Gain and spectral shape adjustment in audio signal processing

• US 8,565,415 B2
• Filed: 10/07/2008
• Issued: 10/22/2013
• Est. Priority Date: 10/08/2007
• Status: Active Grant

• Current Assignee: Cerence Inc., Cerence Operating Company (Cerence Inc.)
• Original Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Inventors: Schmidt, Gerhard Uwe, Iser, Bernd
• Primary Examiner(s): Islam, Mohammad
• Assistant Examiner(s): Huynh, Van D
• Application Number:
  US12/246,994
• Publication Number:
  US 20090225980A1
• Time in Patent Office: 1,841 Days
• Field of Search: 37940601-40602, 379/406.08, 270/288, 708/304, 381/71.1, 381/71.11, 381/92, 381/93, 381/66
• US Class Current: 379/406.03
• CPC Class Codes: G10L 2021/02082 the noise being echo, reverberation of the speech
• CPC Class Codes: H04M 1/6075 adapted for handsfree use in a vehicle H04M1/6058 takes precedence; arrangements for holding telephones in a vehicle B60R11/0241
• CPC Class Codes: H04M 9/082 using echo cancellers echo cancellers per se H04B3/23

Abstract

A signal processing system enhances communication. When an audio signal is detected, an echo component of the detected audio signal may be estimated. A near party communication device may receive an audio signal from a remote party communication device. A characteristic of the received audio signal may be adjusted based on the echo component of the detected audio signal.

Claims

1. A method for processing an audio signal, comprising:

estimating an echo component of a microphone signal detected at a near party communication device;

receiving an audio signal transmitted from a remote party communication device to the near party communication device;

adjusting a gain of the received audio signal based on the echo component of the microphone signal; and

adjusting a spectral shape of the received audio signal based on the echo component of the microphone signal by

modeling a desired spectral shape with an infinite impulse response filter using at least the echo component to obtain a first model result;

modeling the desired spectral shape with a finite impulse response filter using at least the echo component to obtain a second model result;

comparing the first model result with the second model result;

selecting the infinite impulse response filter to adjust a spectral shape of the received audio signal when the first model result is a better match for the desired spectral shape; and

selecting the finite impulse response filter to adjust the spectral shape of the received audio signal when the second model result is a better match for the desired spectral shape.

2. The method of claim 1, where the act of adjusting the spectral shape further comprises modifying a frequency envelope of the received audio signal based on the echo component of the microphone signal.

3. The method of claim 1, where the echo component of the microphone signal is due to a loudspeaker output of an audio signal received from the remote party communication device.

4. The method of claim 1, where the near party communication device is associated with a microphone and a loudspeaker, where the act of estimating the echo component comprises analyzing the microphone signal to identify a portion of the microphone signal that is due to the loudspeaker outputting an audio signal received from the remote party communication device.

5. The method of claim 1, further comprising:

estimating a noise component of the microphone signal; and

where the act of adjusting the gain or spectral shape of the received audio signal is based on the echo component and the noise component of the microphone signal.

6. The method of claim 1, wherein adjusting all the spectral shape further comprises:

passing the received audio signal through the infinite impulse response filter to adjust the spectral shape of the received audio signal based on an inverse discrete Fourier transform of a squared magnitude of the desired spectral shape;

or

passing the received audio signal through the finite impulse response fitter to adjust the spectral shape of the received audio signal based on an inverse discrete Fourier transform of an inverse of a squared magnitude of the desired spectral shape.

7. The method of claim 1, where the act of adjusting a gain or spectral shape of the received audio signal is based on an average peak echo-to-noise ratio of the microphone signal.

8. The method of claim 7, where the average peak echo-to-noise ratio is smoothed in frequency by an infinite impulse response filter.

9. A method for audio signal processing, comprising:

receiving an audio signal at a near party communication device that was transmitted from a remote party communication device;

estimating local background noise at the near party communication device;

estimating local acoustic echo at the near party communication device; and

adjusting a gain of the received audio signal based on the estimated local background noise and the estimated local acoustic echo; and

adjusting a spectral shape of the received audio signal based on the estimated local background noise and the estimated local acoustic echo by

obtaining a first model result using an infinite impulse response filter;

obtaining a second model result using a finite impulse response filter;

comparing the first model result with the second model result for each frame; and

for each frame, adjusting the spectral shape of the received audio signal based on the first model result or the second model result that better matches a desired spectral shape.

10. The method of claim 9, where the local acoustic echo is due to a loudspeaker output of an audio signal received from the remote party communication device.

11. The method of claim 9, where the spectral shape of the received audio signal is adjusted by the infinite impulse response filter based on an inverse discrete Fourier transform of a squared magnitude of the desired spectral shape or the finite impulse response filter based on an inverse discrete Fourier transform of an inverse of a squared magnitude of the desired spectral shape,

12. The method of claim 9, where the gain or spectral shape of the received audio signal is adjusted based on an average peak echo-to-noise ratio of the detected audio signal.

13. A signal processing system, comprising:

an echo compensation filter configured to filter an audio signal detected at a near party communication device and output an estimate of an echo component of the detected audio signal; and

a signal adjustment unit with a finite impulse response filter and an infinite impulse response filter, wherein the signal adjustment unit is coupled with the echo compensation filter and configured to receive an audio signal from a remote party communication device, adjust a gain of the received audio signal based on the echo component of the detected audio signal, and adjust a spectral shape of the received audio signal based on the echo component by

modeling a desired spectral shape with the infinite impulse response filter using at least the echo component to obtain a first model result;

modeling the desired, spectral shape with the finite impulse response filter using, at least the echo component to obtain a second model result;

comparing the first model result with the second model result;

selecting the infinite impulse response filter to adjust the spectral shape of the received audio signal when the first model result is a better match for the desired spectral shape; and

selecting the finite impulse response filter to adjust the spectral shape of the received audio signal when the second model result is a better match for the desired spectral shape.

14. The system of claim 13, where the echo component of the detected audio signal is due to a loudspeaker output of an audio signal received from the remote party communication device.

15. The system of claim 13, further comprising

a noise reduction filter configured to filter the detected audio signal and output an estimate of a noise component of the detected audio signal;

and where the signal adjustment unit is configured to adjust a gain or spectral shape of the received audio signal based on the echo component and the noise component of the detected audio signal.

16. The system of claim 13, where the signal adjustment unit is configured to adjust a gain or spectral shape of the received audio signal based on an average peak echo-to-noise ratio of the detected audio signal.

17. The system of claim 13, where the detected audio signal comprises a microphone signal, the system further comprising:

a first analysis filter bank configured to divide the microphone signal into microphone sub-band, signals;

a second analysis filter bank configured to divide a reference signal into reference sub-band signals to be input into the echo compensation filter to reduce echo in the microphone sub-band signals; and

a synthesis filter bank configured to synthesize the microphone sub-band signals filtered by the echo compensation filter.

18. The system of claim 13, further comprising a transmitter configured to transmit the detected audio signal filtered by the echo compensation filter to a remote party communication. device.

19. The system of claim 13, further comprising an amplifier, where the echo compensation filter is configured to receive a reference signal for adaptation of echo compensation filter coefficients either before or after amplification of the received audio signal processed by the signal adjustment unit.

20. A signal processing system, comprising:

a loudspeaker configured to output a loudspeaker signal;

a microphone configured to convert sound into an operational signal;

a noise reduction filter configured to filter the operational signal and output an estimate of noise;

an echo compensation fitter configured to filter the operational signal and output an estimate of echo; and

a signal adjustment unit with a finite impulse response filter and an infinite impulse response filter, wherein the signal adjustment unit is in communication with the noise reduction filter and the echo compensation filter, where the signal adjustment unit is configured to receive an audio signal from a remote party communication device and adjust a spectral shape of the received audio signal based on the estimate of noise and the estimate of echo by

modeling a desired spectral shape with the infinite impulse response filter using at least the echo component to obtain a first model result;

modeling the desired spectral shape with the finite impulse response filter using, at least the echo component to obtain a second model result;

comparing the first model result with the second model result;

selecting the infinite impulse response filter to adjust the spectral shape of the received audio signal when the first model result is a better match for the desired spectral shape; and

selecting the finite impulse response filter to adjust the spectral shape of the receivedaudio signal when the second model result is a better match for the desired spectral shape.

21. The system of claim 20, where the echo in the operational signal is due to detection of the loudspeaker signal at the microphone, where the loudspeaker signal is based on an audio signal received from the remote party communication device.

22. The system of claim 20, where the microphone comprises a directional microphone of a microphone array, further comprising a beamformer configured to receive the operational signal from the microphone array and obtain a beamformed signal.