Method and system for interference suppression using blind source separation

Patent metadata

Field	Value
Publication number	US-9741358-B2
Application number	US-201414300690-A
Country	US
Kind code	B2
Filing date	Jun 10, 2014
Priority date	Sep 11, 2009
Publication date	Aug 22, 2017
Grant date	Aug 22, 2017

Abstract

A method of interference suppression is provided that includes receiving a first audio signal from a first audio capture device and a second audio signal from a second audio capture device wherein the first audio signal includes a first combination of desired audio content and interference and the second audio signal includes a second combination of the desired audio content and the interference, performing blind source separation using the first audio signal and the second audio signal to generate an output interference signal and an output audio signal including the desired audio content with the interference suppressed, estimating interference remaining in the output audio signal using the output interference signal, and subtracting the estimated interference from the output audio signal to generate a final output audio signal with the interference further suppressed.

First claim

What is claimed is:

1. A digital system comprising:
a primary microphone configured to capture a primary audio signal comprising a first combination of desired audio content and interference;
a secondary microphone configured to capture a secondary audio signal comprising a second combination of the desired audio content and the interference; and
an audio encoder operatively connected to the primary microphone and the secondary microphone to receive the primary audio signal and the secondary audio signal, wherein the audio encoder is configured to generate a final output audio signal comprising the desired audio content with the interference suppressed by:
performing blind source separation using the primary audio signal and the secondary audio signal to generate an output interference signal and an output audio signal comprising the desired audio content with the interference suppressed, wherein the blind source separation uses a tangent hyperbolic function to separate the desired audio content and the interference;
estimating interference remaining in the output audio signal using the output interference signal; and
subtracting the estimated interference from the output audio signal to generate the final output audio signal.

2. The digital system of claim 1 , wherein the digital system is a cellular telephone and the desired audio content is speech.

3. The digital system of claim 2 , wherein the primary microphone is located on a lower front portion of the cellular telephone and the secondary microphone is located on an upper back portion of the cellular telephone.

4. The digital system of claim 1 , wherein subtracting the estimated interference comprises:
estimating a magnitude spectrum of a block of the final output audio signal based on the estimated interference;
estimating a frequency spectrum of the block of the final output audio signal using the estimated magnitude spectrum and a phase of a frequency spectrum of a corresponding block of the output audio signal; and
computing an inverse Fourier transform of the estimated frequency spectrum to generate the block of the final output audio signal.

5. The digital system of claim 1 , wherein the desired audio content is speech.

6. The digital system of claim 1 , wherein estimating the interference includes applying a transform vector to the output interference signal that maps a magnitude spectrum of the interference in the output interference signal to a magnitude spectrum of the interference in the output audio signal.

7. The digital system of claim 1 , wherein estimating the interference includes:
detecting a silence interval of a predetermined length in the output audio signal;
computing a transform vector as a product of a matrix comprising a magnitude spectrum of the silence interval and a pseudo inverse of a matrix comprising a magnitude spectrum of a corresponding interval in the output interference signal; and
applying the transform vector to a block of the output interference signal.

8. The digital system of claim 1 , wherein the tangent hyperbolic function is approximated using a lookup table and interpolation between values in the lookup table.

9. A device comprising one or more processors configured to:
perform blind source separation using a primary audio signal captured by a primary microphone and a secondary audio signal captured by a secondary microphone to generate an output interference signal and an output audio signal including desired audio content with interference suppressed, wherein the blind source separation uses a tangent hyperbolic function to separate the desired audio content and the interference;
estimate interference remaining in the output audio signal using the output interference signal; and
subtract the estimated interference from the output audio signal to generate the final output audio signal.

10. The device of claim 9 , wherein the device is a cellular telephone and the desired audio content is speech.

11. The device of claim 10 , wherein the cellular telephone includes the primary microphone and the secondary microphone, wherein the primary microphone is located on a lower front portion of the cellular telephone, and the secondary microphone is located on an upper back portion of the cellular telephone.

12. The device of claim 9 , wherein the one or more processors are further configured to:
estimate a magnitude spectrum of a block of the final output audio signal based on the estimated interference;
estimate a frequency spectrum of the block of the final output audio signal using the estimated magnitude spectrum and a phase of a frequency spectrum of a corresponding block of the output audio signal; and
compute an inverse Fourier transform of the estimated frequency spectrum to generate the block of the final output audio signal.

13. The device of claim 9 , wherein the desired audio content is speech.

14. The device of claim 9 , wherein the one or more processors are further configured to apply a transform vector to the output interference signal that maps a magnitude spectrum of the interference in the output interference signal to a magnitude spectrum of the interference in the output audio signal.

15. The device of claim 9 , wherein the one or more processors are further configured to:
detect a silence interval of a predetermined length in the output audio signal;
compute a transform vector as a product of a matrix comprising a magnitude spectrum of the silence interval and a pseudo inverse of a matrix comprising a magnitude spectrum of a corresponding interval in the output interference signal; and
apply the transform vector to a block of the output interference signal.

16. The device of claim 9 , wherein the tangent hyperbolic function is approximated using a lookup table and interpolation between values in the lookup table.

Assignees

• Texas Instruments Inc

Inventors

• Parikh Devangi N
• Ikram Muhammad Zubair

Classifications

• H04M1/60

  including speech amplifiers · CPC title

• G10L21/0272 Primary

  Voice signal separating · CPC title

• G10L21/0208 Primary

  Noise filtering · CPC title