Acoustic echo cancellation in an integrated audio and telecommunication system

• US 5,706,344 A
• Filed: 03/29/1996
• Issued: 01/06/1998
• Est. Priority Date: 03/29/1996
• Status: Expired due to Term

• Current Assignee: Digisonix Incorporated
• Original Assignee: Digisonix Incorporated
• Inventors: Finn, Brian M.
• Primary Examiner(s): Zele, Krista M.
• Assistant Examiner(s): SAINT SURIN, JACQUES M
• Application Number:
  US08/626,208
• Time in Patent Office: 648 Days
• Field of Search: 379/410, 379/406, 379/411, 379/420, 379/444, 379/159, 379/167, 379/388, 379/389, 381/71, 381/94, 381/66
• US Class Current: 379/406.09
• CPC Class Codes: G10K 2210/108 Communication systems, e.g. where useful sound is kept and noise is cancelled
• CPC Class Codes: G10K 2210/3039 Nonlinear, e.g. clipping, numerical truncation, thresholding or variable input and output gain
• CPC Class Codes: G10K 2210/3049 Random noise used, e.g. in model identification
• CPC Class Codes: G10K 2210/505 Echo cancellation, e.g. multipath-, ghost- or reverberation-cancellation
• CPC Class Codes: H04M 9/082 using echo cancellers echo cancellers per se H04B3/23

Abstract

Robust acoustic echo cancellation in a single-channel integrated audio and intercom system is provided by adapting echo cancellation filters in accordance with an optimized autocorrelation matrix. This is carried out in a modified FIR LMS filter in which the adaptation regressor signal is formed by a combination of one or more of the received voice signal, the received audio signal, and/or a random noise signal. In a SISO system, it is usually desirable to shelter the regressor signal from tonal components such as voice signals. In a multiple-channel system, it is usually desirable to shelter regressor signals for a particular channel from acoustic components that are likely to be correlated to acoustic components in other parts of the system and are likely to affect that particular channel. In multi-channel applications, it is therefore usually desirable to shelter regressor signals from audio signals.

Claims

1. In a telecommunication system having a near-end zone and a far-end zone, a near-end intercom system having a near-end microphone that senses sound in the near-end zone and a near-end loudspeaker that outputs sound into the near-end zone, and a far-end intercom system having a far-end microphone that senses sound in the far-end zone, and a far-end loudspeaker that outputs sound into the far-end zone, the improvement comprising:

a near-end audio summer that inputs a far-end voice signal from the far-end intercom system and an audio signal from an audio system, and outputs a near-end input signal that is transmitted to the near-end loudspeaker;

an adaptive near-end acoustic echo canceller that receives the near-end input signal and outputs a near-end echo cancellation signal;

a near-end echo cancellation summer that inputs a near-end output signal from a near-end microphone and the near-end echo cancellation signal from the near-end acoustic echo canceller, and outputs an echo-cancelled, near-end voice signal to the far-end intercom system; and

a multiplier that receives the echo-cancelled, near-end voice signal and a regressor signal, and outputs an update signal to adapt the acoustic echo canceller;

wherein the regressor signal is formed by the combination of one or more of the far-end voice signal, the audio signal and a random noise signal.

2. The invention as recited in claim 1 wherein the far-end voice signal inputting the near-end audio summer is an echo-cancelled, far-end voice signal.

3. The invention as recited in claim 1 wherein the near-end adaptive echo canceller is an adaptive FIR filter.

4. The invention as recited in claim 1 further comprising:

a random noise switch that can be closed to combine random noise with the regressor signal;

wherein the random noise switch is closed when a combination of the far-end voice signal and the audio signal exceed a threshold noise level.

5. The invention as recited in claim 4 wherein random noise is further combined with the near-end input signal when the random noise switch is closed.

6. The invention as recited in claim 1 further comprising an audio regressor switch that can be closed to combine the audio signal with the regressor signal.

7. The invention as recited in claim 1 further comprising a voice regressor switch that can be closed to combine the far-end voice signal with the regressor signal.

8. The invention as recited in claim 1 further comprising:

a near-end output gate/switch that monitors the near-end voice signal and terminates transmission of the near-end voice signal to the far-end intercom system when the near-end voice signal falls below a threshold value.

9. The invention as recited in claim 1 further comprising:

a near-end output gate/switch that monitors the near-end voice signal and reduces the level of transmission of the near-end voice signal to the far-end intercom system when the near-end voice signal falls below a threshold value.

10. A multi-channel telecommunications system comprising:

a plurality of zones;

one or more speaking locations in each zone;

a plurality of microphones, each sensing speech at a respective speaking location and outputting a voice signal;

an audio source that provides a plurality of audio signals;

a plurality of acoustic echo cancellers, each outputting an echo cancellation signal and being capable of inputting a combined voice and audio signal from the audio source and a respective microphone;

a plurality of speakers, each inputting one of the combined voice and audio signals and introducing sound into a respective zone at a respective speaking location;

a plurality of echo cancellation summers, each corresponding to a respective microphone, each echo cancellation summer inputting the respective voice signal from the respective microphone and an echo cancellation signal from each adaptive echo canceller associated with the zone in which the microphone is located, and outputting an echo-cancelled voice signal; and

a multiplier associated with each acoustic echo canceller, each multiplier receiving an echo-cancelled voice signal and a regressor signal, and outputting an update signal to the respective acoustic echo canceller;

wherein the acoustic echo canceller inputs a combined voice and audio signal and the regressor signal does not include the audio signal.

11. A multi-channel telecommunications system as recited in claim 10 wherein the regressor signal for each acoustic echo canceller includes random noise when the combination of the voice signal and the audio signal exceeds a threshold noise value.

12. A multi-channel telecommunications system as recited in claim 11 wherein the level of random noise increases as the level of the combined voice signal and audio signal increases.

13. A multi-channel telecommunications system as recited in claim 10 further comprising:

a random noise switch that can be closed to combine random noise with the regressor signal.

14. A multi-channel telecommunications system as recited in claim 10 further comprising:

an output gate associated with each microphone that monitors the level of the echo-cancelled voice signal and terminates transmission of the echo-cancelled voice signal when the echo-cancelled voice signal falls below a threshold value.

15. A multi-channel telecommunications system recited in claim 10 further comprising:

an output gate/switch associated with each microphone that monitors the level of the echo-cancelled voice signal and reduces the level of transmission of the echo-cancelled voice signal when the echo-cancelled voice signal falls below a threshold value.

16. A multi-channel telecommunications system as recited in claim 10 wherein each of the zones in a vehicle.

17. A multi-channel telecommunications system as recited in claim 16 wherein there are two zones and two speaking locations in each zone.

18. A multi-channel telecommunications system as recited in claim 10 further comprising a voice regressor switch that can be closed to combine the voice signal with the regressor signal.

19. In a multi-channel telecommunications system having one or more speaking locations in a plurality of zones, a plurality of microphones, each sensing speech at a respective speaking location, and a plurality of loudspeakers, each introducing sound into a respective zone at a respective speaking location, a method of acoustic echo cancellation comprising the steps of:

outputting a voice signal from each respective microphone in response to speech in the respective speaking location;

combining each respective voice signal with one or more audio signals;

transmitting each combined voice and audio signal to at least one speaker located in other zones;

adaptively modeling the acoustic path between each speaker and each microphone located in each respective zone;

cancelling the echo in a voice signal from each respective microphone by filtering the combined voice and audio signals transmitted to the speaker in the same speaking location through the respective adaptive model for the acoustic path and subtracting the filtered signal from the respective voice signal; and

updating each adaptive model by multiplying the respective echo-cancelled voice signal by a regressor signal wherein the regressor signal is sheltered to not include any audio signal components.

20. A method as recited in claim 19 further comprising the step of:

adding random noise to the regressor signal when the combined value of the voice signal from the other zone and the audio signal exceed threshold value.

21. A method as recited in claim 20 further comprising the step of:

reducing the level of transmission of the echo-cancelled voice signal to the other zones when the echo-cancelled voice signal is less than a threshold value.

22. A method as recited in claim 20 further comprising the step of:

terminating the transmission of the echo-cancelled voice signal to the other zones when the echo-cancelled voice signal is less than a threshold value.