Audio source proximity estimation using sensor array for noise reduction

• US 8,218,397 B2
• Filed: 10/22/2009
• Issued: 07/10/2012
• Est. Priority Date: 10/24/2008
• Status: Active Grant

• Current Assignee: Qualcomm, Inc.
• Original Assignee: Qualcomm, Inc.
• Inventors: Chan, Kwokleung
• Primary Examiner(s): Pihulic, Daniel
• Application Number:
  US12/603,824
• Publication Number:
  US 20100103776A1
• Time in Patent Office: 992 Days
• Field of Search: 367/124
• US Class Current: 367/124
• CPC Class Codes: G01S 11/14 using ultrasonic, sonic, or infrasonic waves
• CPC Class Codes: G10L 21/0208 Noise filtering
• CPC Class Codes: G10L 25/48 specially adapted for particular use
• CPC Class Codes: H04R 3/005 for combining the signals of two or more microphones specially adapted for hearing aids H04R25/407

Abstract

Estimating the proximity of an audio source is accomplished by transforming audio signals from a plurality of sensors to frequency domain. The amplitudes of the transformed audio signals are then determined. The proximity of the audio source is determined based on a comparison of the frequency domain amplitudes. This estimation permits a device to differentiate between relatively distant audio sources and audio sources at close proximity to the device. The technique can be applied to mobile handsets, such as cellular phones or PDAs, hands-free headsets, and other audio input devices. Devices taking advantage of this “close proximity” detection are better able to suppress background noise and deliver an improved user experience.

Claims

1. A method of determining proximity of an audio source, comprising:

band-pass filtering a plurality of audio signals from a plurality of sensors;

transforming the plurality of audio signals from the plurality of sensors to frequency domain;

determining amplitudes of the transformed audio signals;

comparing the amplitudes; and

determining the proximity of the audio source based on the comparison of the amplitudes.

2. The method of claim 1, further comprising:

beamforming each of the audio signals.

3. The method of claim 1, further comprising:

beamforming each of the transformed audio signals.

4. The method of claim 1, further comprising:

amplifying each of the audio signals by a respective correction factor.

5. The method of claim 1, wherein transforming includes applying an FFT, DCT, DFT, wavelet transformation, or any suitable combination of the foregoing transformations to the audio signals.

6. The method of claim 1, further comprising:

dividing the audio signals into a plurality of frames;

determining an amplitude of each transformed audio signal for each of the frames, thereby producing the amplitudes;

smoothing the amplitudes over the frames;

comparing the smoothed amplitudes of the transformed audio signals to one another to produce at least one differential signal; and

determining the proximity of the audio source based on the differential signal.

7. The method of claim 6, further comprising:

normalizing the differential signal.

8. The method of claim 6, further comprising:

converting the smoothed amplitudes to a log domain.

9. The method of claim 6, further comprising:

applying a non-linear function to the differential signal to produce a proximity score.

10. The method of claim 9, wherein the non-linear function is a sigmoid function.

11. A method of determining proximity of an audio source, comprising:

receiving a plurality of audio signals from a plurality of sensors;

transforming the audio signals to frequency domain;

determining amplitudes of the transformed audio signals at a plurality of frequencies;

for each of the frequencies, determining a differential signal by comparing the amplitudes corresponding to the frequency, whereby determining a plurality of differential signals; and

determining the proximity of the audio source based on the differential signals.

12. The method of claim 11, further comprising:

Comparing the differential signal at each of the frequencies to a predetermined threshold; and

determining a near-field flag at each of the frequencies, based on the comparison of the differential signal and the predetermined threshold for the frequency.

13. The method of claim 11, further comprising;

dividing the audio signals into a plurality of frames; and

determining the amplitudes for each of the frames.

14. The method of claim 13, further comprising:

smoothing the amplitudes over the frames.

15. The method of claim 14, further comprising:

converting the smoothed amplitudes to a log domain.

16. The method of claim 11, further comprising:

normalizing the differential signals to determine a proximity score at each of the frequencies.

17. The method of claim 11, further comprising:

applying a non-linear function to the differential signals to produce a proximity score at each of the frequencies.

18. An apparatus, comprising:

a plurality of audio sensors outputting a plurality of audio signals in response to an audio source;

a proximity detection module configured to transform the audio signals to frequency domain and to determine the proximity of the audio source by comparing amplitudes of the transformed audio signals; and

a noise reduction/suppression module responsive to output from the proximity detection module, wherein the reduction/suppression module is configured to estimate an audio signal spectrum and a noise signal spectrum.

19. The apparatus of claim 18, wherein the apparatus is a headset.

20. The apparatus of claim 19, wherein the headset is a wireless headset.

21. The apparatus of claim 18, further comprising:

a microphone calibration module.

22. The apparatus of claim 18, further comprising:

a voice activity detection (VAD) module.

23. The apparatus of claim 18, further comprising:

an echo cancellation module.

24. An apparatus, comprising:

means for transforming a plurality of audio signals from a plurality of sensors to frequency domain;

means for determining amplitudes of the transformed audio signals;

means for comparing the amplitudes; and

means for determining the proximity of the audio source based on the comparison of the amplitudes; and

means for noise reduction/suppression responsive to the means for determining the proximity of the audio source, wherein the means for noise reduction/suppression is configured to estimate an audio signal spectrum and a noise signal spectrum.

25. A non-transitory computer-readable medium embodying a set of instructions executable by one or more processors, comprising:

code for transforming a plurality of audio signals from a plurality of sensors to frequency domain;

code for determining amplitudes of the transformed audio signals;

code for comparing the amplitudes;

code for determining the proximity of the audio source based on the comparison of the amplitudes;

code for noise reduction/suppression responsive to the code for determining the proximity of the audio source; and

code for estimating an audio signal spectrum and a noise signal spectrum.

26. The non-transitory computer-readable medium of claim 25, further comprising:

code for voice activity detection.

27. A method of determining proximity of an audio source, comprising:

receiving a plurality of audio signals from a plurality of sensors;

beamforming the audio signals;

determining amplitudes of the beamformed audio signals;

comparing the amplitudes; and

determining the proximity of the audio source based on the comparison of the amplitudes.

28. The method of claim 27, wherein determining comprising:

computing a near field score; and

determining the proximity of the audio source based on the near field score.

29. The method of claim 27, wherein determining comprising:

computing a near field probability value; and

determining the proximity of the audio source based on the near field probability value.

30. The method of claim 27, further comprising:

amplifying each of the audio signals by a respective correction factor.

31. The method of claim 27, further comprising:

dividing the audio signals into a plurality of frames;

determining an amplitude of each beamformed audio signal for each of the frames, whereby producing the amplitudes;

smoothing the amplitudes over the frames;

comparing the smoothed amplitudes of the beamformed audio signals to one another to produce at least one differential signal; and

determining the proximity of the audio source based on the differential signal.

32. The method of claim 31, further comprising:

normalizing the differential signal.

33. The method of claim 31, further comprising:

applying a non-linear function to the differential signal to produce a proximity score.