Methods and apparatus for source location estimation from microphone-array time-delay estimates

• US 5,737,431 A
• Filed: 03/07/1995
• Issued: 04/07/1998
• Est. Priority Date: 03/07/1995
• Status: Expired due to Term

• Current Assignee: Brown University
• Original Assignee: Brown University
• Inventors: Silverman, Harvey F., Adcock, John E., Brandstein, Michael S.
• Primary Examiner(s): Kuntz, Curtis
• Assistant Examiner(s): NGUYEN, DUC MINH
• Application Number:
  US08/399,427
• Time in Patent Office: 1,127 Days
• Field of Search: 381/91-92, 381/94, 381/122, 381/26
• US Class Current: 381/92
• CPC Class Codes: G01V 1/001 Acoustic presence detection
• CPC Class Codes: H04N 7/15 Conference systems

Abstract

Apparatus for determining the location of a signal-generating source (e.g., a conferee in a telephone conference) includes at least three sensors (e.g., microphones) arranged in a plurality of sets, each having two or more sensors. A surface-finding element responds to receipt at each sensor set of signals (e.g., speech) from the source for identifying a geometric surface (e.g., the surface of a hyperboloid or cone) representing potential locations of the source as a function of sensor locations and time difference of arrival of the signals. A location-approximating element coupled to two or more of the sets identifies a line that further defines potential source locations at the intersection of the surfaces. A location signal representing those potential locations is generated in accord with parameters of that line. Further functionality generates generating the location signal as a function of closest intersections the plural ones of the aforementioned lines.

Claims

1. Apparatus for determining a location of a source that generates source signals, comprising

A. at least three sensors arranged in a plurality of sets, each set including two or more of sensors,

B. surface-finding means, coupled with said sensors, for responding to receipt of a source signal by the sensors of each set for identifying a geometric surface representing a locus of potential source locations, said geometric surface being defined as a function of locations of the sensors in that set and a time difference of arrival of the source signal at those sensors, and

C. location-approximating means for generating a location signal, said location-approximating means being coupled to said surface-finding means and including

i) identification means for identifying a line representing a further constrained locus of potential source locations, said line being defined as a function of an intersection of surfaces identified by two or more of said sensor sets,

ii) intersection-finding means for generating said location signal representing a location of said source as a function of points of closest intersection of a plurality of said lines.

2. Apparatus according to claim 1, wherein

A. each sensor set includes two sensors, and

B. at least one pair of sensor sets are substantially centered and orthogonal to one another, each such pair of substantially centered and orthogonal sets being arranged such that

(i) a linear segment defined by sensor locations of one of the sets in the pair is substantially orthogonal to a linear segment defined by sensor locations of the other of the sets in the pair,

(ii) centerpoints of the linear segments defined by sensor locations of each of the sets are substantially aligned with one another.

3. Apparatus according to claim 2, wherein said surface-finding means includes means for identifying, in connection with receipt of a source signal by each sensor set, a substantially conical surface representing a locus of potential source locations, said substantially conical surface having a direction angle θ

, with respect to a linear segment between m₁ and m₂, in accord with the expression ##EQU20## where c is a speed of propagation of said source signal,

τ

is a measured time difference of arrival of said source signal at the sensors of the sensor set,

m₁ is a location of a first sensor in the sensor set, and

m₂ is a location of a second sensor in the sensor set.

4. Apparatus according to claim 3, wherein said location-approximating means includes means for identifying a line defined as a function of an intersection of said substantially conical surfaces, said line having direction angles α

, β and

γ

from said substantially aligned centerpoints, said direction angles being defined in accord with the expression

space="preserve" listing-type="equation">cos .sup.2 α

+cos .sup.2 β

+cos .sup.2 γ

=1

where

α

is a first direction angle of said line from said substantially aligned centerpoints and is a direction angle of a conical surface for a first sensor set,

β

is a second direction angle of said line from said substantially aligned centerpoints and is a direction angle of a conical surface for a second sensor set that is substantially centered and orthogonal to said first sensor set,

γ

is a third direction angle of said line from said substantially aligned centerpoints.

5. Apparatus according to claim 1, wherein said intersection-finding means includes means for identifying points of closest intersection, s_jk and s_kj, of two such lines, l_j and l_k, in accord with the expression

6. Apparatus according to claim 1, wherein said intersection-finding means further includes weighting means for generating said location signal as a weighted function of said closest points of closest intersection of a plurality of said lines.

7. Apparatus according to claim 1, wherein said intersection-finding means includes weighting means for generating said location signal as a function of a point s defined in accord with the expression ##EQU23## where W_jk is a weight associated with each point of closest intersection in accord with the expression ##EQU24## M is a number of pairs of sensor sets, m₁₁ is the location of the first sensor in the first sensor set of the lth sensor set pair,

m₁₂ is the location of the second sensor in the first sensor set of the lth sensor set pair,

m₁₃ is the location of the first sensor in the second sensor set of the lth sensor set pair,

m₁₄ is the location of the second sensor in the second sensor set of the lth sensor set pair,

τ

₁₁₂ is the measured time difference of arrival of said source signal at the sensors of a first sensor set of the lth pair of sensor sets,

τ

₁₃₄ is the measured time difference of arrival of said source signal at the sensors of a second sensor set of the lth pair of sensor sets,

σ

₁₂ is a variance of a measured time difference of arrival of said source signal at the sensors of a first sensor set of the lth pair of sensor sets,

σ

₃₄ is a variance of a measured time difference of arrival of said source signal at the sensors of a second sensor set of the lth pair of sensor sets,

P(x,m,σ

²) is the value a Gaussian distribution with mean m and variance σ

² in accord with the expression ##EQU25## T({m₁₁, m₁₂ },s) is a true time-difference of arrival of said source signal at the sensors of an lth sensor set pair in accord with the expression ##EQU26## m₁ and m₂ are locations of the first and second sensors in a sensor set.

8. Apparatus for steering a camera for use in a video teleconferencing system, said apparatus comprising

A. at least three microphones arranged in a plurality of sets, each set including two or more of microphones,

B. surface-finding means, coupled with said microphones, for responding to receipt of a source signal by the microphones of each set for identifying a geometric surface representing a locus of potential source locations, said geometric surface being defined as a function of locations of the microphones in that set and a time difference of arrival of the source signal at those microphones,

C. location-approximating means for generating a location signal, said location-approximating means being coupled to said surface-finding means and including

i) identification means for identifying a line representing a further constrained locus of potential source locations, said line being defined as a function of an intersection of surfaces identified by two or more of said sensor sets,

ii) intersection-finding means for generating said location signal representing a location of said source as a function of points of closest intersection of a plurality of said lines, and

D. means, coupled with said camera, for steering it relative to said source as a function of said location signal.

12. Apparatus according to claim 8, wherein said intersection-finding means includes means for identifying points of closest intersection, s_jk and s_kj, of two such lines, l_j and l_k, in accord with the expression

13. Apparatus according to claim 8, wherein said intersection-finding means includes means for generating said location signal as a weighted function of said closest points of closest intersection of a plurality of said lines.

14. Apparatus according to claim 13, wherein said intersection-finding means include means for generating said location signal as a function of a point s defined in accord with the expression ##EQU30## where W_jk is a weight associated with each potential source location in accord with the expression ##EQU31## M is a number of pairs of microphone sets, m₁₁ is the location of the first microphone in the first microphone set of the lth microphone set pair,

m₁₂ is the location of the second microphone in the first microphone set of the lth microphone set pair,

m₁₃ is the location of the first microphone in the second microphone set of the lth microphone set pair,

m₁₄ is the location of the second microphone in the second microphone set of the lth microphone set pair,

τ

₁₁₂ is the measured time difference of arrival of said source signal at the microphones of a first microphone set of the lth pair of microphone sets,

τ

₁₃₄ is the measured time difference of arrival of said source signal at the microphones of a second microphone set of the lth pair of microphone sets,

σ

₁₂ is a variance of a measured time difference of arrival of said source signal at the microphones of a first microphone set of the lth pair of microphone sets,

σ

₃₄ is a variance of a measured time difference of arrival of said source signal at the microphones of a second microphone set of the lth pair of microphone sets,

P(x,m,σ

²) is the value a Gaussian distribution with mean m and variance σ

² in accord with the expression ##EQU32## T({m₁₁, m₁₂ },s) is a true time-difference of arrival of said source signal at the microphones of an lth microphone set pair in accord with the expression ##EQU33## m₁ and m₂ are locations of the first and second microphones in a microphone set.

9. Apparatus according to claim 9, wherein

A. each microphone set includes two microphones, and

B. at least one pair of microphone sets are substantially centered and orthogonal to one another, each such pair of substantially centered and orthogonal sets being arranged such that

(i) a linear segment defined by microphone locations of one of the sets in the pair is substantially orthogonal to a linear segment defined by microphone locations of the other of the sets in the pair,

(ii) centerpoints of the linear segments defined by microphone locations of each of the sets are substantially aligned with one another.

10. Apparatus according to claim 9, wherein said surface-finding means includes means for identifying, in connection with receipt of a source signal by each microphone set, a substantially conical surface representing a locus of potential source locations, said substantially conical surface having a direction angle θ

, with respect to a linear segment defined by m₁ and m₂, in accord with the expression ##EQU27## where c is a speed of propagation of said source signal,

τ

is a measured time difference of arrival of said source signal at the microphones of the microphone set,

m₁ is a location of a first microphone in the microphone set, and

m₂ is a location of a second microphone in the microphone set.

11. Apparatus according to claim 10, wherein said location-approximating means includes means for identifying a line defined as a function of an intersection of said substantially conical surfaces, said line having direction angles α

, β and

γ

from said substantially aligned centerpoints, said direction angles being defined in accord with the expression

space="preserve" listing-type="equation">cos .sup.2 α

+cos .sup.2 β

+cos .sup.2 γ

=1

where

α

is a first direction angle of said line from said substantially aligned centerpoints and is a direction angle of a conical surface for a first microphone set,

β

is a second direction angle of said line from said substantially aligned centerpoints and is a direction angle of a conical surface for a second microphone set that is substantially centered and orthogonal to said first microphone set,

γ

is a third direction angle of said line from said substantially aligned centerpoints.

15. Method for determining a location of a source that generates source signals, comprising

A. an arranging step for arranging at least three sensors in a plurality of sets, each set including two or more of sensors,

B. a surface-finding step for responding to receipt of a source signal by the sensors of each set for identifying a geometric surface representing a locus of potential source locations, said geometric surface being defined as a function of locations of the sensors in that set and a time difference of arrival of the source signal at those sensors,

C. a location-approximating step for

i) identifying a line representing a further constrained locus of potential source locations, said line being defined as a function of an intersection of surfaces identified by two or more of said sensor sets,

ii) intersection-finding step for generating a location signal representing a location of said source as a function of points of closest intersection of a plurality of said identified lines.

16. Method according to claim 15, wherein said arranging step includes

A. providing each sensor set with two sensors, and

B. arranging at least one pair of sensor sets to be substantially centered and orthogonal to one another, each such pair of substantially centered and orthogonal sets being arranged such that

(i) a linear segment defined by sensor locations of one of the sets in the pair is substantially orthogonal to a linear segment defined by sensor locations of the other of the sets in the pair,

(ii) centerpoints of the linear segments defined by sensor locations of each of the sets are substantially aligned with one another.

17. Method according to claim 16, wherein said surface-finding step includes the step of identifying, in connection with receipt of a source signal by each sensor set, a substantially conical surface representing a locus of potential source locations, said substantially conical surface having a direction angle θ

, with respect to a linear segment defined by m₁ and m₂, in accord with the expression ##EQU34## where c is a speed of propagation of said source signal,

τ

is a measured time difference of arrival of said source signal at the sensors of the sensor set,

m₁ is a location of a first sensor in the sensor set, and

m₂ is a location of a second sensor in the sensor set.

18. Method according to claim 17, wherein said location-approximating step includes the step of identifying a line defined as a function of an intersection of said substantially conical surfaces, said line having direction angles α

, β and

γ

from said substantially aligned centerpoints, said direction angles being defined in accord with the expression

space="preserve" listing-type="equation">cos .sup.2 α

+cos .sup.2 β

+cos .sup.2 γ

=1

where

α

is a first direction angle of said line from said substantially aligned centerpoints and is a direction angle of a conical surface for a first sensor set,

β

is a second direction angle of said line from said substantially aligned centerpoints and is a direction angle of a conical surface for a second sensor set that is substantially centered and orthogonal to said first sensor set,

γ

is a third direction angle of said line from said substantially aligned centerpoints.

19. Method according to claim 15, wherein said intersection-finding step includes the step of identifying points of closest intersection, s_jk and s_kj, of two such lines, l_j and l_k, in accord with the expression

20. Method according to claim 19, wherein said intersection-finding step includes the step of generating said location signal as a weighted function of said closest points of closest intersection of a plurality of said lines.

21. Method according to claim 20, wherein said intersection-finding step includes the step of generating said location signal as a function of a point s defined in accord with the expression ##EQU37## where W_jk is a weight associated with each point of closest intersection in accord with the expression ##EQU38## M is a number of pairs of sensor sets, m₁₁ is the location of the first sensor in the first sensor set of the lth sensor set pair,

m₁₂ is the location of the second sensor in the first sensor set of the lth sensor set pair,

m₁₃ is the location of the first sensor in the second sensor set of the lth sensor set pair,

m₁₄ is the location of the second sensor in the second sensor set of the lth sensor set pair,

τ

₁₁₂ is the measured time difference of arrival of said source signal at the sensors of a first sensor set of the/th pair of sensor sets,

τ

₁₃₄ is the measured time difference of arrival of said source signal at the sensors of a second sensor set of the lth pair of sensor sets,

σ

₁₂² is a variance of a measured time difference of arrival of said source signal at the sensors of a first sensor set of the lth pair of sensor sets,

σ

₃₄² is a variance of a measured time difference of arrival of said source signal at the sensors of a second sensor set of the lth pair of sensor sets,

P(x,m,σ

²) is the value a Gaussian distribution with mean m and variance σ

² in accord with the expression ##EQU39## T({m₁₁, m₁₂ },s) is a true time-difference of arrival of a source signal originating a position s at the sensors of an lth sensor set pair in accord with the expression ##EQU40## m₁ and m₂ are locations of the first and second sensors in a sensor set.