Acoustic echo cancellation using a variable length adaptive filter
An audio conferencing system having a base station, a loud speaker and one or more microphones generally operates to receive a F.E. audio signal, play the signal, and to receive a N.E. audio signal at the microphone that includes the F.E. audio signal modified by an echo path and a N.E. audio signal. The base station includes an adaptive filter whose length is controlled to be variable depending upon activity detected at each of one or more filter elements comprising the adaptive filter.
1. In an audio system, a method for cancelling acoustic echo in a microphone signal, comprising:
activating fewer than all of a plurality of filter elements allocated for use by an adaptive filter;
detecting a filter element activity parameter in a last one of the activated filter elements to be greater than a specified threshold value, and activating at least a additional one of the plurality of filter elements allocated for use by the adaptive filter;
calculating an adaptive filter coefficient for each one of the plurality of active filter elements, and each of the active filter elements using the calculated adaptive filter coefficient and audio information in a sample of a far end audio signal to calculate an estimate of an acoustic echo in a microphone signal; and
the adaptive filter subtracting a sum of the estimates of the acoustic echo calculated by each of the active filter elements from the microphone signal.
2. The method of claim 1 , wherein the audio system has two or more microphones.
3. The method of claim 2 , wherein a separate adaptive filter is associated with each of the two or more microphones.
4. The method of claim 1 , wherein the filter element activity parameter is a root-mean-squared value.
5. The method of claim 1 , wherein the specified threshold value is a root-mean-squared value.
6. The method of claim 1 , wherein the audio information in the sample of the far end audio signal comprises a transformed audio signal frequency component and a corresponding amplitude component.
7. In an audio system, a method for cancelling acoustic echo in a microphone signal, comprising:
activating fewer than all of a plurality of filter elements allocated for use by an adaptive filter;
detecting a filter element activity parameter in a last one of the plurality of active filter elements comprising an adaptive filter to be less than a specified threshold value and de-activating the last one of the currently active filter elements;
calculating a different adaptive filter coefficient for each of the remaining plurality of active filter elements, and each of the remaining plurality of activate filter elements using the adaptive filter coefficient and audio information in a sample of a far end audio signal to calculate an estimate of an acoustic echo in a microphone signal; and
the adaptive filter subtracting a sum of the estimates of the acoustic echo calculated by each of the remaining plurality of the active filter elements from the microphone signal.
8. The method of claim 7 , wherein the audio system has two or more microphones.
9. The method of claim 8 , wherein a separate adaptive filter is associated with each of the two or more microphones.
10. The method of claim 7 , wherein the filter element activity parameter is a root-mean-squared value.
11. The method of claim 7 , wherein the specified threshold value is a root-mean-squared value.
12. The method of claim 7 , wherein the audio information in the sample of the far end audio signal comprises a transformed audio signal frequency component and a corresponding amplitude component.
13. An audio system for cancelling acoustic echo in a microphone signal, comprising:
a loudspeaker operating to play a far end audio signal;
a microphone operating to receive a near end audio signal and audio information corresponding to the far end audio signal played by the loudspeaker and received by the microphone over an acoustic path; and
an adaptive filter having a plurality of active and inactive filter element operating to detect an activity parameter for at least a last one of the plurality of active filter elements comprising the adaptive filter to be less than a specified threshold value, and deactivating the last one of the plurality of currently active filter elements, calculating a different adaptive filter coefficient for each of the remaining plurality of active filter elements, each of the remaining plurality of activate filter elements using the adaptive filter coefficient and audio information in a sample of a far end audio signal to calculate an estimate of an acoustic echo in a microphone signal, and the adaptive filter subtracting a sum of the estimates of the acoustic echo calculated by each of the remaining plurality of the active filter elements from the microphone signal.