each Kth sample in the near-end audio signal, where K is an integer.

8. The method of claim 7 , wherein M=1000 and K=1.

9. A communication device comprising:

means for reproducing a far-end audio signal; and

means for receiving a near-end audio signal comprising acoustic echo of the far-end audio signal;

wherein the communication device is configured to perform echo cancellation on the first near-end audio signal to generate a second near-end audio signal with at least some of the acoustic echo removed,

wherein the echo cancellation is performed responsive to presence or absence of double-talk (DT), and wherein a single zero-crossing rate (ZCR), computed at the at the output of the echo canceller, is used to detect the presence or absence of DT.

10. The communication device of claim 9 , wherein the echo cancellation is performed by:

applying an adaptive filter to the far-end audio signal to estimate the acoustic echo in the near-end audio signal, wherein current filter coefficients are used when the ZCR indicates the absence of DT and coefficients of the adaptive filter when current DT was initially detected are used when the ZCR indicates the presence of DT; and

subtracting the estimated acoustic echo from the first near-end audio signal to generate the second near-end audio signal.

11. The communication device of claim 10 , wherein the ZCR is computed over a window of samples of size M in an echo-cancelled near-end audio signal and is computed for every Kth sample in the echo-cancelled near-end audio signal, wherein M and K are integers, and wherein the echo-cancelled near-end audio signal is generated by applying an adaptive filter to the far-end audio signal to estimate the acoustic echo in the first near-end audio signal and subtracting the estimated acoustic echo from the first near-end audio signal, wherein coefficients used by the filter are updated each time a sample of the echo-cancelled near-end audio signal is generated.

12. The communication device of claim 11 , wherein M=1000 and K=1.