Multi-aural MMSE analysis techniques for clarifying audio signals

• US 10,149,047 B2
• Filed: 06/18/2014
• Issued: 12/04/2018
• Est. Priority Date: 06/18/2014
• Status: Active Grant

• Current Assignee: Cirrus Logic Incorporated
• Original Assignee: Cirrus Logic Incorporated
• Inventors: Geiger, Fredrick D., Bunderson, Bryant V., Grundstrom, Carl, Sherwood, William Erik
• Primary Examiner(s): Patel, Yogeshkumar
• Application Number:
  US14/308,541
• Publication Number:
  US 20150373453A1
• Time in Patent Office: 1,630 Days
• Field of Search: 381 7111, 381 7114, 704226
• CPC Class Codes: G10L 2021/02165 Two microphones, one receiving mainly the noise signal and the other one mainly the speech signal
• CPC Class Codes: G10L 21/02 Speech enhancement, e.g. noise reduction or echo cancellation reducing echo effects in line transmission systems H04B3/20; echo suppression in hands-free telephones H04M9/08
• CPC Class Codes: G10L 25/27 characterised by the analysis technique
• CPC Class Codes: H04R 2410/05 Noise reduction with a separate noise microphone
• CPC Class Codes: H04R 2499/11 Transducers incorporated or for use in hand-held devices, e.g. mobile phones, PDA's, camera's
• CPC Class Codes: H04R 3/00 Circuits for transducers , loudspeakers or microphones

Abstract

Techniques for processing audio signals include removing noise from the audio signals or otherwise clarifying the audio signals prior to outputting the audio signals. The disclosed techniques may employ minimum mean squared error (MMSE) analyses on audio signals received from a primary microphone and at least one reference microphone, and to techniques in which the MMSE analyses are used to reduce or eliminate noise from audio signals received by the primary microphone. Optionally, confidence intervals may be assigned to different frequency bands of an audio signal, with each confidence interval corresponding to a likelihood that its respective frequency band includes targeted audio, and each confidence interval representing a contribution of its respective frequency band in a reconstructed audio signal from which noise has been removed.

Claims

1. A method for clarifying an audio signal comprising:

receiving a primary audio signal and a reference audio signal, each audio signal including a plurality of frequency bands, an unknown target component, and an unknown noise component;

determining a noise estimate of the unknown noise component from the reference audio signal;

incorporating the noise estimate into a minimum mean squared error analysis;

subjecting each frequency band of the plurality of frequency bands of the primary audio signal to the minimum mean squared error analysis;

assigning a confidence interval as a measure of statistical likelihood of dominance of the unknown target component in each frequency band of the plurality of frequency bands based on a result of the minimum mean squared analysis;

modifying an audio output level of each frequency band of the primary audio signal based on the confidence interval of that frequency band to provide a modified output frequency band; and

combining the modified output frequency bands for each frequency band of the plurality of frequency bands of the primary audio signal to provide a clarified output audio signal substantially reduced in the unknown noise component.

2. The method of claim 1, wherein the determining the noise estimate includes comparing at least one frequency band from the primary audio signal with at least one corresponding frequency band of the reference audio signal.

3. A method for clarifying an audio signal comprising:

The method of claim 2, wherein determining the noise estimate includes comparing the plurality of frequency bands of the primary audio signal with the corresponding plurality of frequency bands of the reference audio signal.

4. The method of claim 1, further comprising:

subjecting each frequency band of the plurality of frequency bands of the reference audio signal to the minimum mean squared error analysis.

5. The method of claim 1, wherein assigning the confidence interval indicative of the statistical likelihood of presence of the unknown target component includes:

assigning a very low confidence interval to a frequency band of the reference audio signal having a greater power than a corresponding frequency band of the primary audio signal;

assigning a low confidence interval to a frequency band of the reference audio signal having substantially the same power as a corresponding frequency band of the primary audio signal; and

assigning a high confidence interval to a frequency band of the primary audio signal having a greater power than a corresponding frequency band of the reference audio signal.

6. The method of claim 5, wherein assigning the very low confidence interval comprises assigning a gain of less than 0.3 to the frequency band.

7. The method of claim 5, wherein assigning the low confidence interval comprises assigning a gain of about 0.5 or less to the frequency band.

8. The method of claim 5, wherein assigning the high confidence interval comprises assigning a gain of greater than 0.6 to the frequency band.

9. The method of claim 1, wherein assigning the confidence interval comprises assigning an appropriate gain to each frequency band.

10. The method of claim 1, wherein assigning the confidence interval comprises dynamically estimating noise in the audio signals.

11. The method of claim 1, conducted without detecting voice activity.

12. The method of claim 1, further comprising:

subjecting the primary audio signal and the reference audio signal to an adaptive time domain filter.

13. The method of claim 12, wherein subjecting the primary audio signal and the reference audio signal to the adaptive time domain filter comprises subjecting the primary audio signal and the reference audio signal to a least mean square filter.

14. The method of claim 12, wherein subjecting the primary audio signal and the reference audio signal to the adaptive time domain filter comprises subjecting the primary audio signal and the reference audio signal to the adaptive time domain filter before subjecting each frequency band of the pluralities of frequency bands of the primary and the reference audio signals to the minimum mean squared error analyses.

15. A method for clarifying an audio signal comprising:

receiving a primary audio signal and a reference audio signal, each audio signal including a plurality of frequency bands, an unknown target component, and an unknown noise component;

subjecting the primary audio signal to an adaptive time domain filter to provide a filtered audio signal;

determining a noise estimate of the unknown noise component using the reference audio signal;

tailoring a minimum mean squared error analysis based on the noise estimate; and

subjecting each frequency band of the plurality of frequency bands of the filtered audio signal to the minimum mean squared error analysis;

assigning a confidence interval as a measure of statistical likelihood of dominance of the unknown target component in to each frequency band of the plurality of frequency bands of the filtered audio signal based on a result of the minimum mean squared analyses;

modifying an audio output level of each frequency band of the filtered audio signal based on the confidence interval of that frequency band to provide a modified output frequency band; and

combining the modified output frequency bands for each frequency band of the plurality of frequency bands of the filtered audio signal to provide a clarified output audio signal substantially reduced in the unknown noise component.

16. An electronic device configured to receive audio signals, comprising:

a primary audio channel for receiving a primary audio signal;

a reference audio channel for receiving a reference audio signal;

a processor programmed to;

receive the primary audio signal from the primary audio channel and the reference audio signal from the reference audio channel;

process the reference audio signal to provide a noise estimate of an unknown noise component;

generate a minimum mean squared error analysis that accounts for the noise estimate of the unknown noise component;

subject a plurality of frequency bands of the primary audio signal to the minimum mean squared error analysis;

compare a result of the minimum mean squared analysis of each frequency band of the plurality of frequency bands of the primary audio signal to a result of the minimum mean squared analysis of a corresponding frequency band of the plurality of frequency bands of the reference audio signal to provide a frequency band comparison;

assign a confidence interval as a measure of statistical likelihood of dominance of an unknown target component relative to the unknown noise component for each frequency band of the plurality of frequency bands of the primary audio signal based on the frequency band comparison that corresponds to that frequency band;

adjust an output power of the frequency band based on the confidence interval to provide a modified output frequency band; and

combine the modified output frequency bands for each frequency band of the plurality of frequency bands of the primary audio signal to provide a clarified output audio signal substantially reduced in the unknown noise component; and

cause an output element to output the clarified output audio signal; and

wherein the output element is in communication with the processor.

17. The electronic device of claim 16, wherein the processor is further programmed to:

subject a plurality of frequency bands of the reference audio signal to the minimum mean squared error analysis, frequency ranges of the plurality of frequency bands of the primary audio signal and of the plurality of frequency bands of the reference audio signal corresponding to one another.

18. The electronic device of claim 16, comprising a mobile telephone.

19. The electronic device of claim 16, wherein the output element comprises a speaker.

20. The electronic device of claim 16, wherein the speaker is carried by the electronic device.

21. The electronic device of claim 16, wherein the speaker is configured to selectively couple to the electronic device.

22. The electronic device of claim 16, wherein the processor is further programmed to:

apply an adaptive time domain filter to the primary audio signal and to the reference audio signal.

23. The electronic device of claim 22, wherein the processor is programmed to:

apply an adaptive least mean square filter to the primary audio signal and to the reference audio signal.

24. The electronic device of claim 22, wherein the processor is programmed to:

apply the adaptive time domain filter to the primary audio signal and to the reference audio signal before subjecting the plurality of frequency bands of the primary audio signal and the plurality of frequency bands of the reference audio signal to the minimum mean squared error analyses.