Spatial quiescence protection for multi-channel acoustic echo cancellation
A system performs residual echo suppression on a microphone signal that receives, e.g., voice commands, and that is exposed to echo from multiple speakers. An example is a smartphone that receives voice commands while the smartphone is playing music through stereo speakers. The system estimates residual echo level in different ways, and determines which estimate to use. The technique responds well to the difficult to handle scenario of a spatially quiescent image suddenly transitioning to a spatially rich image. Even in the face of such difficult scenarios, the system detects and removes residual echo from the microphone signal, instead of allowing the undesired residual echo to pass through.
1. A system comprising:
multiple receive out channels;
a microphone input channel; and
decision circuitry configured to:
determine a nominal level of echo suppression for the microphone input channel responsive to an estimated receive out signal level and an echo cancellation estimate;
detect a transition from spatial quiescence to spatial richness in the multiple receive out channels; and
in response to the transition, trigger a change comprising additional echo suppression on the microphone input channel over the nominal amount.
2. The system of claim 1 , where the decision circuitry is further configured to:
prevent the change in an absence of the transition.
3. The system of claim 1 , where the microphone input channel comprises a microphone input signal after echo cancellation.
4. The system of claim 1 , where spatial richness comprises a change in spatial image among the multiple receive out channels.
5. The system of claim 1 , where the decision circuitry is configured to determine spatial richness and spatial quiescence according to a correlation measure between the multiple receive out channels.
6. The system of claim 5 , further comprising:
a cross channel estimator configured to provide the correlation measure.
7. The system of claim 6 , where the correlation measure comprises either or both of a prediction gain and a prediction error level.
8. The system of claim 1 , where the decision circuitry is further configured to transition, after triggering increased residual echo suppression, back to the nominal amount.
9. The system of claim 1 , further comprising:
a cross channel estimator configured to perform an analysis of the multiple receive out channels and determine a correlation error level responsive to the analysis;
where the decision circuitry is configured to determine a second residual echo level estimate using the correlation error level; and
where the decision circuitry is configured to detect the transition when:
the second residual echo level estimate exceeds the first residual echo level estimate.
10. The system of claim 1 where the decision circuitry is configured to:
determine a second echo level estimate responsive to a correlation among the multiple receive out channels; and
provide the second echo level estimate to echo suppression circuitry for the microphone input channel.
11. The system of claim 7 , where the cross channel estimator is configured to detect the transition responsive to an increase in the prediction error, a decrease in the prediction gain, or both.
12. The system of claim 1 where the decision circuitry is configured to determine an echo level estimate for each of the multiple receive out channels.
13. A system comprising:
a microphone input;
a first audio output;
a second audio output;
circuitry in communication with the microphone input, first audio output, and the second audio output, the circuitry configured to:
determine a first residual echo level estimate for the microphone input according to an estimated audio out signal level and an echo cancellation estimate;
determine a second residual echo level estimate for the microphone input, the circuitry configured to adjust the second residual echo level estimate responsive to a correlation between the first audio output and the second audio output;
responsive to the correlation, detect a transition between spatial quiescence to spatial richness; and
responsive to the transition,
provide the second residual echo level estimate to residual echo suppression circuitry for the microphone input to trigger a change comprising additional echo suppression over a nominal amount that would be applied when the first residual echo level estimate is provided.
14. The system of claim 13 , where the circuitry is further configured to determine the correlation by:
determining a prediction error level between the first and second audio outputs.
15. The system of claim 14 , where the circuitry is further configured to determine the second residual echo level estimate responsive to a duration of the correlation.
16. The system of claim 14 , where the circuitry is further configured to determine the second residual echo level estimate responsive to duration and strength of the correlation.
17. A method comprising:
detecting a transition from spatial quiescence to spatial richness among multiple receive out channels; and
in response to the transition, triggering a change comprising additional echo suppression on a microphone input channel over a nominal amount that would be applied when a first residual echo level estimate for the microphone input channel is determined according to an estimated receive out signal level and an echo cancellation estimate.
18. The method of claim 17 , where detecting the transition comprises detecting a change in spatial image among the multiple receive out channels.
19. The method of claim 17 , where detecting the transition comprises determining a correlation measure between the multiple receive out channels.
20. The method of claim 17 , further comprising detecting the transition comprises detecting an increase in a prediction error, a decrease in a prediction gain, or both.