Position estimation method, position estimation device, and program

Abstract

Provided is a technology that enables estimation of the position of a sound source even when not all acoustic signals collected by microphones are synchronized. In the present invention: mi (i=1, ..., M, where M is an integer of 2 or greater) is coordinates representing the position of an i-th sound source direction detector, sj (j=1, ..., N, where N is an integer of 2 or greater) is coordinates representing the position of a j-th sound source, and θj=[θ1j, θ2j, ..., θMj]T (j=1, ..., N, where θij is an angle representing the direction of the j-th sound source detected by the i-th sound source direction detector) is an angle vector representing the direction of the j-th sound source. Each of the sound source direction detectors detects the angle representing the direction of the sound source by using an acoustic signal, which is obtained by collecting sounds emitted by the sound source with two more more microphones and synchronizing the sounds. The invention includes: a sound source position updating unit that updates the values of the coordinates sj by using the angle vector θj and the coordinates mi; and a sound source direction detector position updating unit that updates the values of the coordinates mi by using the angle vector θj and the coordinates sj.

Description

[Technical field]

[0001] The present invention relates to a technique for estimating the position of a sound source using two or more microphones. [Background technology]

[0002] Currently, the number of ways to sense the environment is increasing due to the widespread use of various sensors such as smart speakers and sensors for the Internet of Things (IoT). In addition, the need for understanding the environment using information acquired by sensors is also expanding against the backdrop of the promotion of digital transformation (DX) in business. However, inexpensive, space-saving sensors that are currently in widespread use are unable to fully acquire desired information by themselves, so technology is required to integrate information acquired by multiple sensors and generate the desired information.

[0003] One example of environmental understanding using microphones as sensors is a method in which two or more microphones are placed in a space to estimate the position of a sound source. For example, in the method described in Non-Patent Document 1, the position of a sound source is estimated using the time difference between sounds arriving at the microphones. [Prior art documents] [Non-patent literature]

[0004] [Non-Patent Document 1] Kazunori Kobayashi, Kenichi Furuya, and Akitoshi Kataoka, "Blind Source Localization Using Multiple Microphones with Unknown Positions," IEICE Transactions on Information and Communication Engineers, Vol. 86, No. 6, pp. 619-627, 2003. Summary of the Invention [Problem to be solved by the invention]

[0005] However, the method of Non-Patent Document 1 requires that all acoustic signals picked up by the microphones be synchronized. In other words, since processing is performed on the premise that all acoustic signals picked up by microphones placed in a space are synchronized, it is not very user-friendly.

[0006] Therefore, an object of the present invention is to provide a technique that can estimate the position of a sound source even if all of the acoustic signals picked up by the microphones are not synchronized. [Means for solving the problem]

[0007] One aspect of the present invention is i (i=1, …, M, where M is an integer equal to or greater than 2) is the coordinate representing the position of the i-th sound source direction detector, s j (j=1, …, N, where N is an integer equal to or greater than 2) is the coordinate representing the position of the jth sound source, and θ j =[θ 1j , θ 2j , …, θ Mj ] T (j=1, …, N, where θ ij is an angle vector representing the direction of the j-th sound source, and each sound source direction detector detects an angle representing the direction of the sound source by using a synchronized acoustic signal obtained by collecting a sound emitted by the sound source with two or more microphones. The position estimation device detects the angle vector θ j (j=1, …, N) and coordinate m i (i=1, …, M), the coordinate s j A sound source position update step in which the value of (j=1, ..., N) is updated, and the position estimation device updates the angle vector θ j (j=1, …, N) and coordinate s j (j=1, …, N), coordinate m i and updating the sound source direction detector position values ​​(i=1, ..., M). Effect of the Invention

[0008] According to the present invention, it is possible to estimate the position of a sound source even if all of the acoustic signals picked up by the microphones are not synchronized. [Brief description of the drawings]

[0009] [Figure 1] FIG. 1 is a diagram showing an example of the placement of a sound source direction detector and a sound source. [Diagram 2] FIG. 2 is a diagram showing the positional relationship between the i-th sound source direction detector and the j-th sound source. [Diagram 3] FIG. 13 is a diagram showing an image of log-likelihood LLij(φij|θij). [Figure 4] FIG. 2 is a block diagram showing the configuration of a position estimation device 100 / 200. [Diagram 5] 4 is a flowchart showing the operation of the position estimation device 100 / 200. [Figure 6] FIG. 2 is a block diagram showing the configuration of a sound source direction detection unit 110. [Figure 7] 4 is a flowchart showing the operation of the sound source direction detecting unit 110. [Figure 8] FIG. 2 is a block diagram showing a configuration of a position estimation unit 120. [Figure 9] 13 is a flowchart showing the operation of the position estimation unit 120. [Figure 10] FIG. 2 is a diagram illustrating an example of a functional configuration of a computer that realizes each device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] Hereinafter, an embodiment of the present invention will be described in detail. Note that components having the same functions are given the same reference numbers and duplicated explanations will be omitted.

[0011] Before describing each embodiment, the notation used in this specification will be described.

[0012] ^ (caret) represents a superscript. For example, x y^z Yes z is a superscript to x, and xy^z Yes z is a subscript to x. Also, _ (underscore) represents a subscript. For example, x y_z Yes z is a superscript to x, and x y_z Yes z is a subscript on x.

[0013] In addition, the superscripts "^" and "~" such as ^x and ~x for a certain letter x should actually be written directly above the "x", but due to the constraints of the description in the specification, they are written as ^x and ~x. <Technical background> In an embodiment of the present invention, the position of a sound source and the position of the sound source direction detector are estimated using two or more sound source direction detectors. Here, the sound source direction detector detects an angle representing the direction of a sound source by using a synchronized acoustic signal obtained by collecting a sound emitted by the sound source with two or more microphones.

[0014] In an embodiment of the present invention, a situation is assumed in which M sound source direction detectors and N sound sources are installed in the same space (preferably, a two-dimensional plane) (where M and N are integers of 2 or more). Fig. 1 is a diagram showing an example of the installation of sound source direction detectors and sound sources. Fig. 1 shows how six sound source direction detectors and four sound sources are installed in the same space, and how the six sound source direction detectors detect a sound emitted by one sound source.

[0015] Each of the M sound source direction detectors includes two or more microphones, but does not need to include the same number of microphones. Also, while the acoustic signals picked up by all the microphones included in one sound source direction detector are processed in a synchronized manner, the acoustic signals picked up by all the microphones included in the M sound source direction detectors do not need to be processed in a synchronized manner. <<1: Sound source direction detector>> Here, a method for the sound source direction detector to detect an angle representing the direction of a sound source will be described. [Method 1] As described above, the sound source direction detector detects an angle representing the direction of a sound source by using a synchronized acoustic signal obtained by collecting a sound emitted by the sound source with two or more microphones. For example, the method described in Reference Non-Patent Document 1 can be used to detect the angle. (Reference Non-Patent Document 1: CH Knapp and GC Carter, “The generalized correlation method for estimation of time delay,” IEEE Trans. Acoust., Speech & Signal Process., ASSP-24, 4, pp.320-327, 1976.) If it is desired to improve the reliability of the detection, for example, sound pressure can be used. Specifically, when the input level to any one of the microphones included in the sound source direction detector is smaller than a predetermined threshold, the sound source direction detector can output a label indicating "no detection" instead of an angle.

[0016] In addition, the angle detected using the method of Reference Non-Patent Document 1 is corrected using an angle representing a reference direction before being output. For example, when the first microphone included in each sound source direction detector is at an angle θ r If the sound source direction detector detects an angle θ d is corrected by the following formula to obtain the corrected angle θ.

[0017]

number

[0018] Angle θ representing the reference direction r indicates how much the sound source direction detector is rotated on the xy plane, and is necessary because all M sound source direction detectors are not necessarily rotated to the same degree. In order to detect the angle representing the reference orientation, the sound source direction detector includes, for example, a direction sensor.

[0019] The value of the angle θ is increased or decreased by 2πn (n is an integer) as necessary so that the value satisfies 0≦θ<2π.

[0020] In summary, the sound source direction detector of this example detects the direction of a sound source by picking up a sound emitted by the sound source with two or more microphones and estimating an angle θ d , and an angle θ r Using the angle θ d to get the angle θ. [Method 2] In method 1, the sound source direction detector uses a direction sensor to obtain an angle indicating a reference direction that indicates how much the detector has rotated on the xy plane. In method 2, instead of using a direction sensor, a transmitter that emits a specific acoustic signal is used. Here, a specific acoustic signal is an acoustic signal that can be sufficiently distinguished from normal everyday sounds, such as a 1 kHz sine wave. Whether or not a signal is a specific acoustic signal can be determined by, for example, determining whether the spectrum S(f) of the sound detected by the sound source direction detector is

[0021]

number

[0022] It is advisable to judge whether the above condition is satisfied.

[0023] Method 2 will be described below. The first sound source direction detector is used as a reference for measuring the relative positional relationship of the sound source direction detectors. First, the mth (m ≠ 1) sound source direction detector emits a specific acoustic signal, the first sound source direction detector detects a sound based on the acoustic signal, and an angle ψ representing the direction of the mth (m ≠ 1) sound source direction detector as seen from the first sound source direction detector is calculated. 1m Next, the first sound source direction detector transmits a specific acoustic signal, the mth (m ≠ 1) sound source direction detector detects a sound based on the acoustic signal, and the mth (m ≠ 1) sound source direction detector detects an angle ψ m1Here, if the first sound source direction detector and the mth (m ≠ 1) sound source direction detector are on the reference direction, the following formula is satisfied.

[0024]

number

[0025] On the other hand, if the first sound source direction detector and the mth (m ≠ 1) sound source direction detector are not on the reference direction, the angle θ rm The difference will be just the same.

[0026]

number

[0027] This angle θ rm The angle θ detected by the sound source direction detector using d is corrected by the following formula to obtain the corrected angle θ.

[0028]

number

[0029] In summary, the m-th (m ≠ 1) sound source direction detector in this example detects the direction of a sound source by collecting a sound emitted by the sound source with two or more microphones and estimating an angle θ d The angle θ representing the deviation between the first sound source direction detector and the mth sound source direction detector (m ≠ 1) is obtained by using the transmitter of the first sound source direction detector and the transmitter of the mth sound source direction detector (m ≠ 1). rm Using the above, the angle θ d to get the angle θ. <<2: How to generate angle vectors>> Here, an angle vector generation method for generating angle vectors representing the directions of N sound sources will be described.

[0030] When the sound source direction detector detects a sound emitted by one of the N sound sources, it detects the angle representing the direction of the sound source and obtains and outputs the time when the sound was detected. Therefore, t i (i=1, …, M) is the time when the i-th sound source direction detector detects a sound emitted by one of the N sound sources, and θ i (t i )(i=1, …, M) is the sound source direction detector detected at time t i The i-th sound source direction detector detects a sound emitted by one of the N sound sources at time t i and time t i The angle θ represents the direction of the detected sound source. i (t i ) is output.

[0031] The angle vector generation method is i and angle θ i (t i ) (i=1, …, M) as input, and the angle vector θ1, …, θ N Here, θ j =[θ 1j , θ 2j , …, θ Mj ] T (j=1, …, N, where θ ij is an angle vector that represents the direction of the j-th sound source detected by the i-th sound source direction detector. The procedure for generating an angle vector is as follows. (1) Time t i and angle θ i (t i ) (i=1, …, M), the angle vector θ(t)=[θ1(t1), …, θ M (t M )] T Generate. (2) Using the angle vector θ(t), the angle vector θ representing the direction of the jth sound source j =θ(t). (3) Process (1) and process (2) are applied to the angle vectors θ1, θ2, …, θ for all N sound sources. N Repeat until is obtained.

[0032] The process (1) will be described. In the process (1), the angle θ i (t i ) (i=1, …, M) to the angle vector θ(t)=[θ1(t1), …, θ M (t M )] T where the angle θ i (t i ) (i=1, …, M) can be considered to have been detected at the same time, based on the assumption that the time of the M sound source direction detectors is synchronized with a time difference that is sufficiently shorter than the time at which the non-stationarity of the sound becomes a problem (for example, 100 ms) by using, for example, NTP. M The time is determined by whether it falls within the range from time t to time t+Δt (where Δt is a constant that represents the allowable deviation). For example, time t is t1, …, t M The following is an example. Here, M=3, and the sound source direction detector detects a sound emitted by one of the N sound sources at the following times: Times detected by the first sound source direction detector: 0:00:01.010, 0:00:01.042, … Times detected by the second sound source direction detector: 0:00:01.014, 0:00:01.046, … Times detected by the third sound source direction detector: 0:00:01.009, 0:00:01.041, … Here, the time 0:00:00.000 on the clocks of the three sound source direction detectors is slightly off due to NTP errors as an actual time. Therefore, when making a judgment based on the clock of the first sound source direction detector, the sound source direction detector whose detected time is the furthest from the first sound source direction detector is the second sound source direction detector, and the offset is 4 ms. When an NTP error of up to 5 ms is added to this offset, the actual time may be off by up to 9 ms. Therefore, if the allowable offset Δt is 20 ms, the time detected by the first sound source direction detector and the time detected by the second sound source direction detector can be considered to be the same. On the other hand, if the allowable offset Δt is 8 ms, the time detected by the first sound source direction detector and the time detected by the second sound source direction detector cannot be considered to be the same.

[0033] The process (2) will be described. In the process (2), an angle vector θ1, ..., an angle vector θ2, ..., an angle vector θ3, ..., an angle vector θ4, ... j-1 If it is determined using the angle vector θ(t) that the sound source that emitted the sound detected at time t does not match any of the 1st sound source, ..., j-1th sound source, the sound source that emitted the sound detected at time t is the jth sound source, and the angle vector θ j =θ(t). Whether the sound source that emitted the sound detected at time t is a new sound source or not is determined as follows. n Let (n=1, …, j-1) be the time when the sound emitted by the nth sound source is detected, and f n Let (t) be a function of the following equation:

[0034]

number

[0035] f for all n=1, …, j-1 nIf (t)=1, the sound source that emitted the sound detected at time t is determined to be a new sound source, otherwise, the sound source that emitted the sound detected at time t is determined to be not a new sound source. However, if any of the elements of θ(t) contains a label indicating "not detected," the above determination process is not performed. <<3: Location estimation method>> Here, a position estimation method for estimating the position of a sound source and the position of a sound source direction detector will be described.

[0036] The location estimation method is to use an angle vector θ j (j=1, …, N) is input, and the coordinate s j (j=1, …, N) and the coordinate m i (i=1, …, M), where m represents the position of the i-th sound source direction detector. i and the coordinate s representing the position of the jth sound source. j The initial value of θ is assumed to be given. FIG. 2 is a diagram showing the positional relationship between the i-th sound source direction detector and the j-th sound source. As described above, θ ij is the angle representing the direction of the j-th sound source detected by the i-th sound source direction detector. The procedure of the position estimation method is as follows. (1) Angle vector θ j (j=1, …, N) and coordinate m i (i=1, …, M), the coordinate s j Update the value of (j=1, …, N). (2) Angle vector θ j (j=1, …, N) and coordinate s j (j=1, …, N), coordinate m i Update the values ​​of (i=1, …, M). (3) Repeat the process (1) and the process (2) a predetermined number of times.

[0037] By repeating steps (1) and (2), the coordinate s j and coordinate m i Improve the estimation accuracy of.

[0038] In this position estimation method, φ ij Let be the angle that represents the direction of the jth sound source from the ith sound source direction detector, and let angle θ ij and angle φ ij The angle φ represents the deviation from ij -θ ij The probability distribution of P ij (θ ij |φ ij ) is used. This is explained in detail below.

[0039] The angle representing the direction of the jth sound source detected by the ith sound source direction detector is θ ij The probability that ij (θ ij ), then the angle vector representing the direction of the jth sound source is θ j The probability that j (θ j ) is expressed by the following formula:

[0040]

number

[0041] Probability P j (θ j ) can be regarded as a likelihood function with the direction of the sound source as seen by the sound source direction detector as a parameter. In this case, the likelihood function L j (φ j |θ j ) is given by the following formula:

[0042]

number

[0043] (However, φ j =[φ 1j , φ 2j , …, φ Mj ] T (j=1, …, N) Also, the log likelihood LL j (φ j |θ j ) is given by the following formula:

[0044]

number

[0045] Figure 3 shows the log-likelihood LL ij (φ ij |θ ij ) is a diagram showing an image.

[0046] In the following, P ij (θ ij |φ ij ) is a given probability distribution, and the coordinate s j and coordinate m i A method for estimating is described below. [Example 1] Probability distribution P ij (θ ij |φ ij ) is given as the normal distribution:

[0047]

number

[0048] (However, ij =||s j -m i ||, σ is a predetermined constant representing the standard deviation) The probability distribution P of equation (2) ij (θ ij |φ ij ) is the angle θ from the i-th sound source direction detector ij When a straight line is drawn, the angle φ representing the direction of the jth sound source from the ith sound source direction detector is ij This corresponds to assuming a normal distribution according to the degree to which it deviates from the straight line.

[0049] Angle φ ij and coordinate s j The relationship is expressed by the following formula:

[0050]

number

[0051]

number

[0052] (however,(·) x , (·) y represent the x- and y-coordinate components of the vector, respectively) Therefore, a ij = [-sinθ ij , cosθ ij ] T Then, equation (2) can be expressed as follows:

[0053]

number

[0054] In addition, the following equation, which is similar to equation (1), holds true:

[0055]

number

[0056] Therefore, the following equation is obtained from equations (5) and (6).

[0057]

number

[0058] Therefore, the coordinate s that maximizes equation (7) j Let us consider the angle vector θ j and coordinate m i Specifically, for example, the coordinate s that maximizes Eq. (7) can be found by using the steepest descent method or grid search. j It is a good idea to ask for

[0059] In addition, the coordinate s that maximizes equation (7) j can be derived analytically. First, we use the equation (7) for the coordinate sj Differentiating with respect to, we obtain the following equation:

[0060]

number

[0061] Where:

[0062]

number

[0063]

number

[0064] Then, equation (8) can be expressed as follows:

[0065]

number

[0066] therefore,

[0067]

number

[0068] The coordinate s satisfies j is obtained by the following formula:

[0069]

number

[0070] Also, ~θ i =[θ i1 , θ i2 , …, θ iN ] TLet (i=1, ..., M) be the angle vector representing the direction of the i-th sound source direction detector, and taking into account the symmetry between the sound source direction detector and the sound source, the following equation similar to equation (7) is obtained.

[0071]

number

[0072] Therefore, the coordinate m that maximizes equation (11) i Let us consider the angle vector θ j and coordinate s j Specifically, for example, the coordinate m that maximizes Equation (11) can be obtained by using the steepest descent method or grid search. i It is a good idea to ask for

[0073] Also, as before, the coordinate m that maximizes equation (11) is i can be derived analytically. i Differentiating with respect to, we obtain the following equation:

[0074]

number

[0075] however,

[0076]

number

[0077] It is.

[0078] therefore,

[0079]

number

[0080] The coordinate m satisfies i is obtained by the following formula:

[0081]

number

[0082] From the above, using equations (7) and (11), the coordinate s j (j=1, …, N), coordinate m i (i=1, …, M) should be updated repeatedly. Also, the coordinate s j (j=1, …, N), coordinate m i We can repeat the update for (i=1, …, M). However, if we simply repeat each update process, m i =0, s j = 0. So, the coordinate m i (i=1, …, M), the coordinates of two sound source direction detectors, specifically, the coordinates m of the first sound source direction detector and the second sound source direction detector. i For (i=1, 2), the following constraints are applied:

[0083]

number

[0084] (where c is the distance between the first and second sound source direction detectors, and ψ is the angle representing the direction of the second sound source direction detector as seen from the first sound source direction detector) In this way, the coordinate m i By constraining (i=1, 2), the desired solution can be obtained without the solution converging to the origin.

[0085] The angle ψ can be obtained by using the method 2 described in <<1: Sound source direction detector>>. Specifically, the angle ψ, which represents the direction of the second sound source direction detector as seen from the first sound source direction detector, is 12 Let be the angle ψ.

[0086] In addition, the probability distribution P ij (θ ij |φ ij) is given by equation (2), then the coordinate s j It is also possible to simultaneously find coordinates m (i=1, ..., M) and j (j=1, ..., N). Specifically, it is advisable to find X that satisfies the equation GX=C, which is defined using the matrix G obtained by arranging the conditions in equations (9) and (12).

[0087] Here, the matrix G, vectors X and C are given by the following equations.

[0088]

number

[0089]

number

[0090]

number

[0091]

number

[0092]

number

[0093]

number

[0094]

number

[0095] Vector X=[m3, …, m M , s1,…, s N ] T is the inverse matrix G of matrix G -1 Using this, X=G-1 Obtained by C. [Example 2] Probability distribution P ij (θ ij |φ ij ) is given by the following equation:

[0096]

number

[0097]

number

[0098] Here, f is a probability density function called the von Mises distribution, I0 is the first kind of modified Bessel function, β ij is a parameter that represents the distribution shape. Also, q ij (d ij ) is the distance d between the i-th sound source direction detector and the j-th sound source ij is a term that depends on

[0099] q ij (d ij A specific example of distance d ij When information about the sound source direction detector is available (e.g., the estimated distance from the sound pressure detected by the i-th sound source direction detector), q ij (d ij For example, the distance d ij If you know that is around d0, use the gamma distribution

[0100]

number

[0101] In addition, the distance d ij If you cannot find information about

[0102]

number

[0103] Here, L is a sufficiently large real number, and q ij (d ij ) is effectively a uniform distribution.

[0104] Also, b ij = [cosθ ij , sinθ ij ] T Then, from equations (3) and (4), the following equation is obtained from equation (22).

[0105]

number

[0106] However, in equation (24), d ij →The probability density function for 0 is not unique and is discontinuous, so β ij Using the following formula, d ij It is defined as a function of

[0107]

number

[0108] (where α is a given constant) β ij By defining as a function expressed by equation (25),

[0109]

number

[0110] and the probability distribution of equation (24) becomes continuous.

[0111] From the above, q ij (d ij ) is given by equation (23), the log-likelihood LL j (s j |θ j ) is expressed as follows:

[0112]

number

[0113] As in Example 1, when the symmetry between the sound source direction detector and the sound source is taken into consideration, the following equation is obtained from equation (26):

[0114]

number

[0115] As before, the coordinate s that maximizes equation (26) j Let us consider the angle vector θ j and coordinate m i Specifically, for example, the coordinate s that maximizes Equation (26) can be found by using the steepest descent method or grid search. j It is also useful to find the coordinate m that maximizes equation (27). i Let us consider the angle vector θ j and coordinate s j Specifically, for example, the coordinate m that maximizes Equation (27) can be obtained by using the steepest descent method or grid search. i It is a good idea to ask for

[0116] As before, the coordinates of the two sound source direction detectors m i For (i=1, 2), we constrain it using equation (14).

[0117] By using the above position estimation method, it is not necessary to synchronize and process the acoustic signals collected by all the microphones included in the M sound source direction detectors, and the position of the sound source can be estimated by using the sound source direction detectors autonomously. In addition, the position of the sound source direction detector can also be estimated. First Embodiment The position estimation device 100 estimates coordinates representing the positions of N sound sources and coordinates representing the positions of M sound source direction detectors, using angle vectors representing the directions of N sound sources detected using M sound source direction detectors (where M and N are integers equal to or greater than 2). Here, each sound source direction detector detects an angle representing the direction of a sound source, using synchronized acoustic signals obtained by collecting sounds emitted by the sound source with two or more microphones.

[0118] The position estimation device 100 will be described below with reference to Figs. 4 and 5. Fig. 4 is a block diagram showing the configuration of the position estimation device 100. Fig. 5 is a flowchart showing the operation of the position estimation device 100. As shown in Fig. 4, the position estimation device 100 is connected to M sound source direction detectors via a network such as the Internet. The position estimation device 100 also includes a sound source direction detection unit 110, a position estimation unit 120, and a recording unit 190. The recording unit 190 is a component that appropriately records information required for processing by the position estimation device 100.

[0119] The operation of the position estimation device 100 will be described with reference to FIG.

[0120] In S110, the sound source direction detection unit 110 calculates an angle vector θ representing the direction of the j-th sound source, where j=1, . . . , N. j =[θ 1j , θ 2j , …, θ Mj ] T (However, θ ij is an angle representing the direction of the jth sound source detected by the ith sound source direction detector) and is output.

[0121] The sound source direction detection unit 110 will be described below with reference to Figs. 6 to 7. Fig. 6 is a block diagram showing the configuration of the sound source direction detection unit 110. Fig. 7 is a flowchart showing the operation of the sound source direction detection unit 110. As shown in Fig. 6, the sound source direction detection unit 110 includes an initialization unit 111, a first angle vector generation unit 112, a second angle vector generation unit 113, a counter update unit 114, and an angle vector output unit 115.

[0122] The operation of the sound source direction detection unit 110 will be described with reference to FIG.

[0123] In S111, the initialization unit 111 sets an initial value of a counter j. That is, the initialization unit 111 sets j←1.

[0124] In S112, the first angle vector generation unit 112 detects a sound emitted from any one of the N sound sources at a time t i (i=1, …, M) and the i-th sound source direction detector is detected at time t i The angle θ represents the direction of the detected sound source. i (t i ) (i=1, …, M) and time t1, …, t M If the time t falls within a certain time, the angle vector θ(t)=[θ1(t1), …, θ M (t M )] T The first angle vector generation unit 112 is a component that executes the process (1) of the angle vector generation method described in <Technical Background>.

[0125] In S113, the second angle vector generation unit 113 receives the angle vector θ(t) generated in S112 as input, and uses the angle vector θ(t) to determine whether or not the sound source that emitted the sound detected at time t matches any of the j-1 sound sources that emitted sounds detected at times prior to time t. If it is determined that the sound source that emitted the sound detected at time t does not match any of the j-1 sound sources that emitted sounds detected at times prior to time t, it is determined that the sound source that emitted the sound detected at time t is a j-th sound source different from any of the j-1 sound sources that emitted sounds detected at times prior to time t, and generates the angle vector θ j = θ(t) and outputs it. The second angle vector generation unit 113 is a component that executes the process (2) of the angle vector generation method described in <Technical Background>.

[0126] In S114, the counter update unit 114 increments the counter j by 1. That is, the counter update unit 114 sets j←j+1.

[0127] In S115, when the counter j satisfies a predetermined termination condition (that is, when the counter j satisfies j>N), the angle vector output unit 115 outputs the angle vectors θ1, θ2, ..., θ N and ends the process, otherwise, the process returns to S112 and repeats the processes of S112 to S115.

[0128] Below, m i (i=1, …, M, where M is an integer equal to or greater than 2) is the coordinate representing the position of the i-th sound source direction detector, s j (j=1, …, N, where N is an integer equal to or greater than 2) is defined as the coordinates representing the position of the jth sound source.

[0129] In S120, the position estimation unit 120 calculates the angle vectors θ1, θ2, ..., θ N The input is the angle vector θ j (j=1, …, N) using coordinate s j (j=1, …, N) and coordinate m i (i=1, …, M) is estimated and output.

[0130] Hereinafter, the position estimation unit 120 will be described with reference to Figs. 8 and 9. Fig. 8 is a block diagram showing the configuration of the position estimation unit 120. Fig. 9 is a flowchart showing the operation of the position estimation unit 120. As shown in Fig. 8, the position estimation unit 120 includes an initialization unit 121, a sound source position update unit 122, a sound source direction detector position update unit 123, a counter update unit 124, and a position output unit 125.

[0131] The operation of the position estimation unit 120 will be described with reference to FIG.

[0132] In S121, the initialization unit 121 sets the initial value of the counter t and the coordinate s j Initial value of (j=1, …, N), coordinate mi (i=1, ..., M). In other words, the initialization unit 121 sets t←1. The initialization unit 121 calculates the coordinate m i Set the initial value of (i=1,2). Note that the coordinate s j Initial value and coordinate m for (j=1, …, N) i The initial values ​​of (i=3, …, M) can be set to appropriate values. Here, the coordinate m i The value of (i=1,2) will not be updated in subsequent processing.

[0133] In S122, the sound source position update unit 122 updates the angle vector θ j (j=1, …, N) and coordinate m i (i=1, …, M), the coordinate s j Update the value of (j=1, …, N).

[0134] In S123, the sound source direction detector position update unit 123 updates the angle vector θ j (j=1, …, N) and coordinate s j (j=1, …, N), coordinate m i Update the values ​​of (i=1, …, M).

[0135] In S124, the counter update unit 124 increments the counter t by 1. That is, the counter update unit 124 sets t←t+1.

[0136] In S125, when the counter t satisfies a predetermined termination condition (that is, when the counter t satisfies t>T (where T is a predetermined integer representing the upper limit of the number of updates)), the position output unit 125 outputs the coordinate s j (j=1, …, N) and coordinate m i (i=1, . . . , M) is output and the process ends, whereas in other cases the process returns to S122 and repeats the processes of S122 to S125.

[0137] An example of the processes in S122 and S123 will be described below. ijis the angle representing the direction of the jth sound source from the ith sound source direction detector, P ij (θ ij |φ ij ) given by the given formula: angle θ ij and angle φ ij The angle φ represents the deviation from ij -θ ij Let us assume that the probability distribution is (Example 1) Probability distribution P ij (θ ij |φ ij ) is given by

[0138]

number

[0139] (However, ij =||s j -m i ||, σ is a given constant) A ij If is a matrix calculated by the following equation, the processes of S122 and S123 are as follows:

[0140]

number

[0141] In S122, the sound source position update unit 122 calculates the coordinate s j Update the value of (j=1, …, N).

[0142]

number

[0143]

number

[0144] In S123, the sound source direction detector position update unit 123 updates the coordinate m iUpdate the value of (i=3, …, N).

[0145]

number

[0146]

number

[0147] (Example 2) Probability distribution P ij (θ ij |φ ij ) is given by equation (28), then a ij = [-sinθ ij , cosθ ij ] T Then, the processing of S122 and S123 is as follows.

[0148] In S122, the sound source position update unit 122 updates the coordinate s j Update the value of (j=1, …, N).

[0149]

number

[0150] In S123, the sound source direction detector position update unit 123 updates the coordinate m i Update the value of (i=3, …, N).

[0151]

number

[0152] (Example 3) Probability distribution P ij (θ ij |φ ij ) is given by the following formula, b ij = [cosθ ij , sinθij ] T Then, the processing of S122 and S123 is as follows.

[0153]

number

[0154]

number

[0155]

number

[0156]

number

[0157] (However, ij =||s j -m i ||, L is a given constant, I0 is the first kind of modified Bessel function, and α is a given constant) In S122, the sound source position update unit 122 updates the coordinate s j Update the value of (j=1, …, N).

[0158]

number

[0159] In S123, the sound source direction detector position update unit 123 updates the coordinate m i Update the value of (i=3, …, N).

[0160]

number

[0161] According to the embodiment of the present invention, it is possible to estimate the sound source position even if all the acoustic signals picked up by the microphones are not synchronized. In addition, the system based on Non-Patent Document 1 requires simultaneous A / D conversion via wired connection, but in a real environment, it is not always possible to properly connect all the microphones, which causes a problem of poor usability. However, according to the embodiment of the present invention, it is not necessary to wire the sound source direction detector, and usability is improved. <Second embodiment> Like the position estimation device 100, the position estimation device 200 estimates coordinates representing the positions of N sound sources and coordinates representing the positions of M sound source direction detectors using angle vectors representing the directions of N sound sources detected using M sound source direction detectors.

[0162] Hereinafter, the position estimation device 200 will be described with reference to Figs. 4 and 5. Fig. 4 is a block diagram showing the configuration of the position estimation device 200. Fig. 5 is a flowchart showing the operation of the position estimation device 200. As shown in Fig. 4, the position estimation device 200 is connected to M sound source direction detectors via a network such as the Internet. The position estimation device 200 also includes a sound source direction detection unit 110, a position estimation unit 220, and a recording unit 190. The recording unit 190 is a component that appropriately records information required for the processing of the position estimation device 200. In other words, the position estimation device 200 differs from the position estimation device 100 only in that the position estimation device 200 includes the position estimation unit 220 instead of the position estimation unit 120.

[0163] The operation of the position estimation device 200 will be described with reference to Fig. 5. Here, only the operation of the position estimation unit 220 will be described.

[0164] In S220, the position estimation unit 220 calculates the angle vectors θ1, θ2, ..., θ N The input is the angle vector θ j (j=1, …, N) using coordinate s j (j=1, …, N) and coordinate m i(i=1, ..., M) and outputs it. The position estimation unit 220 is a component that obtains X that satisfies the equation GX=C defined using the matrix G and vectors X and C of the equations (15) to (21).

[0165] According to the embodiment of the present invention, it is possible to estimate the sound source position even if all the acoustic signals picked up by the microphones are not synchronized. In addition, the system based on Non-Patent Document 1 requires simultaneous A / D conversion via wired connection, but in a real environment, it is not always possible to properly connect all the microphones, which causes a problem of poor usability. However, according to the embodiment of the present invention, it is not necessary to wire the sound source direction detector, and usability is improved. <Application Examples> A device worn by a person, such as a smartphone or earphones, may be used as a sound source direction detector, and the person wearing the device may be used as the sound source to be estimated.

[0166] For example, if there are multiple people within the error range of the GPS, the position of the person can be estimated by estimating the position of the smartphone using Bluetooth, but the direction in which the person wearing the smartphone is facing cannot be estimated. In addition, if the person is not wearing the smartphone, such as when the smartphone is placed on a desk, the position of the person cannot be estimated. In such a case, by using the smartphone as a sound source direction detector, it is possible to estimate the position and direction of the person who is the sound source. For example, it is possible to estimate the direction in which the person sneezed, and the position of the person can be estimated during a meal regardless of whether the person is wearing a smartphone or not. In other words, it is possible to identify the person affected by droplets caused by sneezing or speaking. Furthermore, if the sound source direction detector can detect sound pressure, it is possible to identify more accurately. <Additional Notes> 10 is a diagram showing an example of the functional configuration of a computer that realizes each of the above-mentioned devices (i.e., each node). The processing in each of the above-mentioned devices can be implemented by having a recording unit 2020 load a program for causing a computer to function as each of the above-mentioned devices, and having the control unit 2010, input unit 2030, output unit 2040, etc. operate.

[0167] The device of the present invention has, as a single hardware entity, an input section to which a keyboard or the like can be connected, an output section to which a liquid crystal display or the like can be connected, a communication section to which a communication device (e.g., a communication cable) capable of communicating with the outside of the hardware entity can be connected, a CPU (which may also have a central processing unit, cache memory, registers, etc.), memories such as RAM and ROM, an external storage device such as a hard disk, and buses connecting the input section, output section, communication section, CPU, RAM, ROM, and external storage device so that data can be exchanged between them. If necessary, the hardware entity may also be provided with a device (drive) capable of reading and writing a recording medium such as a CD-ROM. A physical entity equipped with such hardware resources is, for example, a general-purpose computer.

[0168] The external storage device of the hardware entity stores the programs required to realize the above-mentioned functions and the data required in the processing of these programs (not limited to the external storage device, but for example the programs may be stored in a ROM, which is a read-only storage device). Data obtained by the processing of these programs is stored appropriately in the RAM, the external storage device, etc.

[0169] In the hardware entity, each program stored in an external storage device (or ROM, etc.) and the data required for processing each program are loaded into memory as necessary, and interpreted, executed, and processed by the CPU as appropriate, so that the CPU realizes a predetermined function (each component represented as the above, "unit," "means," etc.).

[0170] The present invention is not limited to the above-described embodiment, and can be modified as appropriate without departing from the spirit of the present invention. In addition, the processes described in the above-described embodiment are not limited to being executed in chronological order according to the order described, but may be executed in parallel or individually depending on the processing capacity of the device that executes the processes or as necessary.

[0171] As described above, when the processing functions of the hardware entities (the devices of the present invention) described in the above embodiments are realized by a computer, the processing contents of the functions that the hardware entities should have are described by a program. Then, by executing this program on a computer, the processing functions of the hardware entities are realized on the computer.

[0172] The program describing the processing contents can be recorded on a computer-readable recording medium. The computer-readable recording medium may be, for example, a magnetic recording device, an optical disk, a magneto-optical recording medium, a semiconductor memory, or any other suitable recording medium. Specifically, for example, a hard disk drive, a flexible disk, a magnetic tape, or the like can be used as the magnetic recording device, a DVD (Digital Versatile Disc), a DVD-RAM (Random Access Memory), a CD-ROM (Compact Disc Read Only Memory), a CD-R (Recordable) / RW (ReWritable), or the like can be used as the optical disk, an MO (Magneto-Optical disc), or the like can be used as the magneto-optical recording medium, and an EEP-ROM (Electronically Erasable and Programmable-Read Only Memory) or the like can be used as the semiconductor memory.

[0173] The program may be distributed, for example, by selling, transferring, lending, etc. portable recording media such as DVDs and CD-ROMs on which the program is recorded. Furthermore, the program may be distributed by storing the program in a storage device of a server computer and transferring the program from the server computer to other computers via a network.

[0174] A computer that executes such a program, for example, first stores the program recorded on a portable recording medium or the program transferred from a server computer in its own storage device. Then, when executing a process, the computer reads the program stored in its own storage device and executes a process according to the read program. As another execution form of the program, the computer may directly read the program from the portable recording medium and execute a process according to the program, or may execute a process according to the received program each time a program is transferred from the server computer to the computer. The server computer may not transfer the program to the computer, but may execute the above-mentioned process by a so-called ASP (Application Service Provider) type service that realizes the processing function only by issuing an execution instruction and obtaining the result. Note that the program in this embodiment includes information used for processing by an electronic computer that is equivalent to a program (data that is not a direct instruction to the computer but has a nature that specifies the processing of the computer, etc.).

[0175] In addition, in this embodiment, a hardware entity is configured by executing a specific program on a computer, but at least a part of the processing contents may be realized by hardware.

[0176] The foregoing description of the embodiments of the invention has been presented for purposes of illustration and description. It is not intended to be exhaustive or to limit the invention to the precise form disclosed. Modifications and variations are possible in light of the above teachings. The embodiments have been chosen and depicted to provide a best illustration of the principles of the invention and to enable those skilled in the art to utilize the invention in various embodiments and with various modifications as may be suitable for the practical use contemplated. All such modifications and variations are within the scope of the invention as defined by the appended claims interpreted in accordance with the breadth to which they are fairly, legally, and equitably entitled.

Claims

1. m i (i=1, …, M, where M is an integer equal to or greater than 2) is the coordinate representing the position of the i-th sound source direction detector, s j (j=1, …, N, where N is an integer equal to or greater than 2) is the coordinate representing the position of the jth sound source. θ j =[θ 1j , θ 2j , …, θ Mj ] T (j=1, …, N, where θ ij is the angle representing the direction of the jth sound source detected by the i-th sound source direction detector) is the angle vector representing the direction of the jth sound source, Each sound source direction detector detects an angle representing the direction of a sound source by using a synchronized acoustic signal obtained by collecting a sound emitted by the sound source with two or more microphones, an initialization step in which a position estimation unit of the position estimation device sets an initial value of a counter t, an initial value of a coordinate sj (j=1, ..., N), and an initial value of a coordinate mi (i=1, ..., M); The position estimation unit of the position estimation device calculates an angle vector θ j (j=1, …, N) and coordinate m i (i=1, …, M), the coordinate s j A sound source position update step in which the values ​​of (j=1, …, N) are updated; The position estimation unit of the position estimation device calculates an angle vector θ j (j=1, …, N) and coordinate s j (j=1, …, N), coordinate m i A sound source direction detector position update step for updating the values ​​of (i=1, …, M); a counter update step in which a position estimation unit of the position estimation device increments a counter t by 1; a position output step in which, if the counter t satisfies a predetermined termination condition, the position estimation unit of the position estimation device outputs coordinates sj (j=1, ..., N) and coordinates mi (i=1, ..., M) and ends the process, and otherwise returns to the sound source position update step; A location estimation method comprising:

2. 2. A position estimation method according to claim 1, comprising: t i (i=1, …, M) is the time when the i-th sound source direction detector detects a sound emitted by one of the N sound sources, and θ i (t i )(i=1, …, M) is the sound source direction detector detected at time t i is the angle that represents the direction of the detected sound source, an initialization step in which a sound source direction detection unit of the position estimation device sets an initial value of a counter j; The sound source direction detection unit of the position estimation device detects the sound source direction at time t 1 , …, t M If the time t falls within a certain time, the angle vector θ(t) = [θ 1 (t 1 ), …, θ M (t M )] T A first angle vector generation step of generating A sound source direction detection unit of the position estimation device uses the angle vector θ(t) to determine whether or not the sound source that emitted the sound detected at time t matches any of the j-1 sound sources that emitted sounds detected at times prior to time t. If it is determined that the sound source that emitted the sound detected at time t does not match any of the j-1 sound sources that emitted sounds detected at times prior to time t, the sound source that emitted the sound detected at time t is a j-th sound source different from any of the j-1 sound sources that emitted sounds detected at times prior to time t, and the angle vector θ j = θ(t), a counter update step in which a sound source direction detection unit of the position estimation device increments a counter j by 1; an angle vector output step in which, if the counter j satisfies a predetermined termination condition, the sound source direction detection unit of the position estimation device outputs angle vectors θ 1 , θ 2 , ..., θ N and terminates the process, and otherwise returns to the first angle vector generation step; Includes A position estimation method comprising:

3. 2. A position estimation method according to claim 1, comprising: φ ij is the angle representing the direction of the jth sound source from the ith sound source direction detector, P ij (θ ij |φ ij ) is given by the following equation: ij and angle φ ij The angle φ represents the deviation from ij -θ ij Let the probability distribution be [0,58] (However, d ij =||s j -m i ||, σ is a given constant) A ij Let be a matrix calculated as follows: [0,59] The sound source position update step is performed by calculating the coordinate s j (j=1, …, N) [0.60] [0061] The sound source direction detector position update step is performed by calculating the coordinate m i (i=3, …, N) [0062] [0063] A position estimation method comprising:

4. 2. A position estimation method according to claim 1, comprising: φ ij is the angle representing the direction of the jth sound source from the ith sound source direction detector, P ij (θ ij |φ ij ) is given by the following equation: ij and angle φ ij The angle φ represents the deviation from ij -θ ij Let the probability distribution be [0,64] (However, d ij =||s j -m i ||, σ is a given constant) a ij = [-sinθ ij , cosθ ij ] T year, The sound source position updating step is to update the coordinate s j (j=1, …, N) [0.65] The sound source direction detector position update step is performed by maximizing the following equation: i (i=3, …, N) [0.66] A position estimation method comprising:

5. 2. A position estimation method according to claim 1, comprising: φ ij is the angle representing the direction of the jth sound source from the ith sound source direction detector, P ij (θ ij |φ ij ) is given by the following equation: ij and angle φ ij The angle φ represents the deviation from ij -θ ij Let the probability distribution be [0.67] [0068] [0,69] [Number 70] (However, d ij =||s j -m i ||, L is a given constant, I 0 is the first kind of modified Bessel function, α is a given constant) b ij = [cosθ ij , sinθ ij ] T year, The sound source position updating step is to update the coordinate s j (j=1, …, N) [Equation 71] The sound source direction detector position update step is performed by maximizing the following equation: i (i=3, …, N) [Equation 72] A position estimation method comprising:

6. m i (i=1, …, M, where M is an integer equal to or greater than 2) is the coordinate representing the position of the i-th sound source direction detector, s j (j=1, …, N, where N is an integer equal to or greater than 2) is the coordinate representing the position of the jth sound source. θ j =[θ 1j , θ 2j , …, θ Mj ] T (j=1, …, N, where θ ij is the angle representing the direction of the jth sound source detected by the i-th sound source direction detector) is the angle vector representing the direction of the jth sound source, Each sound source direction detector detects an angle representing the direction of a sound source by using a synchronized acoustic signal obtained by collecting a sound emitted by the sound source with two or more microphones, A ij Let be a matrix calculated as follows: [Equation 73] Let X be a vector defined as [74] The position estimation unit of the position estimation device obtains the vector X that satisfies the equation GX=C, and calculates the coordinate s j (j=1, …, N) and coordinate m i (i=3, …, M) (However, the matrix G and vector C are calculated using the following formula.) [Number 75] [76] [77] [78] [79] [Number 80] A location estimation method comprising:

7. m i (i=1, …, M, where M is an integer equal to or greater than 2) is the coordinate representing the position of the i-th sound source direction detector, s j (j=1, …, N, where N is an integer equal to or greater than 2) is the coordinate representing the position of the jth sound source. θ j =[θ 1j , θ 2j , …, θ Mj ] T (j=1, …, N, where θ ij is the angle representing the direction of the jth sound source detected by the i-th sound source direction detector) is the angle vector representing the direction of the jth sound source, Each sound source direction detector detects an angle representing the direction of a sound source by using a synchronized acoustic signal obtained by collecting a sound emitted by the sound source with two or more microphones, an initialization unit that sets an initial value of a counter t, an initial value of a coordinate sj (j=1, ..., N), and an initial value of a coordinate mi (i=1, ..., M); Angle vector θ j (j=1, …, N) and coordinate m i (i=1, …, M), the coordinate s j A sound source position update unit that updates the values ​​of (j=1, …, N); Angle vector θ j (j=1, …, N) and coordinate s j (j=1, …, N), coordinate m i A sound source direction detector position update unit that updates the values ​​of (i=1, …, M); a counter update unit that increments a counter t by 1; a position output unit that outputs coordinates sj (j=1, ..., N) and coordinates mi (i=1, ..., M) and ends the process when the counter t satisfies a predetermined termination condition, and otherwise returns to the process by the sound source position update unit; A position estimation device comprising:

8. A program for causing a computer to execute the position estimation method according to any one of claims 1 to 6.

Images