JP2011232238A - Sound source direction estimation device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a sound source direction estimation device which can estimate a direction of a sound source even when a non-directional microphone is used.SOLUTION: In a sound source direction estimation device 20, a sound generated by a sound source located outside a first housing 10a is collected in the housing 10a through each of apertures 11 to 14. Each of microphones 1 to 4 embedded in the first housing 10a converts the sound propagated into the first housing 10a to an electric signal. A high-pass filter (HPF) 21 passes a frequency component of 2.5 kHz or higher out of the electric signal outputted by each of the microphones 1 to 4. The frequency component of a bandwidth having a high correlation with an angle of the position of the sound source is extracted. A sound source direction calculation unit 23 estimates the direction of the sound source based on the frequency component having passed through the HPF 21. Accordingly, even when the microphones 1 to 4 are non-directional, the sound source direction estimation device 20 can estimate the direction of the sound source.

Description

  The present invention relates to a sound source direction estimating apparatus.

  2. Description of the Related Art Conventionally, as shown in Patent Document 1 below, an electronic device incorporating a directional microphone that can collect sound on the front side and the back side is known. In this device, the microphone is held such that the front surface of the microphone comes into contact with the inner surface of the front cabinet (housing) in which the first opening is formed. A space portion is formed, and a second opening is formed in the portion of the front cabinet that faces the space portion.

  With such a configuration, in the above-described device, the first opening allows the outside of the housing to communicate with the front side of the microphone, and the second opening connects the outside of the housing and the space on the back side of the microphone. Communicate. Then, sound generated outside the front cabinet is collected on the front side of the microphone through the first opening, and collected on the back side of the microphone through the second opening, thereby improving the directivity of the microphone. Yes.

Japanese Patent Laid-Open No. 10-56681

  However, the above-described device using a directional microphone has a structure in which sound is collected on the front side and the back side of the microphone, so it is necessary to form a space on the back side of the microphone, and there are significant structural restrictions. . Therefore, in order to secure the space on the back side of the microphone, the casing is enlarged or the microphone is installed at a position protruding from the casing. For example, the microphone is directed toward the sound source used in the video conference system. When applied to a camera, there is a problem that it is difficult to make the camera compact. Therefore, in order to reduce such structural constraints, a technique that can estimate the direction of a sound source even when an omnidirectional microphone is used has been desired.

  An object of this invention is to provide the sound source direction estimation apparatus which can estimate the direction of a sound source, even when an omnidirectional microphone is used.

  The sound source direction estimating device according to the present invention is a case in which an opening for capturing sound generated by an external sound source is formed, and is vibrated by sound embedded in the opening of the case and propagating through the opening into the case. A microphone that vibrates a plate to convert sound into an electrical signal, a high-pass filter that passes a frequency component of a predetermined frequency or higher in the audible range of the electrical signal output from the microphone, and a frequency component that passes through the high-pass filter And a sound source direction estimating unit that estimates the direction of the sound source based on the sound source.

  According to the sound source direction estimating apparatus according to the present invention, sound generated by a sound source outside the housing is taken into the housing through the opening, and the sound propagated into the housing is converted into an electric signal by the microphone embedded in the housing. Is done. Here, the sound generated at a position shifted by a certain angle with respect to the direction line connecting the opening and the microphone diffracts near the opening and reaches the microphone. Although sound is attenuated by this diffraction, the angle at which the sound source is located and the sound attenuation amount have a correlation according to the frequency of the sound. That is, for a sound with a low frequency, the amount of attenuation is small even when the angle is large, but for a sound with a high frequency, the amount of attenuation increases as the angle increases. Therefore, a frequency component in a band having a high correlation with the angle at which the sound source is located can be extracted by passing a frequency component of a predetermined frequency or higher in the electrical signal output from the microphone through the high-pass filter. Since the direction of the sound source is estimated by the sound source direction estimation unit based on the frequency component that has passed through the high-pass filter, the direction of the sound source can be estimated even if the microphone itself is omnidirectional. Furthermore, since the microphone needs to collect sound only on the front side facing the opening, it is not necessary to provide a space for collecting sound on the back side of the microphone, and the space inside the enclosure is arranged with wiring and other parts. As a result, the housing can be made compact. Further, since the microphone is embedded in the housing, the appearance can be improved as compared with the case where the microphone is installed at a position protruding from the housing.

  Here, the casing is provided with a cylindrical body that forms a propagation path extending from the opening toward the casing, and the microphone is disposed at the end of the cylindrical body. It is preferable that it is larger than the diaphragm.

  In this case, since the cylindrical body is provided, the sound propagates from the opening to the microphone through the propagation path having a larger cross section than the diaphragm of the microphone, so that the sound is reliably transmitted to the diaphragm and the sound source The direction can be estimated with higher accuracy.

  The shortest distance from the opening to the microphone is preferably larger than the width of the opening.

  If the shortest distance from the opening to the microphone is smaller than the width of the opening, high-frequency sound may reach the microphone directly without being diffracted. If the shortest distance from the opening to the microphone is larger than the width of the opening, high-frequency sound is prevented from reaching the microphone directly, and sound attenuated by diffraction near the opening reaches the microphone. The effect of estimating the direction is more suitably exhibited.

  According to the present invention, the direction of a sound source can be estimated even when an omnidirectional microphone is used.

1 is a perspective view of a video conference camera to which a sound source direction estimating device according to an embodiment of the present invention is applied. 2A is a cross-sectional view of the position including the opening in the video conference camera of FIG. 1, and FIG. 2B is a side view of the cylindrical body in FIG. 2A viewed from the axial direction. . It is a block diagram which shows the function structure of the camera for video conferences of FIG. It is a schematic diagram which shows the angle which the direction of a sound source makes with respect to the axis line of a cylindrical body. FIG. 5A is a diagram showing the position of the sound source with respect to the video conference camera, and FIG. 5B is a diagram showing the output characteristics of the microphone in the case of FIG.

  Hereinafter, a sound source direction estimating apparatus according to an embodiment of the present invention will be described with reference to the drawings. In the following description, a case where the sound source direction estimating device is applied to a video conference camera will be described.

  As shown in FIGS. 1 to 3, the video conference camera A is installed at a position surrounded by, for example, a plurality of participants participating in the video conference, and speaks (hereinafter referred to as “speaker”). The video of the speaker is acquired by directing the camera 35 in that direction. The video conference camera A can communicate with the other party's video conference system via a network.

  The video conference camera A has a substantially cylindrical shape with a diameter of about 10 cm placed on a table or the like, and a resin-made first housing 10a, and an axis (not shown) arranged in the vertical direction along the axis of the first housing 10a. And a second housing 10b made of resin having a substantially rectangular parallelepiped shape that is rotatable about the axis of the first housing 10a. The camera 35 is housed in the second housing 10b.

  On the side surface of the first housing 10a, four circular openings 11 to 14 are formed at the same height and spaced apart by 90 ° in the circumferential direction. Each opening 11-14 is a hole for taking in the voice of the speaker uttered around the video conference camera A into the first housing 10a, and has an inner diameter of about 8 mm.

  Four cylindrical bodies 16 to 19 having a cylindrical shape extending in the radial direction of the first casing 10a from the openings 11 to 14 into the first casing 10a are fixed to the inner surface side of the first casing 10a. . The inner diameter of each cylindrical body 16-19 is about 8 mm, and the inner peripheral surface of each opening 11-14 and the inner wall surface of each cylindrical body 16-19 are substantially flush. Each cylindrical body 16-19 forms propagation paths B1-B4 on a cylindrical shape having a diameter of about 8 mm inside thereof. That is, each propagation path B1-B4 is formed so that it may extend horizontally in the radial direction of the 1st housing | casing 10a toward the inside of the 1st housing | casing 10a from each opening 11-14. In addition, each opening 11-14 means the hole formed in the outer wall of the 1st housing 10a, and each cylindrical body 16-19 and each propagation path B1-B4 are included in each opening 11-14. I can't.

  Furthermore, four cylindrical microphones (hereinafter referred to as “hereinafter referred to as“ microphones ”) are formed at the ends of the cylindrical bodies 16 to 19 located on the center side of the first casing 10a so that the front faces are parallel to the openings 11 to 14, respectively. 1-4) are embedded. Each of the microphones 1 to 4 is an omnidirectional electric condenser microphone having a diameter of about 7 mm and a thickness of about 4 mm and capable of collecting sound only on the front side. Each of the microphones 1 to 4 incorporates diaphragms 1a to 4a made of a circular diaphragm having a diameter of about 5 mm. The shortest distance from each opening 11-14 to the front surface of each microphone 1-4 is about 10 mm, which is longer than the inner diameter of each opening 11-14. Moreover, as above-mentioned, the internal diameter (cross section) of propagation path B1-B4 is larger than the diaphragms 1a-4a (refer FIG. 2). Thereby, the sound passing through the propagation paths B1 to B4 is surely transmitted to the diaphragms 1a to 4a. A spacer (not shown) is preferably disposed between the microphones 1 to 4 and the inner wall surfaces of the cylindrical bodies 16 to 19.

  With such a configuration, the video conference camera A has a mechanism in which the voice of the speaker in the conference is input to the microphones 1 to 4 through the openings 11 to 14 and the propagation paths B1 to B4. Then, the diaphragms 1a to 4a are vibrated by the sound input to the microphones 1 to 4, and the sound is converted into an electrical signal according to the vibration, and the electrical signal is output from the microphones 1 to 4 to the amplifiers 5 to 8. (See FIG. 3). In addition, it is the structure which the sound outside the 1st housing 10a does not propagate to the internal space S of the back side of the microphones 1-4.

  As shown in FIG. 3, the video conference camera A has a function of estimating the direction of the speaker by performing a predetermined process based on the sound input to the microphones 1 to 4. Specifically, the video conference camera A receives the electric signals output from the microphones 1 to 4 and amplifies the input electric signals by the four amplifiers 5 to 8 and the amplifiers 5 to 8. A high-pass filter (hereinafter referred to as “HPF”) 21 that passes a frequency component equal to or higher than a predetermined frequency in the received electrical signal, and an AD conversion unit 22 that performs analog-digital conversion on the frequency component that has passed through the HPF 21. A sound source direction calculation unit (sound source direction estimation unit) 23 that estimates the direction of the sound source based on the frequency component digitally converted by the AD conversion unit 22, a histogram registration unit 24, and a sound source direction recalculation unit 25. Yes.

  The amplifiers 5 to 8 are connected to the microphones 1 to 4, respectively, and amplify the electrical signals output from the microphones 1 to 4 by an amplifier circuit. Each amplifier 5-8 may have an adjustment knob for adjusting the amplification degree of an electric signal. Each of the amplifiers 5 to 8 outputs the amplified electric signal to the HPF 21.

  The HPF 21 receives the electrical signal output from each of the amplifiers 5 to 8, blocks the frequency component of, for example, less than 2.5 kHz, and passes the frequency component of 2.5 kHz or more from the input electrical signal. The cutoff frequency in the HPF 21 is a predetermined frequency within the audible range. The HPF 21 may have a function of identifying which one of the amplifiers 5 to 8 is a frequency component output, or a plurality of HPFs 21 may be provided for each of the amplifiers 5 to 8.

  The AD converter 22 converts the frequency component that has passed through the HPF 21 from an analog signal to a digital signal. The AD conversion unit 22 generates a frequency component converted into a digital signal, and outputs the generated frequency component to the sound source direction calculation unit 23. The AD converter 22 may have a function of identifying which one of the amplifiers 5 to 8 is a frequency component output. When a plurality of HPFs 21 are provided for each of the amplifiers 5 to 8, A plurality of amplifiers 5 to 8 may be provided.

  The sound source direction calculation unit 23, the histogram registration unit 24, and the sound source direction recalculation unit 25 are configured by, for example, a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), and the like. The sound source direction calculation unit 23, the histogram registration unit 24, and the sound source direction recalculation unit 25 estimate the speaker direction by calculating the speaker direction based on the frequency component output from the AD conversion unit 22. .

  Specifically, the sound source direction calculation unit 23 compares the output values (sound volume) of frequency components corresponding to the microphones 1 to 4 based on the frequency components output from the AD conversion unit 22. The direction of the sound source from which the sound is emitted is calculated. As the calculation processing of the direction of the sound source in the sound source direction calculation unit 23, a known technique can be used. The direction of the sound source in the video conference camera A is positive in the clockwise direction when viewed from above the first housing 10a with reference to a predetermined position in the circumferential direction of the first housing 10a (for example, the center position of the opening 11). It is expressed as an angle (0 ° to 360 °). The calculation of the direction of the sound source by the sound source direction calculation unit 23 is executed every predetermined time, for example, every 40 milliseconds.

  Here, in the video conference camera A of the present embodiment, since the microphones 1 to 4 are embedded in the first casing 10a through the openings 11 to 14 at a predetermined length (for example, about 10 mm), As the frequency increases, the sound reaching the microphones 1 to 4 is attenuated by diffraction. For example, as shown in FIG. 4, if the direction in which the sound source is located is shifted by an angle θ with respect to the direction line connecting the center of the microphone 1 and the center of the opening 11 (that is, the axis of the cylindrical body 16), Sound that is high is attenuated by the angle θ shift and reaches the microphone 1. Since the frequency component of 2.5 kHz or more is passed by the HPF 21, an output value affected by the angle is input to the sound source direction calculation unit 23, and the sound source is compared by comparing the output values corresponding to the microphones 1 to 4. The direction calculation unit 23 calculates the direction of the sound source with high accuracy. The sound source direction calculation unit 23 outputs the calculated direction of the sound source every predetermined time to the histogram registration unit 24.

  The histogram registration unit 24 inputs the direction of the sound source output from the sound source direction calculation unit 23 and sequentially stores the input direction of the sound source. In storing the direction of the sound source by the histogram registration unit 24, the oldest data is always overwritten with the latest data. Then, the histogram registration unit 24 classifies the stored sound source directions into 20-degree angle ranges in increments of 20 °, and generates a histogram according to the classification result. The histogram registration unit 24 outputs the generated histogram to the sound source direction recalculation unit 25.

  The sound source direction recalculation unit 25 receives the histogram output from the histogram registration unit 24, and converts the frequency in the angle range where the frequency is maximum in the input histogram and the frequency in the angle range near the angle range. The direction of the sound source is recalculated by obtaining the average value of the angle range based on it. The sound source direction recalculation unit 25 outputs the recalculated sound source direction to the motor control unit 31 as the speaker direction.

  In this way, the sound source direction calculation process is performed by the sound source direction calculation unit 23, the histogram registration unit 24, and the sound source direction recalculation unit 25, and the direction of the speaker is estimated. As shown in FIGS. 1 and 3, in the video conference camera A, the first casing 10 a, openings 11 to 14, cylindrical bodies 16 to 19, microphones 1 to 4, amplifiers 5 to 8, HPF 21, AD conversion unit 22. The sound source direction estimating unit 20 includes a sound source direction calculating unit 23, a histogram registration unit 24, and a sound source direction recalculating unit 25.

  Further, as shown in FIG. 3, the video conference camera A has a function of directing the camera 35 in the direction of the sound source estimated by the sound source direction estimating device 20 and acquiring the video of the speaker by the camera 35. Specifically, the video conference camera A includes a motor control unit 31 that controls the motor driving unit 32 based on the direction of the sound source output from the sound source direction recalculation unit 25, and a motor controlled by the motor control unit 31. A motor drive unit 32 that drives the motor 33, a motor 33 that is driven by the motor drive unit 32 to apply a rotational force to the camera rotation mechanism 34, and is connected to the motor 33 to rotate the camera 35 together with the second casing 10b by rotation. A camera rotation mechanism 34, a camera 35 that acquires video, a video processing unit 36 that performs video processing on video data acquired from the camera 35, and a data transfer unit that transfers video data processed by the video processing unit 36 37 and a network communication unit 38 that transmits the video data transferred by the data transfer unit 37 to the network.

  Next, the operation of the video conference camera A will be described. In the following description, as shown in FIG. 5A, a case where a voice is uttered from a speaker located at 45 ° opposite to the openings 13 and 14 at an equal distance from the openings 11 and 12 is used. This will be described as an example. As shown in FIG. 5B, the output values of the frequency components corresponding to the microphones 1 to 4 have a difference in peak value between 2.5 kHz and 10 kHz. More specifically, the output values seen between 2.5 kHz and 10 kHz are the same for the microphones 1 and 2, and are attenuated more than the microphones 1 and 2 for the microphones 3 and 4. On the other hand, the output values seen below 2.5 kHz are substantially equal for the microphones 1 to 4.

  In the sound source direction estimating apparatus 20, the frequency component of less than 2.5 kHz is blocked by the HPF 21, and the frequency component of 2.5 kHz or more is passed. In the case of the example shown in FIG. 5B, the direction of the sound source is calculated based on the output value seen between 2.5 kHz and 10 kHz that is greatly affected by diffraction, and the sound source is at a position of 45 °. Presumed. The sound source direction estimating device 20 outputs a signal indicating 45 ° to the motor control unit 31.

  The motor control unit 31 inputs a signal indicating 45 ° output from the sound source direction estimating device 20, controls the motor driving unit 32 to drive the motor 33, and the camera 35 faces in the 45 ° direction. The camera 35 is rotated together with the second casing 10b by the camera rotation mechanism 34. Note that the turning control of the camera 35 by the motor control unit 31 is executed using the average value of the direction of the sound source within the certain range when the direction of the sound source within the certain range is input continuously for a predetermined time. May be.

  The camera 35 acquires a front video and generates video data indicating the acquired video. The camera 35 outputs the generated video data to the video processing unit 36. The video processing unit 36 receives the video data output from the camera 35, performs video processing on the input video data, and outputs the video data after the video processing to the data transfer unit 37. The data transfer unit 37 receives the video data output from the video processing unit 36 and transfers the input video data to the network communication unit 38. Then, the network communication unit 38 transmits the video data transferred from the data transfer unit 37 to the other party's video conference system via the network.

  According to the sound source direction estimating device 20 of the present embodiment, the sound generated by the sound source outside the first housing 10a is taken into the first housing 10a through the openings 11 to 14 and embedded in the first housing 10a. The sound propagated into the first housing 10a is converted into an electric signal by the microphones 1 to 4. And the frequency component of the band which has a high correlation with the angle where a sound source is located is extracted because the frequency component of 2.5 kHz or more passes through HPF 21 among the electric signals output from each microphone 1-4. Since the direction of the sound source is estimated by the sound source direction calculation unit 23 based on the frequency component that has passed through the HPF 21, the direction of the sound source can be estimated even if each of the microphones 1 to 4 is omnidirectional. . Furthermore, since each microphone 1-4 should collect sound only on the front side directed to each opening 11-14, there is no need to provide a space for collecting sound on the back side of each microphone 1-4. The internal space S in the first housing 10a can be used for the arrangement of wiring and other components, and as a result, the first housing 10a can be made compact. Moreover, since each microphone 1-4 is embed | buried in the 1st housing 10a, the improvement of an external appearance is achieved compared with the case where a microphone is installed in the position protruded from the housing.

  Furthermore, by providing the cylindrical bodies 16 to 19, the microphones 1 to 4 pass from the openings 11 to 14 through the propagation paths B1 to B4 having a larger cross section than the diaphragms 1a to 4a of the microphones 1 to 4, respectively. Since the sound is transmitted to the sound, this sound is reliably transmitted to each of the diaphragms 1a to 4a, and the direction of the sound source can be estimated with higher accuracy.

  Moreover, since the shortest distance from each opening 11-14 to each microphone 1-4 is larger than the internal diameter of each opening 11-14, it is prevented that the sound with a high frequency reaches each microphone 1-4 directly, The sound attenuated by the diffraction in the vicinity of each opening 11-14 reaches each microphone 1-4, and the above-described effect of estimating the direction of the sound source is more suitably exhibited.

  The preferred embodiment of the present invention has been described above, but the present invention is not limited to the above embodiment.

  For example, in the above-described embodiment, the cylindrical bodies 16 to 19 are arranged in the radial direction of the first casing 10a, and the propagation paths B1 to B4 extend horizontally. However, the cylindrical bodies 16 to 19 are You may arrange | position so that it may incline upward or downward toward each opening 11-14 from the center side of the 1 housing 10a. According to such a configuration, even when the video conference camera is installed at a position lower or higher than the position of the head of the participant of the conference, the voice of the speaker is placed in the first housing 10a. It can be taken in effectively.

  Moreover, although the cylindrical bodies 16-19 demonstrated the case where it was a cylindrical shape in the said embodiment, a rectangular tube shape may be sufficient. In this case, the propagation path has a prismatic shape. Moreover, although the microphones 1-4 demonstrated the case where it was cylindrical shape, a rectangular parallelepiped shape may be sufficient.

  Moreover, although the said embodiment demonstrated the case where the four microphones 1-4 were provided, the installation number of microphones may be 1-3, and may be five or more. For example, when the volume of sound generated around the sound source direction estimation device is assumed in advance, the direction of the sound source is estimated based on the output value of the frequency component of a predetermined frequency or more with one microphone. Can do.

  Further, the cutoff frequency in the HPF 21 can be set as appropriate within a range within the audible range according to the specification of the microphone. In addition to the HPF 21, a low-pass filter that passes a frequency component that is lower than the cutoff frequency of the HPF 21 (for example, 7 kHz or less) among frequency components that have passed through the HPF 21 may be provided. In this case, a band pass filter is configured by the HPF 21 and the low pass filter.

  Furthermore, in the above-described embodiment, the case where the sound source direction estimating device 20 is applied to the video conference camera A has been described. However, the sound source direction estimating device 20 is a device that includes a microphone and estimates the direction of the sound source and uses the estimated sound source direction. If there is, it is not limited to this. For example, the sound source direction estimating apparatus of the present invention can be applied to a monitoring camera or the like that detects abnormal sound and reflects the direction of the sound source in which the abnormal sound has occurred.

  DESCRIPTION OF SYMBOLS 1-4 ... Microphone, 1a-4a ... Diaphragm, 10a ... Housing, 11-14 ... Opening, 16-19 ... Cylindrical body, 20 ... Sound source direction estimation apparatus, 21 ... HPF, 23 ... Sound source direction calculating part (sound source Direction estimation unit), B1 to B4... Propagation path.

Claims (3)

A housing in which an opening for capturing sound generated by an external sound source is formed;
A microphone that is embedded inside the opening of the housing and oscillates a diaphragm by the sound that propagates through the opening and into the housing, and converts the sound into an electrical signal;
A high-pass filter that passes a frequency component of a predetermined frequency or higher in the audible range of the electrical signal output from the microphone;
A sound source direction estimation unit that estimates the direction of the sound source based on the frequency component that has passed through the high-pass filter;
A sound source direction estimating apparatus comprising: The casing is provided with a cylindrical body that forms a propagation path extending from the opening toward the casing,
The microphone is disposed at an end of the cylindrical body;
The sound source direction estimating apparatus according to claim 1, wherein a cross section of the propagation path is larger than the diaphragm of the microphone.   3. The sound source direction estimating apparatus according to claim 1, wherein a shortest distance from the opening to the microphone is larger than a width of the opening.

Images