Apparatus and method to track positions of multiple sound sources

• US 8,773,952 B2
• Filed: 10/26/2010
• Issued: 07/08/2014
• Est. Priority Date: 10/30/2009
• Status: Active Grant

• Current Assignee: Samsung Electronics Co. Ltd.
• Original Assignee: Samsung Electronics Co. Ltd.
• Inventors: Shin, Ki Hoon
• Primary Examiner(s): Alsomiri, Isam
• Assistant Examiner(s): HULKA, JAMES R
• Application Number:
  US12/911,836
• Publication Number:
  US 20110103191A1
• Time in Patent Office: 1,351 Days
• Field of Search: 367/125
• US Class Current: 367/125
• CPC Class Codes: G01S 3/8006 Multi-channel systems specially adapted for direction-finding, i.e. having a single aerial system capable of giving simultaneous indications of the directions of different signals
• CPC Class Codes: G01S 3/8083 determining direction of source

Abstract

Disclosed herein is an apparatus and method to track positions of multiple sound sources which can simultaneously track the positions of multiple sound sources. Sound signals input to an array of microphones are separated into independent sound source signals, correlation coefficients between the sound signals input to the microphones and the separated sound source signals are estimated, and the estimated correlation coefficients are analyzed to track the positions of the sound sources.

Claims

1. An apparatus to track positions of multiple sound sources, the apparatus comprising:

a microphone array to which sound signals of a plurality of sound sources are input;

a sound source separator to separate the sound signals input to the microphone array into independent sound source signals;

a target sound source selector to select a target sound source to be tracked by analyzing patterns of the separated sound source signals;

a correlation coefficient estimator to estimate correlation coefficients between the sound signals input to the microphone array and the separated sound source signals and to create a correlation coefficient matrix using the estimated correlation coefficients; and

a sound source position estimator to estimate a position of the selected target sound source using the estimated correlation coefficients,

wherein the sound source position estimator rearranges all correlation coefficients of the target sound source according to magnitudes of the correlation coefficients thereof.

2. The apparatus according to claim 1, wherein the sound source separator performs sound source separation using a blind source separation (BSS) method.

3. The apparatus according to claim 2, wherein the BSS method includes one of independent component analysis (ICA) and geometric source separation (GSS).

4. The apparatus according to claim 1, wherein the target sound source selector selects the target sound source by matching patterns of the separated sound source signals with a preset target sound source signal pattern.

5. The apparatus according to claim 4, wherein the target sound source selector generates an index for identifying the selected target sound source.

6. The apparatus according to claim 1, wherein the correlation coefficient estimator obtains correlation coefficients by performing cross-correlation between the sound signals input to the microphone array and the separated sound source signals.

7. The apparatus according to claim 1, wherein, if a plurality of target sound sources to be tracked is present, the sound source position estimator performs simultaneous position tracking on the plurality of target sound sources.

8. The apparatus according to claim 1, wherein the sound source position estimator further tracks a position of the target sound source using a global coordinate of the target sound source estimated using a preset number of correlation coefficients among the rearranged correlation coefficients.

9. The apparatus according to claim 1, wherein the microphone array includes a plurality of microphones arranged at substantially equal intervals in a circle and the sound source position estimator estimates a global coordinate ψ

_i of an ith target sound source using Expression 1;

ψ

_i=ω

_j+φ

_i, 

 

(Expression

     1)

where ω

_j is a global coordinate of a jth microphone corresponding to a largest correlation coefficient with respect to a reference position of the microphone array, φ

_i is a local coordinate of the ith target sound source with respect to the jth microphone, the global coordinate ψ

_i of the ith target sound source represents an angle of the ith target sound source relative to the reference position of the microphone array, the global coordinate ω

_j of the jth microphone represents an angle of the jth microphone corresponding to the largest correlation coefficient relative to the reference position of the microphone array, and the local coordinate φ

_i of the ith target sound source represents an angle of the ith target sound source relative to the jth microphone corresponding to the largest correlation coefficient.

10. The apparatus according to claim 9, wherein the sound source position estimator estimates the local coordinate of the ith target sound source using Expression 2:

φ

_i=sign(φ

_i)×

θ

/2×

(r_i2−

r_i3)/(r′

_i1−

r′

_i3), 0≦

φ

_i≦

θ

/2, 

 

(Expression

     2)

where sign(φ

_i) is the sign of the local coordinate φ

_i of the ith target sound source, r′

_i1 is a largest correlation coefficient among correlation coefficients of the ith target sound source rearranged according to magnitudes of the correlation coefficients, r′

_i2 is a second largest correlation coefficient among the correlation coefficients of the ith target sound source rearranged according to the magnitudes thereof, and r′

_i3 is a third largest correlation coefficient among the correlation coefficients of the ith target sound source rearranged according to the magnitudes thereof.

11. The apparatus according to claim 10, wherein the local coordinate φ

_i of the ith target sound source is positive if the ith target sound source is located at one side of a microphone corresponding to the largest correlation coefficient r′

_i1, the one side thereof being the same as where the global coordinate ψ

_i is located, and is negative if the ith target sound source is located at the other side of the microphone corresponding to the largest correlation coefficient r′

_i1, the other side thereof being opposite to where the global coordinate ψ

_i is located.

12. The apparatus according to claim 11, wherein an absolute value of the local coordinate φ

_i of the ith target sound source is defined as a linear function of the second largest correlation coefficient r′

_i2 according to relations of Expression 3;

r′

_i2=r′

_i3, if |φ

_i|=0

r′

_i2=r′

_i1, if |φ

_i|=θ

/2, 

 

(Expression

     3)

where θ

is an angle between each of the microphones arranged at substantially equal intervals.

13. The apparatus according to claim 1, wherein the microphone array includes a plurality of microphones arranged at substantially equal intervals in a line and the sound source position estimator estimates a global coordinate D_i of an ith target sound source using Expression 4:

D_i=m_j+l_j, 

 

(Expression

     4)

where m_j is a global coordinate of a jth microphone corresponding to a largest correlation coefficient with respect to a reference position of the microphone array, l_i is a local coordinate of an ith target sound source with respect to the jth microphone, D_i represents a distance of the ith target sound source from the reference position of the microphone array, the global coordinate m_j of the jth microphone represents a distance of the jth microphone corresponding to the largest correlation coefficient from the reference position of the microphone array, and the local coordinate l_i of the ith target sound source represents a distance of the ith target sound source from the jth microphone corresponding to the largest correlation coefficient.

14. The apparatus according to claim 13, wherein the sound source position estimator estimates the local coordinate of the ith target sound source using Expression 5:

l_i=sign(l_i)×

d/2×

(r′

_i2−

r′

_i3)/(r′

_i1−

r′

_i3), 0≦

l_i≦

d/2, 

 

(Expression

     5)

where sign(l_i) is the sign of the local coordinate of the ith target sound source, r′

_i1 is a largest correlation coefficient among correlation coefficients of the ith target sound source rearranged according to magnitudes of the correlation coefficients, r′

_i2 is a second largest correlation coefficient among the correlation coefficients of the ith target sound source rearranged according to the magnitudes thereof, and r′

_i3 is a third largest correlation coefficient among the correlation coefficients of the ith target sound source rearranged according to the magnitudes thereof.

15. The apparatus according to claim 14, wherein the local coordinate l_i of the ith target sound source is positive if the ith target sound source is located at one side of a microphone corresponding to the largest correlation coefficient r′

_i1, the one side thereof being the same as where the global coordinate D_i is located, and is negative if the ith target sound source is located at the other side of the microphone corresponding to the largest correlation coefficient r′

_i1, the other side thereof being opposite to where the global coordinate D_i is located.

16. The apparatus according to claim 13, wherein, if a target sound source is located at an end of the microphone array, the sound source position estimator roughly estimates a position of the target sound source, determining that the target sound source is located at the end of the microphone array.

17. A method to track positions of multiple sound sources, the method comprising:

separating sound signals input to a microphone array into independent sound source signals;

selecting a target sound source to be tracked by analyzing patterns of the separated sound source signals;

estimating correlation coefficients between the sound signals input to the microphone array and the separated sound source signals and creating a correlation coefficient matrix using the estimated correlation coefficients; and

estimating a position of the selected target sound source using the estimated correlation coefficients,

wherein the estimation of the position of the target sound source includes rearranging all correlation coefficients of the target sound source according to magnitudes of the correlation coefficients thereof.

18. The method according to claim 17, wherein the separating sound source signals uses a blind source separation (BSS) method.

19. The method according to claim 18, wherein the BSS method is performed using one of independent component analysis (ICA) and geometric source separation (GSS).

20. The method according to claim 17, wherein the target sound source selection is performed by matching patterns of the separated sound source signals with a preset target sound source signal pattern.

21. The method according to claim 17, wherein the estimation of the correlation coefficients includes obtaining correlation coefficients by performing cross-correlation between the sound signals input to the microphone array and the separated sound source signals.

22. The method according to claim 17, wherein the estimation of the position of the target sound source further includes tracking a position of the target sound source using a global coordinate of the target sound source estimated using a preset number of correlation coefficients among the rearranged correlation coefficients.

23. The method according to claim 17, wherein, when the microphone array includes a plurality of microphones arranged at substantially equal intervals in a circle, a global coordinate ψ

_i of an ith target sound source and a local coordinate φ

_i of the ith target sound source are estimated using Expression 1 and Expression 10;

ψ

_i=ω

_j+φ

_i, 

 

(Expression

     1)

where ω

_j is a global coordinate of a jth microphone corresponding to a largest correlation coefficient with respect to a reference position of the microphone array, φ

_i is a local coordinate of an ith target sound source with respect to the jth microphone, ψ

_i represents an angle of the ith target sound source relative to the reference position of the microphone array, the global coordinate ω

_j of the jth microphone represents an angle of the jth microphone corresponding to the largest correlation coefficient relative to the reference position of the microphone array, and the local coordinate φ

_i of the ith target sound source represents an angle of the ith target sound source relative to the jth microphone corresponding to the largest correlation coefficient, and

φ

_i=sign(φ

_i)×

θ

/2×

(r′

_i2−

r′

_i3)/(r′

_i1−

r′

_i3), 0≦

φ

_i≦

θ

/2, 

 

(Expression

     2)

where sign(φ

_i) is the sign of the local coordinate φ

_i of the ith target sound source, r′

_i1 is a largest correlation coefficient among correlation coefficients of the ith target sound source rearranged according to magnitudes of the correlation coefficients, r′

_i2 is a second largest correlation coefficient among the correlation coefficients of the ith target sound source rearranged according to the magnitudes thereof, and r′

_i3 is a third largest correlation coefficient among the correlation coefficients of the ith target sound source rearranged according to the magnitudes thereof.

24. The method according to claim 17, wherein, when the microphone array includes a plurality of microphones arranged at substantially equal intervals in a line, a global coordinate D_i of an ith target sound source and a local coordinate l_i of the ith target sound source are estimated using Expression 4 and Expression 5:

D_i=m_j+l_i, 

 

(Expression

     11)

where m_j is a global coordinate of a jth microphone corresponding to a largest correlation coefficient with respect to a reference position of the microphone array, l_i is a local coordinate of an ith target sound source with respect to the jth microphone, D_i represents a distance of the ith target sound source from the reference position of the microphone array, the global coordinate m_j of the jth microphone represents a distance of the jth microphone corresponding to the largest correlation coefficient from the reference position of the microphone array, and the local coordinate l_i of the ith target sound source represents a distance of the ith target sound source from the jth microphone corresponding to the largest correlation coefficient, and

l_i=sign(l_j)×

d/2×

(r′

_i2−

r′

_i3)/(r′

_i1−

r′

_i3), 0≦

l_i≦

d/2, 

 

(Expression

     5)

where sign(l_i) is the sign of the local coordinate l_i of the ith target sound source, r′

_i1 is a largest correlation coefficient among correlation coefficients of the ith target sound source rearranged according to magnitudes of the correlation coefficients, r′

_i2 is a second largest correlation coefficient among the correlation coefficients of the ith target sound source rearranged according to the magnitudes thereof, and r′

_i3 is a third largest correlation coefficient among the correlation coefficients of the ith target sound source rearranged according to the magnitudes thereof.

25. A method to track positions of multiple sound sources, the method comprising:

separating sound signals input to a microphone array into independent sound source signals;

selecting a target sound source to be tracked by analyzing patterns of the separated sound source signals; and

estimating a position of the selected target sound source using three largest magnitudes of correlation coefficients of a cross correlation between microphones of the array and the target sound source,

wherein the estimation of the position of the target sound source includes rearranging all correlation coefficients of the target sound source according to magnitudes of the correlation coefficients thereof.