WO2022118367A1 - Sound source direction estimation device, program, and sound source direction estimation method - Google Patents

Abstract

A sound source direction estimation device (130) comprises: an acquisition unit (131) that acquires first sound data in which the sound inside an elevator shaft through which an elevator car travels a first time is acquired by a sound collection unit pointed in a first direction on a predetermined plane, and second sound data in which the sound inside the elevator shaft through which the elevator car travels a second time is acquired by the sound collection unit pointed in a second direction on the plane; a score calculation unit (134) that calculates, from the first sound, a plurality of first scores, which are a plurality of scores indicating the possibility of being a sound source direction for each angle relative to the first direction in the plane, and calculates, from the second sound, a plurality of second scores, which are a plurality of scores indicating the possibility of being a sound source direction for each angle relative to the second direction in the plane; and a sound source direction estimation unit (135) that uses the first scores and the second scores to estimate the sound source direction of a target sound when the car is in a particular location in the elevator shaft.

Description

Sound source direction estimation device, program and sound source direction estimation method

The present disclosure relates to a sound source direction estimation device, a program, and a sound source direction estimation method.

Conventionally, there is a technique of estimating the sound source direction of abnormal sound using a 2ch (channel) microphone array in the hoistway of an elevator (see, for example, Patent Document 1). However, with the 2ch microphone array, basically, the sound source direction can only be estimated up to the range of 180 degrees in the horizontal direction.
Therefore, there is a technique for estimating the sound source direction by changing the direction of the microphone array. For example, there is a technique of changing the direction of the 2ch microphone array between the first round trip and the second round trip to integrate the sound source estimation result of the first round trip and the sound source estimation result of the second round trip.

Japanese Unexamined Patent Publication No. 2013-060295

However, when estimating the sound source position using a 2ch microphone array in an environment where a lot of sound reflection occurs such as an elevator hoistway and an error in sound source direction estimation is likely to occur, one round trip is caused by the error in sound source direction estimation. It may not be possible to integrate the sound source direction estimation result of the eye and the sound source direction estimation result of the second round trip.

Therefore, one or more aspects of the present disclosure are intended to ensure that the sound source direction estimation result of the first round trip and the sound source direction estimation result of the second round trip can be integrated.

In the sound source direction estimation device according to one aspect of the present disclosure, the sound in the hoistway where the elevator car moves for the first time is acquired by the sound collecting unit directed in the first direction on a predetermined plane. Acquisition of acquiring the sound data of 1 and the second sound data acquired by the sound collecting unit directed in the second direction in the plane for the sound in the hoistway where the car moves for the second time. When the car is in a specific position in the hoistway, it is included in the first sound, which is the sound indicated by the first sound data, for each angle with respect to the first direction in the plane. A plurality of first scores, which are a plurality of scores indicating the possibility of the sound source direction of the target sound being recorded, are calculated from the first sound, and the above-mentioned is performed for each angle with respect to the second direction in the plane. A plurality of second scores, which are a plurality of scores indicating the possibility of the target sound included in the second sound, which is the sound indicated by the second sound data, in the direction of the sound source, are referred to as the second score. A score calculation unit calculated from sound and a sound source direction estimation unit that estimates the sound source direction using the first score and the second score are provided, and the sound source direction estimation unit is described in the plane. A first straight line extending from the sound collecting unit at an angle corresponding to the first selected score, which is one score selected from the plurality of first scores, is orthogonal to the first direction in the plane. From the first locus in which the point where the first straight line comes into contact with the unit sphere centered on the sound collecting portion moves by rotating the axis around the sound, and from the plurality of second scores in the plane. A second straight line extending from the sound collecting unit at an angle corresponding to the second selected score, which is one selected score, is rotated about an axis orthogonal to the second direction in the plane. This is characterized in that the intersection of the second straight line with the second locus on which the point of contact with the unit sphere moves is set to the sound source direction.

In the program according to one aspect of the present disclosure, the computer acquires the sound in the hoistway where the elevator car moves for the first time at the sound collecting unit directed in the first direction in a predetermined plane. Acquisition of acquiring the sound data of 1 and the second sound data acquired by the sound collecting unit directed in the second direction in the plane for the sound in the hoistway where the car moves for the second time. A unit, when the car is in a specific position in the hoistway, it is included in the first sound, which is the sound indicated by the first sound data, for each angle with respect to the first direction in the plane. A plurality of first scores, which are a plurality of scores indicating the possibility of being the sound source direction of the target sound, are calculated from the first sound, and the first score is calculated for each angle with respect to the second direction in the plane. A plurality of second scores, which are a plurality of scores indicating the possibility of the target sound being directed to the sound source, included in the second sound, which is the sound indicated by the sound data of 2, are obtained from the second sound. The score calculation unit calculated from the above and the sound source direction are made to function as a sound source direction estimation unit estimated by using the first score and the second score, and the sound source direction estimation unit is the above-mentioned in the plane. A first straight line extending from the sound collecting unit at an angle corresponding to the first selected score, which is one score selected from the plurality of first scores, is orthogonal to the first direction in the plane. From the first locus in which the point where the first straight line comes into contact with the unit sphere centered on the sound collecting portion moves by rotating the axis around the sound, and from the plurality of second scores in the plane. A second straight line extending from the sound collecting unit at an angle corresponding to the second selected score, which is one selected score, is rotated about an axis orthogonal to the second direction in the plane. This is characterized in that the intersection of the second straight line with the second locus on which the point of contact with the unit sphere moves is set to the sound source direction.

In the sound source direction estimation method according to one aspect of the present disclosure, the sound in the hoistway where the elevator car moves for the first time is acquired by a sound collecting unit directed in the first direction on a predetermined plane. The sound data of 1 and the second sound data acquired by the sound collecting unit in which the sound in the hoistway where the car moves for the second time is directed to the second direction in the plane are acquired. When the car is in a specific position in the hoistway, it is included in the first sound, which is the sound indicated by the first sound data, for each angle with respect to the first direction in the plane. A plurality of first scores, which are a plurality of scores indicating the possibility of the sound source direction of the target sound, are calculated from the first sound, and the second score is calculated for each angle with respect to the second direction in the plane. A plurality of second scores, which are a plurality of scores indicating the possibility of the target sound included in the second sound, which is the sound indicated by the sound data, in the direction of the sound source, are calculated from the second sound. It is a sound source direction estimation method that estimates the sound source direction using the first score and the second score, and is one score selected from the plurality of first scores in the plane. The first straight line, which is a straight line extending from the sound collecting portion at an angle corresponding to the first selection score, is rotated around an axis orthogonal to the first direction in the plane. Corresponds to the first locus in which the point of contact with the unit sphere centered on the sound collecting unit moves, and the second selection score, which is one score selected from the plurality of second scores in the plane. By rotating the second straight line, which is a straight line extending from the sound collecting portion at an angle, about an axis orthogonal to the second direction in the plane, the second straight line comes into contact with the unit sphere. It is characterized in that the intersection with the second locus on which the point moves is the sound source direction.

According to one or more aspects of the present disclosure, the sound source direction estimation result of the first round trip and the sound source direction estimation result of the second round trip can be reliably integrated.

It is a block diagram which shows roughly the structure of the sound source direction estimation system in Embodiments 1 to 4. It is a schematic diagram which shows the installation example of the microphone array. It is a block diagram which shows an example of the sound source direction estimation apparatus in Embodiment 1. FIG. It is a top view which shows the 1st example which put the microphone array on the car of an elevator. It is a top view which shows the 2nd example which put the microphone array on the car of an elevator. It is the first graph which shows the relationship between a score and an angle. It is a second graph which shows the relationship between a score and an angle. It is a first perspective view which shows the relationship between the microphone array, the unit sphere centered on the center point of the microphone array, and the score. 2 is a second perspective view showing the relationship between the microphone array, the unit sphere centered on the center point of the microphone array, and the score. It is a third perspective view which shows the relationship between the microphone array, the unit sphere centered on the center point of the microphone array, and the score. It is a fourth perspective view which shows the relationship between the microphone array, the unit sphere centered on the center point of the microphone array, and the score. It is a flowchart which shows the process at the time of acquiring a sound data in a sound source direction estimation system. It is a flowchart which shows the process of estimating the sound source direction by the sound source direction estimation apparatus in Embodiment 1. FIG. It is a block diagram which shows an example of the sound source direction estimation apparatus in Embodiment 2. It is a perspective view which shows the state which the score of the 1st round trip was reversed. It is a perspective view which shows the state which the score of the 2nd round trip was reversed. It is a perspective view which shows the example which shifted the angle in the score of the 2nd round trip. (A) to (E) are tables showing the relationship between the angle and the score. It is a block diagram which shows an example of the sound source direction estimation apparatus in Embodiment 3. FIG. It is a flowchart which shows the process of estimating the sound source direction by the sound source direction estimation apparatus in Embodiment 3. FIG. It is a block diagram which shows an example of the sound source direction estimation apparatus in Embodiment 4.

Hereinafter, embodiments will be described with reference to the drawings. The following embodiments are merely examples, and various modifications can be made within the scope of the present disclosure.

Embodiment 1.
FIG. 1 is a block diagram schematically showing the configuration of the sound source direction estimation system 100 according to the first embodiment.
The sound source direction estimation system 100 includes a computer 101 that functions as a sound source direction estimation device, a microphone array 110, an input device 111, an output device 112, and a sensor 113.

The sound source direction estimation device realized by the computer 101 is a device that executes the sound source direction estimation method. For example, the computer 101 may be a portable terminal device such as a smartphone.

The microphone array 110 is a device including a plurality of microphones. In the microphone array 110, the sound data output from each of the plurality of microphones is output in a completely synchronized state.
The microphone array 110 may be separated from the computer 101 or may be incorporated into the computer 101 to be integrated. Further, the microphone array 110 may be equipped with an acceleration sensor or an image sensor. These sensors may be installed in the vicinity of the microphone array 110.

The input device 111 is a device that receives input from the user. The input device 111 may be separated from the computer 101 or may be incorporated into the computer 101 to be integrated.

The output device 112 is a device that outputs the processing result of the computer 101. The output device 112 may be separated from the computer 101 or may be incorporated into the computer 101 to be integrated. For example, the output device 112 is a display, but may be a speaker.

The input device 111 and the output device 112 may be configured by, for example, a touch panel.

The sensor 113 is a sensor that acquires a synchronization signal for synchronizing the sound data acquired by the microphone array 110. For example, the sensor 113 is an acceleration sensor or an image sensor.

As shown in FIG. 1, the computer 101 includes a processor 102, a main storage device 103, an auxiliary storage device 104, a communication device 105, and an interface (I / F) 106.

The processor 102 controls the entire computer 101. For example, the processor 102 is a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), an FPGA (Field Programmable Gate Array), or the like. The processor 102 may be a multiprocessor. The computer 101 may have a processing circuit instead of the processor 102. The processing circuit may be a single circuit or a composite circuit. In other words, the computer can be configured by a processing network.

The main storage device 103 is, for example, a RAM (Random Access Memory).
The auxiliary storage device 104 is, for example, a ROM (Read Only Memory), an HDD (Hard Disk Drive), or an SSD (Solid State Drive).
The main storage device 103 and the auxiliary storage device 104 store various data and programs necessary for processing by the computer 101.

The communication device 105 executes communication via the network. The communication device 105 is, for example, a wired LAN (Local Area Network) adapter, a wireless LAN adapter, or a Bluetooth (registered trademark) adapter. The communication device 105 may be referred to as a communication interface. The communication device 105 communicates with an external device.

The microphone array 110 is a sound collecting device that functions as a sound collecting unit installed in the hoistway of an elevator. For example, the microphone array 110 is installed above the cage, below the cage, above the counterweight, or below the counterweight.
When the microphone array 110 is installed on the car, the microphone array 110 is interlocked with the car. Here, the car may be referred to as an elevator car. The hoistway may be referred to as an elevator hoistway.
Further, the microphone array 110 may be installed in a pit portion, near a hoist, or in a place where the position does not change due to the elevator operation. The microphone array 110 may be fixed to the hoistway. Further, the microphone array 110 brought by the user may be installed in the hoistway by the user.

FIG. 2 is a schematic view showing an installation example of the microphone array 110.
FIG. 2 shows a case where the microphone array 110 is installed on a car in the first embodiment.

FIG. 2 shows a wall surface 120, a car guide rail 121, a car 122, a car upper guide shoe 123, a car lower guide shoe 124, a counterweight guide rail 125, a counterweight 126, and a counterweight upper guide shoe. 127 and a counterweight lower guide shoe 128 are shown.

The wall surface 120 is a concrete wall surface.
The car guide rail 121 is a rail for moving the car 122 up and down. The car 122 moves up and down along the car guide rail 121. Therefore, the car guide rail 121 serves as a hoistway for the elevator.
The car upper guide shoe 123 and the car lower guide shoe 124 are joints between the car guide rail 121 and the car 122.

The counterweight guide rail 125 is a rail for moving the counterweight 126 up and down.
The counterweight upper guide shoe 127 and the counterweight lower guide shoe 128 are joints between the counterweight guide rail 125 and the counterweight 126.

FIG. 3 is a block diagram showing an example of the sound source direction estimation device according to the first embodiment.
As shown in FIG. 3, the sound source direction estimation device 130 includes an acquisition unit 131, a synchronization unit 132, a data storage unit 133, a score calculation unit 134, a sound source direction estimation unit 135, and an input unit 140. , The output unit 141 is provided.

The acquisition unit 131 acquires sound data. For example, the acquisition unit 131 acquires sound data from the microphone array 110. Further, for example, the acquisition unit 131 may acquire the sound data from a recording medium in which the sound data is recorded.
The sound data is data showing the sound in the hoistway where the elevator car 122 moves. Further, the sound data is sound data of a plurality of channels.

Specifically, the acquisition unit 131 is a microphone array 110 which is a sound collection unit in which the sound in the hoistway where the elevator car 122 moves for the first time is directed to the first direction in a predetermined plane. The first sound data, which is the acquired sound data, and the sound in the hoistway where the car 122 moves for the second time are the sound data acquired by the microphone array 110 directed in the second direction on the plane. Acquire a certain second sound data. In the present embodiment, the predetermined plane is a horizontal plane, but the present invention is not limited to such an example.

Further, the acquisition unit 131 acquires synchronization data indicating a signal or the like for synchronization processing performed by the synchronization unit 132. The synchronization data is acquired from, for example, a sensor 113 such as an acceleration sensor or an image sensor.

The microphone array 110 and the sensor 113 are mounted on the elevator car 122, for example, and output sound data and synchronization data when the elevator car 122 is reciprocated a plurality of times.

4 and 5 show a top view showing an example of a case where the microphone array 110 is placed on the elevator car 122 and reciprocated twice.
In FIG. 4, the direction from the microphone array 110 toward the door 129 is referred to as the door direction. As shown in FIG. 4, the microphone array 110 is directed toward the door in the first round trip, which is the first movement.

Further, as shown in FIG. 5, in the second round trip, which is the second movement, the microphone array 110 is directed to the right where the counterweight 126 is present.
As described above, the orientation of the microphone array 110 is rotated by 90 degrees between the first round trip and the second round trip. In this way, the orientation of the microphone array 110 in the horizontal plane is changed each time the elevator car 122 is reciprocated.

Returning to FIG. 3, the synchronization unit 132 performs synchronization processing for aligning the start points of the respective sound data when the microphone array 110 is placed on the elevator car 122 and reciprocates in the hoistway a plurality of times. In the synchronization process here, the acceleration data obtained from the acceleration sensor, or the image data or the moving image data obtained from the image sensor is used.

For example, when the elevator car 122 is reciprocated twice, the synchronization unit 132 obtains the mutual correlation coefficient from the acceleration data acquired for each of the first and second round trips. By applying the shift amount having the highest mutual correlation coefficient to the sound data, synchronization can be achieved even between the sound data.

Further, in the case of moving image data, the synchronization unit 132 obtains, for example, an optical flow. Then, the synchronization unit 132 obtains the mutual correlation coefficient between the first round trip and the second round trip with respect to the time change of the optical flow, and applies the shift amount at which the mutual correlation coefficient becomes the highest to the sound data. It is possible to synchronize between sound data.

The synchronization unit 132 may directly calculate the mutual correlation coefficient from the sound data and shift one of the sound data by the shift amount at which the mutual correlation coefficient is the highest to achieve synchronization.
Further, the synchronization unit 132 may receive input of a shift amount of sound data from the user via the input device 111 so that the sound data for two round trips are synchronized.

Regarding the synchronization accuracy of the synchronization unit 132, it is not necessary that the sound data of the first round trip and the second round trip are completely synchronized. The guideline for the accuracy of synchronization is about the duration of the target sound, which is the target sound for which the sound source direction is estimated. For example, when the duration of the target sound is 0.1 second, it is necessary to synchronize so that the target sound is generated at the same timing in the sound data of the first round trip and the second round trip. In this case, it is necessary to synchronize the sound data of the first round trip and the second round trip with an accuracy of about 0.01 to 0.05 seconds.

The data storage unit 133 stores data necessary for processing in the sound source direction estimation device 130. For example, the data storage unit 133 includes sound data acquired when the elevator car 122 reciprocates in the hoistway a plurality of times, acceleration data or image data from the sensor 113, and orientation information indicating the orientation of the microphone array 110. Remember.
Further, the data storage unit 133 stores parameters such as the relationship between the positions of the microphones required for the score calculation process in the score calculation unit 134.

The score calculation unit 134 calculates the score of the target sound acquired by the microphone array 110 in the target direction. For example, the score calculation unit 134 identifies target directions that are a plurality of directions on a horizontal plane in which the microphone array 110 is arranged, and scores indicating that each of the plurality of target directions may be a sound source of the target sound. Is calculated. Specifically, when the range of the horizontal angle of the sound acquired by the microphone array 110 is 180 degrees and the resolution in the horizontal direction is 10 degrees, the score calculation unit 134 calculates the score in 19 directions. In this way, the score calculation unit 134 calculates the scores in the plurality of target directions. The horizontal angle is an angle in the horizontal plane with respect to the direction in which the microphone array 110 is facing.

In the present embodiment, when the car 122 is in a specific position in the hoistway, the score calculation unit 134 makes an angle with respect to the first direction, which is the direction in which the microphone array 110 is facing in the horizontal plane in the first round trip. For each, a plurality of first scores, which are a plurality of scores indicating the possibility of being the sound source direction of the target sound included in the first sound, which is the sound indicated by the first sound data, are set to the first. It is calculated from the first sound indicated by the sound data.
Further, in the second round trip, the score calculation unit 134 converts the second sound, which is the sound indicated by the second sound data, into the second sound for each angle with respect to the second direction, which is the direction in which the microphone array 110 is facing in the horizontal plane. A plurality of second scores, which are a plurality of scores indicating the possibility of being the sound source direction of the included target sound, are calculated.

Beamforming, delay sum method, maximum likelihood method, minimum variance method, MUSIC (Multiple SIgnal Classification) method, root-MUSIC method, minimum norm method, CSP (CSP) are used to calculate the score. The Power Spectram Phase Analysis) method or the trained model may be used. The values calculated by these methods may be used as they are as scores, or may be converted into probability densities so that the total value of the scores in the 19 directions becomes 1.0. The calculated score is stored in the data storage unit 133.

When the car 122 is in a specific position in the hoistway, the sound source direction estimation unit 135 includes a target sound included in the first sound, which is a sound indicated by the first sound data, and a second sound source direction estimation unit 135. The sound source direction of the target sound included in the second sound indicated by the sound data is estimated.

For example, the sound source direction estimation unit 135 draws a first straight line extending from the microphone array 110 at an angle corresponding to the first selection score, which is one score selected from the plurality of first scores in the horizontal plane. A first locus in which the point where the first straight line contacts the unit sphere centered on the microphone array 110 moves by rotating around an axis orthogonal to the first direction in the horizontal plane, and a plurality of first trajectories in the horizontal plane. A second straight line extending from the microphone array 110 at an angle corresponding to the second selection score, which is one score selected from the two scores, is rotated about an axis orthogonal to the second direction in the horizontal plane. By doing so, the intersection with the second locus where the point where the second straight line contacts the unit sphere moves is set as the sound source direction.

Here, the sound source direction estimation unit 135 may select the maximum value of the plurality of first scores as the first selection score and the maximum value of the plurality of second scores as the second selection score.
However, when the sound source direction estimation unit 135 selects the maximum value of the plurality of first scores as the first selection score and selects the maximum value of the plurality of second scores as the second selection score, the first When the locus of 1 and the second locus do not intersect, the first selection score is selected in order from the one having the largest first score until the first locus and the second locus intersect. May be good.

Further, the sound source direction estimation unit 135 selects the maximum value of the plurality of first scores as the first selection score, and selects the maximum value of the plurality of second scores as the second selection score. When the locus of 1 and the second locus do not intersect, the second locus is selected as the second selection score in order from the one having the largest second score until the first locus and the second locus intersect. May be good.
Here, the sound source direction estimation unit 135 includes an intersection calculation unit 136 and a change unit 137.

The intersection calculation unit 136 estimates the sound source direction by integrating the sound source direction candidates specified from the scores calculated when the elevator car 122 reciprocates in the hoistway a plurality of times. Here, the integration method when the microphone array 110 is installed on the elevator car 122 and reciprocates twice will be described.

It is assumed that the microphone array 110 faces the door direction as shown in FIG. 4 on the first round trip, and faces the right direction where the counterweight 126 exists as shown in FIG. 5 on the second round trip.
Since the sound data for two round trips is synchronized by the synchronization unit 132, for example, when the elevator car 122 passes through a specific place in the hoistway for each of the two round trips, a target sound which is some kind of sound is generated. If so, the target sound is recorded at the same time on the sound data for each of the two round trips.

When the score calculation unit 134 performs the score calculation process at the timing when the target sound is recorded, the score in the direction in which the target sound exists in each sound data is output as shown in FIGS. 6 and 7.

The horizontal axis of FIGS. 6 and 7 is 0 degrees in front of the microphone array 110, −90 degrees to the right, and 90 degrees to the left.
As shown in FIGS. 6 and 7, the scores estimated by the score calculation unit 134 show different transitions because the directions in which the microphone array 110 is facing are different.

8 and 9 are perspective views showing a unit sphere centered on the center point 150 of the microphone array 110.
FIG. 8 shows the direction of the microphone array 110 in the first round trip and the solid arrow 151 indicating the sound source direction candidate.
The intersection calculation unit 136 sets the target direction having the highest score estimated from the sound acquired by the microphone array 110 at the specific time of the first round trip as the sound source direction candidate among the plurality of target directions.

FIG. 9 shows the direction of the microphone array 110 in the second round trip and the solid arrow 152 indicating the sound source direction candidate.
The intersection calculation unit 136 sets the target direction having the highest score estimated from the sound acquired by the microphone array 110 at the specific time of the second round trip among the plurality of target directions as the sound source direction candidate.

In FIG. 8, when the arrow 151 indicating the sound source direction candidate is rotated 360 degrees about the straight line L1 passing through the two microphones constituting the microphone array 110, the locus 153 is obtained. As shown in FIG. 8, the straight line L1 is a straight line orthogonal to the horizontal plane in the direction in which the microphone array 110 faces.
Similarly, in FIG. 9, when the arrow 152 indicating the sound source direction candidate is rotated 360 degrees about the straight line L2 passing through the two microphones constituting the microphone array 110, the locus 154 is obtained. As shown in FIG. 9, the straight line L2 is a straight line orthogonal to the horizontal plane in the direction in which the microphone array 110 faces.
The locus 153 and the locus 154 are directions in which an actual sound source can exist.

The score calculation unit 134 calculates a score used for estimating the sound source direction based on the difference in distance between each microphone of the microphone array 110 and the actual sound source position. The locus 153 and the locus 154 are loci of sound source position candidates in which the difference in distance between the microphone and the sound source position is equal.
Therefore, as shown in FIG. 10, the intersection 155 between the locus 153 and the locus 154 is the sound source position. That is, the direction of the sound source position indicated by the intersection 155 is estimated as the sound source direction.

Here, as shown in FIG. 8, when the locus 153 is projected onto the horizontal plane of the unit sphere including the microphone array 110 from the direction perpendicular to the horizontal plane, the line segment 156 is obtained.
Further, as shown in FIG. 9, when the locus 154 is projected onto the horizontal plane of the unit sphere including the microphone array 110 from the direction perpendicular to the horizontal plane, the line segment 157 is obtained.

When considering three-dimensional Cartesian coordinates with the center point 150 as the origin, the intersection point 158 between the line segment 156 and the line segment 157 is used.
When the coordinates of the intersection point 158 are (x, y), the horizontal angle θ and the elevation angle φ of the intersection point 155, which is the actual sound source position, can be obtained by the following equations (1) and (2).

(1)

(2)

Here, arctan2 is represented by the following equation (3) based on the inverse function tan ^-1 of the tan function, and arccos indicates the inverse function of the cos function.

(3)

The direction indicated by the horizontal angle θ and the elevation angle φ as seen from the center point 150 is the sound source direction, and is the direction of the intersection 155 of the locus 153 and the locus 154.
Here, when the line segment 156 and the line segment 157 intersect, the intersection calculation unit 136 gives the calculation results of the horizontal angle θ and the elevation angle φ to the output unit 141 as the sound source direction. On the other hand, the intersection calculation unit 136 gives information to the change unit 137 that the line segment 156 and the line segment 157 do not intersect when they do not intersect.

The change unit 137 changes the sound source direction candidate when the line segment 156 and the line segment 157 do not intersect in the intersection calculation unit 136.
FIG. 11 is a schematic view showing an example in which the line segment 156 and the line segment 157 do not intersect.
There is reverberation in the hoistway, and there is a possibility that the score calculation unit 134 cannot calculate an accurate score. If the score is not calculated correctly on the first round trip, the second round trip, or both, there is a possibility that the line segment 156 and the line segment 157 do not intersect as shown in FIG. In this case, the intersection calculation unit 136 cannot estimate the sound source direction. Therefore, the change unit 137 changes the sound source direction candidate so that the intersection calculation unit 136 can estimate the sound source direction.

The change unit 137 extracts a plurality of scores calculated for each of the first round trip and the second round trip in descending order, and gives an angle corresponding to the extracted score to the intersection calculation unit 136 as a new sound source direction candidate. The intersection calculation unit 136 calculates the intersection with a new sound source direction candidate.

Specifically, the change unit 137 may set the angle corresponding to the score in descending order of the score as a new sound source direction candidate only for the score of the first round trip. In this case, for the second round trip, the angle corresponding to the highest score is fixed as a sound source direction candidate.
Further, the change unit 137 may set the angle corresponding to the score in descending order of the score as a new sound source direction candidate only for the score of the second round trip. In this case, for the first round trip, the angle corresponding to the highest score is fixed as a sound source direction candidate.
Further, the changing unit 137 may set the corresponding angles as new sound source direction candidates in descending order of the scores for both the score of the first round trip and the score of the second round trip.

As described above, the score of the first round trip calculated by the score calculation unit 134 and 2 by repeating the process of calculating the intersection point by the intersection calculation unit 136 using the new sound source direction candidate specified by the change unit 137. Based on the score of the round trip, the sound source direction can be reliably estimated in the range of the horizontal angle of 0 to 360 degrees and the elevation angle of 0 to 90 degrees.

The input unit 140 receives the input of information necessary for processing in the sound source direction estimation device 130 via the input device 111.
The input information is, for example, in which direction the microphone array 110 is rotated 90 degrees in the second round trip, or the rotation angle of the microphone array 110. The information input by the input unit 140 is stored in the data storage unit 133.

The output unit 141 outputs information indicating the sound source direction to the output device 112. For example, when the output device 112 is a display, the output unit 141 outputs information indicating the sound source direction to the display. As a result, the display displays information indicating the direction of the sound source. Further, for example, when the output device 112 is a speaker, the output unit 141 outputs information indicating the sound source direction to the speaker. As a result, the speaker outputs information indicating the direction of the sound source by voice.

The sound source direction estimation device 130 described above can be realized by the computer 101 shown in FIG.
For example, the acquisition unit 131 can be realized by the I / F 106.
Further, the data storage unit 133 can be realized by the main storage device 103 or the auxiliary storage device 104.

A part or all of the synchronization unit 132, the score calculation unit 134, the intersection calculation unit 136, the change unit 137, the input unit 140, and the output unit 141 read the program stored in the auxiliary storage device 104 into the main storage device 103. , It can be realized by executing the program by the processor 102.

For example, the program executed by the processor 102 is also referred to as a sound source specifying program. For example, the sound source specifying program is recorded on a recording medium. Such a program may be provided through a network, or may be recorded and provided on a recording medium. That is, such a program may be provided, for example, as a program product.
A part or all of the synchronization unit 132, the score calculation unit 134, the intersection calculation unit 136, the change unit 137, the input unit 140, and the output unit 141 may be realized by a processing circuit (not shown).
That is, these may be realized by a processing network.

FIG. 12 is a flowchart showing a process for acquiring sound data in the sound source direction estimation system 100.
The flowchart here assumes that the user has installed the microphone array 110 on the car 122.

First, the microphone array 110 starts recording the sound data of the first round trip (S10).
Then, while the car 122 is moving on the hoistway, the microphone array 110 acquires sound data (S11). Here, the car 122 reciprocates in the hoistway by descending after ascending or ascending after descending. Then, the sound in the hoistway is input to the microphone array 110.

The microphone array 110 finishes recording the sound data of the first round trip (S12). The recorded sound data is input to the sound source direction estimation device 130 via the acquisition unit 131 and stored in the data storage unit 133.

Next, the user rotates the microphone array 110 horizontally by 90 degrees (S13). When the user inputs the rotation angle, the user inputs the rotation direction and the rotation angle (here, 90 degrees) of the microphone array 110 via the input device 111. Such information is acquired by the input unit 140 and stored in the data storage unit 133.

The microphone array 110 starts recording the sound data for the second round trip (S14).
Then, while the car 122 is moving on the hoistway, the microphone array 110 acquires sound data (S15). The car 122 moves in the same direction as in step S11 and at the same speed.

The microphone array 110 ends the recording of the sound data for the second round trip (S16). The recorded sound data is input to the sound source direction estimation device 130 via the acquisition unit 131 and stored in the data storage unit 133.

FIG. 13 is a flowchart showing a process of estimating the sound source direction by the sound source direction estimation device 130 in the first embodiment.
The flowchart here is based on the premise that the sound data for two round trips is stored in the data storage unit 133.

First, the synchronization unit 132 synchronizes the sound data for two round trips (S20).
Next, the score calculation unit 134 calculates a score from each of the synchronized sound data for two round trips (S21).

Next, the intersection calculation unit 136 identifies a sound source direction candidate using the score calculated by the score calculation unit 134 (S22). Here, the intersection calculation unit 136 sets the target direction corresponding to the highest score from the plurality of scores calculated for the plurality of target directions of the first round trip as the sound source direction candidate, and the plurality of second round trips. The target direction corresponding to the highest score from the plurality of scores calculated for the target direction of is set as the sound source direction candidate.

Next, the intersection calculation unit 136 refers to the locus of the two sound source position candidates specified from the two sound source direction candidates with respect to the horizontal plane of the microphone array 110 in the unit sphere centered on the center point 150 of the microphone array 110. The two line segments projected by the above are specified (S23).

Next, the intersection calculation unit 136 determines whether or not there is an intersection in the two line segments specified in step S23 (S24). If the two line segments have an intersection (Yes in S24), the process proceeds to step S25, and if the two line segments do not have an intersection (No in S24), the process proceeds to step S27.

In step S25, the intersection calculation unit 136 estimates the sound source direction from the intersection of the two line segments. The information indicating the sound source direction estimated in this way is given to the output unit 141.
Then, the output unit 141 performs a process of outputting the sound source direction based on the given information (S26).

In step S27, the changing unit 137 changes at least one sound source direction candidate among the two sound source direction candidates. Then, the process returns to step S23. In step S23, the intersection calculation unit 136 identifies two line segments using the changed sound source direction candidates.

According to the first embodiment, the sound source direction estimation device 130 uses a 2ch microphone array 110 to generate a sound source in the range of 0 to 360 degrees in the horizontal direction and 0 to 90 degrees in the elevation direction from the sound data indicating the sound in the hoistway. The direction can be estimated with certainty.

Further, even if there is an error in the calculation of the scores of the first round trip or the second round trip, or both the first round trip and the second round trip, the sound source direction candidate is changed and the first round trip and the second round trip are performed. The final sound source direction can be estimated by integrating the sound source direction candidates.

Embodiment 2.
Next, the second embodiment will be described. In the second embodiment, the matters different from the first embodiment will be mainly described. Then, in the second embodiment, the description of the matters common to the first embodiment will be omitted.

In the first embodiment, when the intersection 158 as shown in FIG. 10 cannot be obtained from the scores for two round trips in the intersection calculation unit 136, the change unit 137 changes the sound source direction candidate. There is. In the second embodiment, a method of estimating the sound source direction by a method different from that of the first embodiment will be described.

As shown in FIG. 1, the sound source direction estimation system 200 according to the second embodiment includes a computer 101 functioning as a sound source direction estimation device, a microphone array 110, an input device 111, an output device 112, and a sensor 113. And prepare.
The sound source direction estimation system 200 according to the second embodiment is different from the sound source direction estimation system 100 according to the first embodiment in the processing by the computer 101.

FIG. 14 is a block diagram showing an example of the sound source direction estimation device according to the second embodiment.
As shown in FIG. 14, the sound source direction estimation device 230 includes an acquisition unit 131, a synchronization unit 132, a data storage unit 133, a score calculation unit 134, a sound source direction estimation unit 235, and an input unit 140. , The output unit 141 is provided.
The acquisition unit 131, synchronization unit 132, data storage unit 133, score calculation unit 134, input unit 140, and output unit 141 of the sound source direction estimation device 230 according to the second embodiment acquire the sound source direction estimation device 130 according to the first embodiment. This is the same as the unit 131, the synchronization unit 132, the data storage unit 133, the score calculation unit 134, the input unit 140, and the output unit 141.

When the car 122 is in a specific position in the hoistway, the sound source direction estimation unit 235 includes a target sound included in the first sound, which is a sound indicated by the first sound data, and a second sound source direction estimation unit 235. The sound source direction of the target sound included in the second sound indicated by the sound data is estimated.

The sound source direction estimation unit 235 selects the maximum value of the plurality of first scores as the first selection score, and selects the maximum value of the plurality of second scores as the second selection score. When the locus and the second locus do not intersect, a plurality of angles in the range from the first direction to 360 ° with respect to the first direction in the horizontal plane and a range in which a plurality of first scores are not calculated are obtained. By complementing from the first score of, a plurality of first complementary scores are calculated. Further, the sound source direction estimation unit 235 has a plurality of positions for each angle in a range from the second direction to 360 ° with respect to the second direction in the horizontal plane and in a range in which a plurality of second scores have not been calculated. By complementing from the score of 2, a plurality of second complementary scores are calculated. Then, the sound source direction estimation unit 235 corrects the angle with respect to the second direction so that the angle with respect to the second direction matches the angle with respect to the first direction, and the plurality of first scores and the plurality of firsts are corrected. The direction in the horizontal plane corresponding to the angle in which the value obtained by averaging the complementary score, the plurality of second scores, and the plurality of second complementary scores for each modified angle is the largest is defined as the sound source direction.

The sound source direction estimation unit 235 includes an intersection calculation unit 136 and an estimation unit 238.
The intersection calculation unit 136 of the sound source direction estimation unit 235 in the second embodiment is the same as the intersection calculation unit 136 of the sound source direction estimation unit 135 in the first embodiment.
However, as shown in FIG. 11, when the line segment 156 and the line segment 157 do not intersect, the intersection calculation unit 136 gives information to the estimation unit 238 that the line segment 156 and the line segment 157 do not intersect.

The estimation unit 238 estimates the sound source direction in the intersection calculation unit 136 when the intersection 158 as shown in FIG. 10 cannot be obtained.
Specifically, the score for the first round trip is [A _-90 , A- ₈₀ , ..., A _{- 10} , _A0 , A10, ..., _A80 , _A90 ], and the score for the second round trip. Is [B _-90 , B- ₈₀ , ..., B _{- 10} , _B0 , B10, ..., _B80 , _B90 ]. Here, for example, "A ₀ " indicates a score of a horizontal angle of 0 degrees for the first round trip. Further, "B ₀ " indicates a score of a horizontal angle of 0 degrees for the second round trip.

For such a score, the estimation unit 238 sets the score of the first round trip to [A ₀ , A- ₁₀ , A- ₂₀ , ..., A- ₆₀ , A- ₇₀ , A _{- 80} , A-. ₉₀ , A _{- 80} , ..., A _{- 10} , _A0 , _A10 , ..., _A80 , _A90 , A80, A70, _A60 , ..., _A20 , _A10 , A [ ₀ ], so that it is folded back at each of A _-90 and _A90 , [A- ₁₈₀ , A- ₁₇₀ , A- ₁₆₀ , ..., A- ₁₃₀ , A _-120 , A _-110 , A. As the score of [ _-100 ], the score of [A ₀ , A- ₁₀ , A- ₂₀ , ..., A- ₆₀ , A- ₇₀ , A- ₈₀ ] is assigned, and the score of [A ₁₀₀ , A ₁₁₀ , A ₁₂₀ , ..., A ₁₆₀ , A ₁₇₀ , A ₁₈₀ ] are assigned the scores of [A ₈₀ , A ₇₀ , A ₆₀ , ..., A ₂₀ , A ₁₀ , A ₀ ]. As a result, the estimation unit 238 expands the score from −90 degrees to 90 degrees to the score from −180 degrees to 180 degrees. That is, the estimation unit 238 complements the score by inverting the score so that the straight line passing through the two microphones is symmetrical.

FIG. 15 is a perspective view schematically showing how the score of the first round trip is inverted as described above.
As shown in FIG. 15, the line SL1 showing the score of 90 ° to −90 ° is symmetrical with respect to the straight line L1 passing through the two microphones constituting the microphone array 110 from −90 ° to −90 °. It is extended like the line SL2 showing a score of 90 °.

In addition, the estimation unit 238 also describes the score for the second round trip as [B ₀ , B- ₁₀ , B- ₂₀ , ..., B- ₆₀ , B- ₇₀ , B- ₈₀ , B _-90 , B- ₈₀ . , ..., B _-10 , B ₀ , B ₁₀ , ..., B ₈₀ , B ₉₀ , B ₈₀ , B ₇₀ , B ₆₀ , ..., B ₂₀ , B ₁₀ , B ₀ ] and so on. , The score from -90 degrees to 90 degrees is expanded to the score from -180 degrees to 180 degrees to perform complementation.

FIG. 16 is a perspective view schematically showing how the score of the second round trip is inverted as described above.
As shown in FIG. 16, the line SL3 showing the score of 90 degrees to −90 degrees is symmetrical with respect to the straight line L2 passing through the two microphones constituting the microphone array 110 from −90 degrees to −90 degrees. It is extended like the line SL4 showing a score of 90 degrees.

Further, the estimation unit 238 shifts the angle of the second round trip by 90 degrees so that the horizontal angle of the score of the second round trip matches the horizontal angle of the score of the first round trip. Although the angle arrangements are different in FIGS. 15 and 16, by performing such processing, as shown in FIG. 17, the angle arrangement corresponding to the score of the second round trip is obtained in FIG. Consistent with the angular arrangement shown in.

18 (A) to 18 (E) are tables showing the relationship between the angle processed as described above and the score.
FIG. 18A is a table showing the relationship between the score and the angle of the first round trip inverted by the estimation unit 238. As shown in FIG. 18A, the score of the first round trip horizontal angle of 90 degrees to −90 degrees is expanded from the horizontal angle of −180 degrees to 170 degrees.

FIG. 18B is a table showing the relationship between the score and the angle of the second round trip inverted by the estimation unit 238. As shown in FIG. 18B, the score of the second round trip horizontal angle of 90 degrees to −90 degrees is expanded from the horizontal angle of −180 degrees to 170 degrees.

FIG. 18C is a table showing the relationship between the score and the angle of the second round trip when the angle shown in FIG. 18B is shifted by 90 degrees.
Then, FIG. 18 (D) is a table in which the relationship between the angle and the score shown in FIG. 18 (C) is rearranged so as to be similar to the arrangement of the angles in FIG. 18 (A).
Further, FIG. 18E shows a table showing the relationship between the angle shown in FIG. 18A and the score of the first round trip, and the angle shown in FIG. 18C and the second round trip. It is a table that summarizes the table showing the relationship with the score.

Here, the estimation unit 238 averages the score of the first round trip and the score of the second round trip for each corresponding angle in FIG. 18 (E). For example, the estimation unit 238 calculates an average value by a synergistic average or an arithmetic mean, and sets the horizontal angle having the highest average value as the horizontal angle in the sound source direction and the elevation angle in the sound source direction as 0 degree.
Based on the above, the estimation unit 238 estimates the sound source direction. The estimation result of the sound source direction is given to the output unit 141.

In this example, the horizontal angle is divided every 10 degrees, but even if the division angle is different, the range of the horizontal angle can be expanded by folding back at -90 degrees and 90 degrees in the same manner. ..

In the second embodiment, by fixing the elevation angle to 0 degrees, the score is calculated for either the first round trip, the second round trip, or both the first round trip and the second round trip, as in the first round trip. Even if there is an error, the sound source direction can be estimated.

Embodiment 3.
Next, the third embodiment will be described. In the third embodiment, the matters different from the first and second embodiments will be mainly described. Then, in the third embodiment, the description of the matters common to the first and second embodiments will be omitted.

In the first embodiment, when the intersection point calculation unit 136 does not obtain the intersection point 158 in FIG. 10 from the scores for two round trips, the change unit 137 changes the sound source direction candidate. On the other hand, in the second embodiment, in such a case, the sound source direction is estimated by processing the score for two round trips. In other words, in the first and second embodiments, the method of estimating the sound source direction is different when the intersection 158 in FIG. 10 cannot be obtained from the scores for two round trips by the intersection calculation unit 136. Here, in the third embodiment, a method of determining which of the sound source directions calculated by the methods described in the first embodiment and the second embodiment is selected in such a case will be described.

As shown in FIG. 1, the sound source direction estimation system 300 according to the third embodiment includes a computer 101 functioning as a sound source direction estimation device, a microphone array 110, an input device 111, an output device 112, and a sensor 113. And prepare.
The sound source direction estimation system 300 according to the third embodiment is different from the sound source direction estimation system 100 according to the first embodiment in the processing by the computer 101.

FIG. 19 is a block diagram showing an example of the sound source direction estimation device according to the third embodiment.
As shown in FIG. 19, the sound source direction estimation device 330 includes an acquisition unit 131, a synchronization unit 132, a data storage unit 133, a score calculation unit 334, a sound source direction estimation unit 335, and an input unit 140. , An output unit 341, a car position estimation unit 342, and a priority order specifying unit 343.
The acquisition unit 131, synchronization unit 132, data storage unit 133, and input unit 140 of the sound source direction estimation device 330 according to the third embodiment are the acquisition unit 131, synchronization unit 132, and data storage of the sound source direction estimation device 130 according to the first embodiment. This is the same as the unit 133 and the input unit 140.

The score calculation unit 334 cuts out a section for performing sound source direction estimation from the sound data for two round trips synchronized by the synchronization unit 132. At this time, the score calculation unit 334 specifies the section information including the start time, the end time, and the cutout length for cutting out the section, and cuts out the section sound data which is the sound data of the section.
The score calculation unit 334 calculates the score in the target direction of the sound acquired by the microphone array 110 for each section sound data cut out. The score calculated here is stored in the data storage unit 133.

The sound source direction estimation unit 335 includes a sound source direction candidate specified from the sound data acquired by the microphone array 110 on the first round trip of the elevator car 122 and sound data acquired by the microphone array 110 on the first round trip of the elevator car 122. Combine with the sound source direction candidates specified from. For example, the sound source direction estimation unit 335 has a sound source direction candidate specified from the score calculated from the sound data acquired by the microphone array 110 in the first round trip of the elevator car 122, and the microphone array 110 is 2 of the elevator car 122. The sound source direction is estimated by integrating with the sound source direction candidate specified from the score calculated from the sound data acquired on the round trip.

The sound source direction estimation unit 335 in the third embodiment both estimates the sound source direction in the first embodiment and estimates the sound source direction in the second embodiment. Therefore, the sound source direction estimation unit 335 includes an intersection calculation unit 136 that performs the same processing as the intersection calculation unit 136 of the first embodiment, and a change unit 137 that performs the same processing as the change unit 137 of the first embodiment. The estimation unit 238 that performs the same processing as the estimation unit 238 of the second embodiment is provided.
Specifically, when the intersection point 158 in FIG. 10 cannot be obtained from the scores for two round trips by the intersection calculation unit 136, the sound source direction candidate is changed by the change unit 137, and the intersection calculation unit 136 is changed. The sound source direction is estimated from the sound source direction candidates. Further, when the intersection point calculation unit 136 does not obtain the intersection point 158 in FIG. 10 from the scores for two round trips, the estimation unit 238 estimates the sound source direction by processing the scores for the two round trips. ..
As a result, two estimation results in the sound source direction can be obtained in the same section. The two sound source direction estimation results are recorded in the data storage unit 133.

The car position estimation unit 342 estimates the car position, which is the position of the car 122 in the hoistway, based on the data from the sensor 113 acquired by the acquisition unit 131.
For example, when the sensor 113 is an acceleration sensor, the car position can be obtained by integrating the acceleration in the vertical direction twice.
Further, when the sensor 113 is an image sensor, the car position can be estimated from the amount of change in the image data at each time.
The existing method may be used for these calculations.

When the sound source direction estimation unit 335 estimates two sound source directions in a certain section, the priority order specifying unit 343 specifies the priority in the two sound source directions. Here, the priority specifying unit 343 is the same as that of the first embodiment by comparing the direction in which the car 122 is advancing with the first sound source direction obtained by the same method as that of the first embodiment. The priority order between the first sound source direction obtained by the method and the second sound source direction obtained by the same method as in the second embodiment is specified.
For example, the priority order specifying unit 343 specifies the priority by comparing the sound source direction in the section where the two sound source directions are estimated with the car position.

First, the priority specifying unit 343 specifies the car moving direction, which is the moving direction of the car in the section where the two sound source directions are estimated, from the car position obtained from the car position estimation unit 342. Since the elevation angle direction of the sound source is obtained in the sound source direction obtained by the same method as in the first embodiment, the priority specifying unit 343 is in a section in which two sound source directions are estimated from the elevation angle direction. Specify the elevation angle change direction, which is the direction in which the elevation angle direction changes.

Then, the priority specifying unit 343 compares the car moving direction with the elevation angle changing direction. Specifically, when the car moving direction is upward and the elevation angle changing direction is also upward, there is no object that moves upward faster than the car, so it is estimated by the same method as in the first embodiment. There is a contradiction in the direction of the sound source. In this case, the priority specifying unit 343 raises the priority of the sound source direction estimated by the same method as in the second embodiment to higher than the priority of the sound source direction estimated by the same method as the first embodiment. To.

Similarly, even when the car moving direction and the elevation angle changing direction are both downward, the priority order specifying unit 343 sets the priority of the sound source direction estimated by the same method as in the second embodiment to the first embodiment. It should be higher than the priority of the sound source direction estimated by the same method.

On the other hand, in the priority order specifying unit 343, when the car moving direction and the elevation angle changing direction are different, in other words, the car moving direction is upward and the elevation angle changing direction is downward, or the car moving direction is downward. When the elevation angle change direction is upward, the priority of the sound source direction estimated by the same method as in the first embodiment is higher than the priority of the sound source direction estimated by the same method as the second embodiment. Also on top.

Then, the priority specifying unit 343 gives the output unit 341 priority information indicating the priority specified as described above.

The output unit 341 outputs information indicating the sound source direction to the output device 112.
In the third embodiment, when the sound source direction estimation unit 335 estimates two sound source directions in a certain section, the sound source direction is such that the priority order specified by the priority specifying unit 343 can be known as the upper sound source direction. Is output to the output device 112. This allows the user to know which sound source direction has the higher priority.

FIG. 20 is a flowchart showing a process of estimating the sound source direction by the sound source direction estimation device 330 in the third embodiment.
The flowchart here is based on the premise that the sound data for two round trips is stored in the data storage unit 133.
First, the synchronization unit 132 synchronizes the sound data for two round trips (S30).

Next, the score calculation unit 334 cuts out the section sound data to be processed in order from the beginning of the synchronized sound data for two round trips (S31).
Next, the score calculation unit 334 calculates the score from the section sound data (S32).

Then, the sound source direction estimation unit 335 estimates the sound source direction using the score calculated by the score calculation unit 334 (S33). Here, the sound source direction estimation unit 335 estimates the sound source direction by the same method as in the first embodiment when the intersection point 158 in FIG. 10 cannot be obtained from the scores for two round trips by the intersection calculation unit 136. , The sound source direction is estimated by the same method as in the second embodiment. The sound source direction estimated by the same method as in the first embodiment is also referred to as a first sound source direction, and the sound source direction estimated by the same method as in the second embodiment is also referred to as a second sound source direction.

Next, the car position estimation unit 342 estimates the car position (S34).
Then, the score calculation unit 334 determines whether or not the section in which the sound source direction is estimated is the last section (S35). For example, the score calculation unit 334 may make such a determination depending on whether or not the section sound data has reached the end time. Then, when the section in which the sound source direction is estimated is the last section (Yes in S35), the process proceeds to step S36. On the other hand, when the section in which the sound source direction is estimated is not the last section (No in S35), the process returns to step S31, and the score calculation unit 334 cuts out the sound data of the next section as the section sound data. , Calculate the score.

In step S36, the priority order specifying unit 343 specifies the priority of the first sound source direction and the second sound source direction for each section in which the first sound source direction and the second sound source direction are estimated.

Then, the output unit 341 outputs information indicating the estimated sound source direction (S37). Here, for the section in which the priority order is specified, the output unit 341 outputs so that the priority order of the first sound source direction and the second sound source direction can be known.

As described above, according to the third embodiment, by specifying the priority order in the sound source direction estimated by a plurality of methods, it is possible to present a more reliable sound source direction estimation result to the user.

Embodiment 4.
Next, the fourth embodiment will be described. In the fourth embodiment, the matters different from the first and second embodiments will be mainly described. Then, in the fourth embodiment, the description of the matters common to the first and second embodiments will be omitted.

In the first to third embodiments, the user inputs the angle and direction in which the microphone array 110 is rotated to the input unit 140. In the fourth embodiment, the orientation of the microphone array 110 is estimated so that the user's input can be omitted.

As shown in FIG. 1, the sound source direction estimation system 400 according to the fourth embodiment includes a computer 101 functioning as a sound source direction estimation device, a microphone array 110, an input device 111, an output device 112, and a sensor 413. And prepare.

The computer 101, microphone array 110, input device 111 and output device 112 of the sound source direction estimation system 400 according to the fourth embodiment are the computer 101, microphone array 110, input device 111 and output of the sound source direction estimation system 100 according to the first embodiment. It is the same as the device 112.

However, in the sound source direction estimation system 400 in the fourth embodiment, the processing by the computer 101 is different from the sound source direction estimation system 100 in the first embodiment.
Further, the sensor 413 in the fourth embodiment may include a geomagnetic sensor or an angular velocity sensor.

FIG. 21 is a block diagram showing an example of the sound source direction estimation device according to the fourth embodiment.
As shown in FIG. 21, the sound source direction estimation device 430 includes an acquisition unit 431, a synchronization unit 132, a data storage unit 133, a score calculation unit 134, a sound source direction estimation unit 135, and an input unit 140. , The output unit 141 and the microphone array direction estimation unit 444 are provided.

The synchronization unit 132, the data storage unit 133, the score calculation unit 134, the sound source direction estimation unit 135, the input unit 140, and the output unit 141 of the sound source direction estimation device 430 according to the fourth embodiment are the sound source direction estimation device 130 according to the first embodiment. This is the same as the synchronization unit 132, the data storage unit 133, the score calculation unit 134, the sound source direction estimation unit 135, the input unit 140, and the output unit 141.
However, the sound source direction estimation unit 135 uses the direction of the microphone array 110 estimated by the microphone array direction estimation unit 444 instead of the angle and direction in which the microphone array 110 input by the input unit 140 is rotated. Estimate.

The acquisition unit 431 acquires sound data and synchronization data as in the first embodiment.
Further, the acquisition unit 431 acquires estimation data for estimating the direction of the microphone array 110. For example, the acquisition unit 431 may acquire data as estimation data from at least one of the geomagnetic sensor, the angular acceleration sensor, and the image sensor included in the sensor 413. Then, the acquired estimation data is stored in the data storage unit 133. The estimation data is acquired every time the car is reciprocated along the hoistway.

The microphone array direction estimation unit 444 is a direction estimation unit that estimates the direction in which the microphone array 110 as a sound collecting unit is facing.
For example, the microphone array direction estimation unit 444 estimates the direction of the microphone array using the estimation data.
First, a method of estimating the orientation of the microphone array 110 when the sensor 413 is an image sensor will be described.

When installing the microphone array 110, for example, the image sensor is fixedly installed on the microphone array 110 so as to face upward in the hoistway. In this state, the image data in the first round trip is acquired. In the second round trip, the user manually rotates the microphone array 110 around 90 degrees. In this state, the image data in the second round trip is acquired.

The microphone array direction estimation unit 444 uses the intercorrelation function between the image shown by the image data of the first round trip and the image shown by the image data of the second round trip, and the image shown by the image data of the first round trip. And the rotation angle with the image shown by the image data of the second round trip can be obtained. As a result, the microphone array direction estimation unit 444 can determine how much the second round trip has rotated in which direction with respect to the first round trip.

When the sensor 413 is an angular acceleration sensor, the actual angle is calculated by integrating the angular acceleration twice when the user rotates and installs the microphone array 110 to which the angular acceleration sensor is fixed. Can be done. Thereby, the microphone array direction estimation unit 444 can obtain the angle and the rotation direction of the first round trip and the second round trip.

Further, when the sensor 413 is a geomagnetic sensor, the microphone array direction estimation unit 444 directs the microphone array 110 from the X-axis magnetic field and the Y-axis magnetic field output from the geomagnetic sensor from the following equation (4). You can ask.

(4)
Here, Y is a magnetic value on the Y axis, and X is a magnetic value on the X axis.

Note that the intersection calculation unit 136 performs the same processing described in the first embodiment. However, the difference between the angle of the first round-trip microphone array and the angle of the second round-trip microphone array 110 is not set to 90 degrees, but the angle difference estimated by the microphone array direction estimation unit 444.

Further, in the intersection calculation unit 136 of the first embodiment, the rotation direction of the second round-trip microphone array 110 when viewed from the first round-trip microphone array 110 is clockwise, but in the fourth embodiment, the microphone array 110 The rotation direction is also set to match the rotation direction estimated by the microphone array direction estimation unit 444.

As described above, according to the fourth embodiment, the user does not need to input the rotation angle of the microphone array 110, and the sound source direction can be estimated more easily.
Further, by calculating the rotation angle of the microphone array 110 by the microphone array direction estimation unit 444, the actual rotation angle when the microphone array 110 is manually rotated can be accurately reflected in the intersection calculation unit 136. , It will be possible to estimate the sound source direction more accurately.

In the fourth embodiment, an example in which the microphone array direction estimation unit 444 is provided in the sound source direction estimation device 130 in the first embodiment is shown, but the fourth embodiment is not limited to such an example.
For example, the sound source direction estimation device 230 in the second embodiment or the sound source direction estimation device 330 in the third embodiment may be provided with the microphone array direction estimation unit 444.

In the above-described embodiments 1 to 4, the angle φ has been described as an elevation angle, but the angle φ may be a depression angle.
Therefore, the sound source direction estimation units 135 to 335 may treat the angle φ calculated as described above as the elevation angle and the depression angle, and may estimate the two upper and lower directions as the sound source direction.
Further, the sound source direction estimation units 135 to 335 may treat the angle φ as a depression angle and estimate the sound source direction.

100, 200, 300, 400 sound source direction estimation system, 101 computer, 110 microphone array, 111 input device, 112 output device, 113 sensor, 120 wall surface, 121 car guide rail, 122 car, 123 car top guide shoe, 124 car bottom Guide shoe, 125 counter weight guide rail, 126 counter weight, 127 counter weight upper guide shoe, 128 counter weight lower guide shoe, 130, 230, 330, 430 sound source direction estimation device, 131, 431 acquisition unit, 132 synchronization unit, 133 Data storage unit, 134,334 score calculation unit, 135,235,335 sound source direction estimation unit, 136 intersection point calculation unit, 137 change unit, 238 estimation unit, 140 input unit, 141 output unit, 342 car position estimation unit, 343 priority Ranking identification unit, 444 microphone array direction estimation unit.

Claims (9)

1. The sound in the hoistway where the elevator car moves for the first time is the first sound data acquired by the sound collector pointed in the first direction on a predetermined plane, and the car moves for the second time. The acquisition unit for acquiring the sound in the hoistway and the second sound data acquired by the sound collecting unit directed in the second direction on the plane, and the acquisition unit.
   When the car is in a specific position in the hoistway, it is included in the first sound, which is the sound indicated by the first sound data, for each angle with respect to the first direction in the plane. A plurality of first scores, which are a plurality of scores indicating the possibility of the sound source direction of the target sound, are calculated from the first sound, and the second score is calculated for each angle with respect to the second direction in the plane. A plurality of second scores, which are a plurality of scores indicating the possibility of the target sound included in the second sound, which is the sound indicated by the sound data, in the direction of the sound source, are calculated from the second sound. Score calculation unit and
   A sound source direction estimation unit that estimates the sound source direction using the first score and the second score is provided.
   The sound source direction estimation unit draws a first straight line extending from the sound collecting unit at an angle corresponding to the first selection score, which is one score selected from the plurality of first scores on the plane. The first locus in which the point where the first straight line comes into contact with the unit sphere centered on the sound collecting portion moves by rotating the axis orthogonal to the first direction on the plane. A second straight line extending from the sound collecting unit at an angle corresponding to the second selection score, which is one score selected from the plurality of second scores on the plane, is drawn on the plane. A sound source characterized in that the intersection with the second locus where the point where the second straight line contacts the unit sphere moves by rotating around an axis orthogonal to the direction of is set as the sound source direction. Direction estimation device.
2. The sound source direction estimation unit selects the maximum value of the plurality of first scores as the first selection score, and selects the maximum value of the plurality of second scores as the second selection score. The sound source direction estimation device according to claim 1.
3. When the sound source direction estimation unit selects the maximum value of the plurality of first scores as the first selection score and selects the maximum value of the plurality of second scores as the second selection score. When the first locus and the second locus do not intersect, the first locus has the highest score until the first locus and the second locus intersect. The sound source direction estimation device according to claim 2, wherein the selection score is selected as 1.
4. When the sound source direction estimation unit selects the maximum value of the plurality of first scores as the first selection score and selects the maximum value of the plurality of second scores as the second selection score. When the first locus and the second locus do not intersect, the first locus has the highest second score until the first locus and the second locus intersect. The sound source direction estimation device according to claim 2, wherein the selection score is selected as 2.
5. When the sound source direction estimation unit selects the maximum value of the plurality of first scores as the first selection score and selects the maximum value of the plurality of second scores as the second selection score. When the first locus and the second locus do not intersect, the plurality of first scores are calculated in the range from the first direction to 360 ° with respect to the first direction on the plane. By complementing from the plurality of first scores for each angle in the range not set, a plurality of first complement scores are calculated, and the second direction to the second direction on the plane. A plurality of second complementary scores are calculated by complementing from the plurality of second scores for each angle in a range up to 360 ° in a range in which the plurality of second scores are not calculated. The angle with respect to the second direction is modified so that the angle with respect to the second direction matches the angle with respect to the first direction, and the plurality of first scores and the plurality of first complementary scores are combined with the plurality of first scores. The direction in the plane corresponding to the angle having the largest value averaged for each of the modified angles of the plurality of second scores and the plurality of second complementary scores is defined as the sound source direction. The sound source direction estimation device according to claim 2, which is characterized.
6. Further provided with a car position estimation unit for estimating the car position, which is the position of the car in the hoistway,
   When the sound source direction estimation unit selects the maximum value of the plurality of first scores as the first selection score and selects the maximum value of the plurality of second scores as the second selection score. When the first locus and the second locus do not intersect, the plurality of first trajectories having the highest score until the first locus and the second locus intersect. The sound source estimated by performing at least one of selecting the first selection score in order and selecting the plurality of second selection scores in order from the one having the largest second selection score. While specifying the first sound source direction which is the direction, the angle in the range from the first direction to 360 ° with respect to the first direction in the plane and in the range where the plurality of first scores are not calculated. Each time, by complementing from the plurality of first scores, a plurality of first complement scores are calculated, and in the range from the second direction to 360 ° with respect to the second direction on the plane. , By complementing from the plurality of second scores for each angle in the range in which the plurality of second scores are not calculated, a plurality of second complementary scores are calculated, and the angles with respect to the second direction are calculated. Corrects the angle with respect to the second direction so as to match the angle with respect to the first direction, and the plurality of first scores and the plurality of first complementary scores and the plurality of second The second sound source direction, which is the sound source direction estimated by the direction in the plane, corresponding to the angle at which the average value of the score and the plurality of second complementary scores for each of the modified angles is the largest. Identify and
   By comparing the direction in which the car is advancing, which is specified by the car position, with the first sound source direction, the priority order between the first sound source direction and the second sound source direction can be determined. The sound source direction estimation device according to claim 2, further comprising a priority specifying unit to be specified.
7. The sound source direction estimation device according to any one of claims 1 to 6, further comprising a direction estimation unit that estimates the direction in which the sound collecting unit is facing.
8. Computer,
   The sound in the hoistway where the elevator car moves for the first time is the first sound data acquired by the sound collector pointed in the first direction on a predetermined plane, and the car moves for the second time. Acquiring unit for acquiring the sound in the hoistway and the second sound data acquired by the sound collecting unit directed in the second direction on the plane.
   When the car is in a specific position in the hoistway, it is included in the first sound, which is the sound indicated by the first sound data, for each angle with respect to the first direction in the plane. A plurality of first scores, which are a plurality of scores indicating the possibility of the sound source direction of the target sound, are calculated from the first sound, and the second score is calculated for each angle with respect to the second direction in the plane. A plurality of second scores, which are a plurality of scores indicating the possibility of the target sound included in the second sound, which is the sound indicated by the sound data, in the direction of the sound source, are calculated from the second sound. Score calculation unit and
   The sound source direction is made to function as a sound source direction estimation unit that estimates using the first score and the second score.
   The sound source direction estimation unit draws a first straight line extending from the sound collecting unit at an angle corresponding to the first selection score, which is one score selected from the plurality of first scores on the plane. A first locus in which the point where the first straight line comes into contact with the unit sphere centered on the sound collecting portion moves by rotating the axis orthogonal to the first direction on the plane. A second straight line extending from the sound collecting unit at an angle corresponding to the second selection score, which is one score selected from the plurality of second scores on the plane, is drawn on the plane. A program characterized in that the intersection with the second locus where the point where the second straight line contacts the unit sphere moves by rotating around an axis orthogonal to the direction of is set as the sound source direction. ..
9. The sound in the hoistway where the elevator car moves for the first time is the first sound data acquired by the sound collector pointed in the first direction on a predetermined plane, and the car moves for the second time. The sound in the hoistway is acquired with the second sound data acquired by the sound collecting unit directed in the second direction on the plane.
   When the car is in a specific position in the hoistway, it is included in the first sound, which is the sound indicated by the first sound data, for each angle with respect to the first direction in the plane. A plurality of first scores, which are a plurality of scores indicating the possibility of the sound source direction of the target sound, are calculated from the first sound.
   A plurality of scores indicating the possibility of the sound source direction of the target sound included in the second sound, which is the sound indicated by the second sound data, for each angle with respect to the second direction on the plane. A plurality of second scores are calculated from the second sound.
   A sound source direction estimation method for estimating the sound source direction using the first score and the second score.
   A first straight line extending from the sound collecting unit at an angle corresponding to the first selection score, which is one score selected from the plurality of first scores in the plane, is drawn from the first straight line in the plane. A first locus in which a point where the first straight line comes into contact with a unit sphere centered on the sound collecting portion moves by rotating the axis orthogonal to the direction of A second straight line, which is a straight line extending from the sound collecting unit at an angle corresponding to the second selection score, which is one score selected from the two scores, is an axis orthogonal to the second direction in the plane. A sound source direction estimation method characterized in that the intersection with a second locus where a point where the second straight line comes into contact with the unit sphere moves by rotating the second straight line is the sound source direction.

Images