close

Вход

Забыли?

вход по аккаунту

?

JP2016114426

код для вставкиСкачать
Patent Translate
Powered by EPO and Google
Notice
This translation is machine-generated. It cannot be guaranteed that it is intelligible, accurate,
complete, reliable or fit for specific purposes. Critical decisions, such as commercially relevant or
financial decisions, should not be based on machine-translation output.
DESCRIPTION JP2016114426
Abstract: The present invention provides a method of accurately estimating the direction of a
sound source from sound pressure signals of sound from sound sources input to a plurality of
microphones installed on a flat plate. SOLUTION: The incident elevation angle φ to the plane
plate 13 is changed at every predetermined angle, and the arrival time difference of sound
between the microphones M1 and M3 and the arrival time difference of sound between the
microphones M2 and M4 are measured. After setting the virtual coordinates with the plane
including the acoustic center of M1 to M4 as the virtual horizontal plane, measure the arrival
time difference of the sound input to the microphones M1 to M4, and set the virtual horizontal
angle θ 'and the virtual elevation angle φ' The virtual horizontal angle θ ′ and the virtual
elevation angle φ ′ are calculated and converted to the horizontal angle θ and the elevation
angle φ of physical coordinates, respectively, and then the elevation angle φ is compared with
the condition angle φ. It is estimated that (θ, φ) having a close elevation angle φ is the
direction of the sound source. [Selected figure] Figure 1
Source direction estimation method
[0001]
The present invention relates to a method of estimating a sound source direction from sound
pressure signals of sound from sound sources input to a plurality of microphones installed on a
flat plate.
[0002]
Conventionally, as shown in FIG. 13, four microphones M1 to M4 are arranged at predetermined
intervals on two straight lines orthogonal to each other, and a fifth microphone M5 is a square
03-05-2019
1
formed by the microphones M1 to M4 as a bottom surface. The sound pressure means of the
sound propagating from the sound source is detected by the sound collecting means disposed at
the position of the apex of the quadrangular pyramid, and the sound source is detected from the
arrival time difference Dij corresponding to the phase difference between the two microphones
(Mi, Mj) Of the horizontal direction θ and elevation angle φ, and an image pickup means such
as a CCD camera is taken to capture an image of the estimated sound source direction, and this
image data and data of the sound source direction are synthesized There is known an apparatus
for creating an image for sound source estimation in which a sound source direction (θ, φ)
estimated in an image and a sound pressure level are displayed in a graphic form (see, for
example, Patent Document 1).
[0003]
In the above-described conventional method, the sound source direction and the size of the
arriving sound can be measured for each frequency, so information of the sound source can be
reliably grasped, but in narrow spaces such as indoors, the influence of the reflected sound from
a wall etc. Because of the large size, arithmetic processing is required to distinguish between
direct sound and reflected sound.
Therefore, if multiple microphones are installed on a flat plate and the sound pressure signal
input to the microphones is limited to only 180 ° in the front, which is the shooting direction of
the camera, the effect of the reflected sound from the back or side wall etc. Can be significantly
reduced, so it can be expected to significantly improve the estimation accuracy of the sound
source direction.
In addition, since the microphone and the camera can be arranged in the same plane, the sound /
image collecting means can be miniaturized.
[0004]
JP, 2011-238985, A
[0005]
However, when the plurality of microphones are installed on a flat plate, the estimated sound
source direction deviates from the actual sound source direction depending on the sound
incident direction due to the influence of reflection and diffraction by the flat plate. There was a
03-05-2019
2
point.
[0006]
The present invention has been made in view of the conventional problems, and provides a
method of accurately estimating the direction of a sound source from sound pressure signals of
sound from sound sources input to a plurality of microphones installed on a flat plate. The
purpose is to
[0007]
The inventors of the present application, as a result of intensive studies, have found that the
cause of the error generated between the sound source direction estimated from the sound
pressure signals of the sounds input to the plurality of microphones installed on the flat plate
and the actual sound source direction is The actual acoustic center of each microphone is offset
from the geometric center (hereinafter referred to as the geometric center) which is the center of
each microphone and the position of the diaphragm, and this offset is Since the relationship
between the incident elevation angle and the arrival time difference between the acoustic centers
is previously obtained, and the sound source direction is estimated using this relationship,
because it depends on the incident elevation angle of the incident angles of the sound to the flat
plate. The present invention has been found to improve the estimation accuracy of the direction
of a sound source.
[0008]
That is, according to the present invention, at least three non-straight lines which are installed on
a flat plate and collect only sound pressure signals of sound propagated from a sound source
forward from one side of a plane including the flat plate. A method of estimating the direction of
the sound source from the time difference of arrival of sound pressure signals input to
microphones, wherein two microphones disposed on two straight lines crossing each other
among the at least three microphones. A virtual coordinate setting step of setting virtual
coordinates with respect to a plurality of incident elevation angles as virtual coordinates in which
a plane including the acoustic centers of three or four microphones constituting the pair is a
virtual horizontal plane, and each of the two microphone pairs Measuring an arrival time
difference of a sound pressure signal input to the microphone and temperature or sound speed;
and a virtual seat set for each of the plurality of incident elevation angles A virtual direction
calculating step of calculating a virtual horizontal angle and a virtual elevation angle, which are
incident directions of sound in the virtual coordinates, using the measured arrival time difference
and the temperature or sound speed, and for each of the incident elevation angle, Converting the
03-05-2019
3
virtual horizontal angle and the virtual elevation angle into a horizontal angle and an elevation
angle of physical coordinates by rotating the virtual coordinates and returning the virtual
coordinates to physical coordinates that are an actual coordinate system; A sound source
direction estimating step of estimating a sound source direction by comparing the converted
elevation angle with the incident elevation angle; and in the virtual coordinate setting step,
changing the incident elevation angle to the flat plate at every predetermined angle, After
measuring the arrival time difference of sound pressure signals input to the microphones
constituting the two microphone pairs, the two microphones are determined from the measured
arrival time difference. A virtual vector in which an axial vector connecting acoustic centers of
microphones constituting a phone pair is calculated, and a plane including the acoustic centers of
microphones constituting two microphone pairs from the calculated two axial vectors is a virtual
horizontal plane It is characterized in that coordinates are set.
As a result, since the sound source direction can be estimated in consideration of the deviation of
the acoustic center due to the sound incident direction, the sound source direction can be
estimated with high accuracy.
[0009]
Further, the virtual coordinate setting step is performed before the measurement step, and the
set virtual coordinate is stored in the storage unit, and when estimating the sound source
direction, the virtual coordinate extracted from the storage unit is used. Since the virtual
horizontal angle and the virtual elevation angle are respectively calculated, the sound source
direction can be efficiently estimated.
Further, according to the present invention, each of the microphones constituting the two
microphone pairs is disposed at each vertex of a square centered on the intersection of two
straight lines intersecting each other, and in the measurement step, one of the squares is Arrival
time difference of sound pressure signal input to each microphone forming the first microphone
pair on diagonal and sound pressure signal input to each microphone forming the second
microphone pair on the other diagonal of the square Measuring the arrival time difference of
As described above, when the number of microphones is four, the sound pressure signal of the
same microphone does not have to be used redundantly for calculation of the arrival time
difference, so that the estimation accuracy of the sound source direction can be further
improved.
03-05-2019
4
[0010]
The summary of the invention does not enumerate all necessary features of the present
invention, and a subcombination of these feature groups can also be an invention.
[0011]
It is a figure which shows the structure of the image display apparatus for sound source
estimation which concerns on this Embodiment.
It is a figure which shows the example of arrangement | positioning of the microphone which
concerns on this Embodiment, and a camera.
It is a figure which shows an example of the image for sound source estimation. Fig. 2 shows the
geometric center and the acoustic center of the microphone; It is a figure which shows the
measuring method of the arrival time difference for calculating | requiring virtual coordinate. It is
a figure which shows the method of calculating the component of axis vector. It is a figure which
shows the relationship between the 1st and 2nd axis vector and virtual coordinates. It is a figure
which shows the calculation method of horizontal angle (theta) 'in virtual coordinates, and
elevation angle (phi)'. It is a figure which shows the method of converting a sound source
direction from a virtual coordinate to a physical coordinate. It is a figure which shows the specific
example of the image for sound source estimation. It is a figure which shows an example of the
plane board which performed the sound absorption process. FIG. 6 shows another configuration
of the sound collecting means according to the present invention. It is a figure which shows the
example of arrangement | positioning of the microphone which comprises the conventional
sound extraction means.
[0012]
FIG. 1 is a diagram showing the configuration of a sound source direction estimation image
display device 1. The sound source direction estimation image creation device 1 includes a sound
/ video sampling unit 10, a sound data input / output unit 21, and a video input / output unit 22.
, Measurement data storage means 23, sound source direction estimation means 24, sound
source estimation image creation means 25, and display means 26. Each of the measurement
03-05-2019
5
data storage means 23 to the sound source estimation image creation means 25 is composed of,
for example, software of a personal computer and a memory, and the display means 26 is
composed of, for example, a display device such as a liquid crystal display. The sound / image
collecting unit 10 is equipped with a sound collecting means 11 provided with four microphones
M1 to M4, a CCD camera (hereinafter referred to as a camera) 12 as a photographing means,
microphones M1 to M4 and a camera 12 And a support base 14 for supporting the plane board
13 and a support leg (tripod) 15 for supporting the support base 14. In this example, as shown in
FIG. 2A, the flat plate 13 has a disk shape, the camera 12 is mounted at the center thereof, and
the microphones M1 to M4 are arranged at each vertex of a square centered on the camera 12
did. Further, as shown in FIG. 2B, the microphones M1 to M4 were mounted such that the
vibrating film surface was at substantially the same position as one surface (hereinafter referred
to as the front surface 13a) of the plane plate 13. A direction from the rear surface 13b of the
flat plate 13 toward the front surface 13a will be referred to as a front direction, as indicated by
the rear surface 13b, and the arrow in the (b) view on the other surface of the flat plate 13. The
camera 12 is mounted on the flat plate 13 such that the shooting direction is the forward
direction.
[0013]
Here, coordinates where the plate surface of the flat plate 13 is the x p y p plane, the direction
perpendicular to the plate surface is the z p axis direction, the center of the camera 12 is the
origin, and the y p axis is the vertical direction (hereinafter referred to as physical coordinates
Set). The coordinates of the position P1 of the microphone M1 in the physical coordinates are (0,
L / 2, 0), and the coordinates of the position P2 of the microphone M2 are (L / 2, 0, 0). The
coordinates of the position P3 of the microphone M3 are (0, -L / 2, 0), and the coordinates of the
position P4 of the microphone M4 are (-L / 2, 0, 0). As described above, in this example, the
microphones M1 to M4 are disposed in the order of M1 to M4 clockwise as viewed from the
front (front side), and the microphone M1 and the microphone M3 constituting the microphone
pair (M1, M3) And the distance between the microphone M2 and the microphone M4
constituting the microphone pair (M2, M4) are both L. When coordinates viewed from the
camera 12 are optical coordinates (real space coordinates), the y p axis of physical coordinates in
the vertical direction is the z axis of optical coordinates, and the z p axis of physical coordinates
in the imaging direction is optical coordinates. The x axis of the physical coordinates is the x axis
of the optical coordinates. Here, as the microphones M1 to M4, any of omnidirectional and
unidirectional microphones may be used. Also, the microphones M1 to M4 may be generally
used small microphones or thin microphones such as surface microphones. Further, as the flat
plate 13, it is preferable to use a material having high rigidity and totally reflecting sound.
Moreover, it is preferable to make the front surface 13a a flat surface without unevenness.
03-05-2019
6
[0014]
The sound data input / output means 21 includes an amplifier 21a and an A / D converter 21b.
The amplifier 21a includes a low pass filter, removes high frequency noise components from the
sound pressure signal of the sound sampled by the microphones M1 to M4, amplifies the sound
pressure signal, and outputs the amplified signal to the A / D converter 21b. The A / D converter
21 b A / D converts the sound pressure signal, and sends the A / D converted sound pressure
signal to the measurement data storage unit 23 as sound pressure waveform data. The video
input / output unit 22 inputs a video signal continuously photographed by the camera 12 and
performs A / D conversion, and sends the A / D converted video signal to the measurement data
storage unit 23 as image data. The measurement data storage means 23 stores sound pressure
waveform data and image data. The sound source direction estimation unit 24 includes a virtual
coordinate storage unit 24a, a virtual sound source direction calculation unit 24b, a sound source
direction conversion unit 24c, and a sound source direction estimation unit 24d. The sound
pressure signal stored in the measurement data storage unit 23 While estimating the horizontal
angle θ p and elevation angle φ p which are the sound source direction when viewed from the
plane plate 13 for each frequency, the sound pressure level of the sound transmitted from the
sound source is measured. The virtual coordinate storage unit 24a stores virtual coordinates with
the acoustic centers P1 'to P4' of the microphones M1 to M4 as virtual horizontal planes (x'y
'planes). The virtual sound source direction calculation unit 24b uses the sound arrival time
difference M13 between the microphone M1 and the microphone M3 and the sound arrival time
difference M24 between the microphone M2 and the microphone M4 to generate the sound
source direction (horizontal angle θ ′, elevation angle Calculate φ ′). The sound source
direction conversion unit 24 c converts the sound source direction in the virtual coordinates into
the sound source direction (horizontal angle θ p, elevation angle φ p) in the physical
coordinates by rotating each coordinate axis of the virtual coordinates. The virtual coordinates,
the sound source direction in the virtual coordinates, and the sound source direction in the
physical coordinates are set or determined for each incident elevation angle φ in. The sound
source direction estimating unit 24 selects the most probable sound source direction by
comparing the elevation angle φ p obtained for each incident elevation angle φ in with the
incident elevation angle φ in as a conditional angle, and estimates this as the estimated sound
source direction (Horizontal angle θ p, elevation angle φ p). The details of the method of
estimating the horizontal angle θ p and the elevation angle φ p will be described later.
[0015]
The sound source estimation image creation means 25 is data of the sound source direction
03-05-2019
7
when the horizontal angle θ p and the elevation angle φ p which are data of the sound source
direction estimated by the sound source direction estimation means 24 are viewed from the
camera (horizontal And the data of the horizontal angle θ and the elevation angle φ are
converted to the angle θ and the elevation angle φ), and the image data as shown in the upper
diagram of FIG. A sound source estimation image G in which a plurality of figures (here, a circle)
indicating the direction of the sound source is drawn is created and output to the display means
26. The sound source estimation image G is displayed on the display screen 26 M of the display
means 26. The horizontal axis of the upper diagram of FIG. 3 is the horizontal angle θ, and the
vertical axis is the elevation angle φ. The central coordinates of the circle C j in the figure are (θ
j, φ j), the magnitude of the circle C j represents the sound pressure level, and the pattern
represents the frequency. The lower side of FIG. 3 shows the distribution of sound pressure
levels, where the horizontal axis is the horizontal angle θ and the vertical axis is the sound
pressure level (dB). In order to estimate the sound source, as long as a circle C displaying the
sound source direction, the frequency and the size of the sound is displayed on the image,
coordinate conversion is not necessary.
[0016]
Next, a method of setting virtual coordinates stored in the virtual coordinate storage unit 24a,
and details of operations of the virtual sound source direction calculation unit 24b, the sound
source direction conversion unit 24c, and the sound source direction estimation unit 24d will be
described. Usually, the acoustic center of the microphone Mi has been considered as the center of
the microphone Mi and the position Pi of the diaphragm, but in fact it is apparently affected by
the incident angle of sound because it is affected by reflection and diffraction near the
microphone . As shown in FIG. 4, assuming that the position Pi of the center of the microphone
Mi and the diaphragm is the geometric center of the microphone and the position of the actual
acoustic center is the acoustic center Pi ', the microphone pair actually measured (M1, M3) The
arrival time difference M13 between the microphone M1 and the microphone M3 constituting
the second embodiment, and the arrival time difference M24 between the microphone M2 and
the microphone M4 constituting the microphone pair (M2, M4) are microphone Mi (i = 1 It is an
arrival time difference when a sound enters into acoustic center Pi 'of -4). The arrival time
difference M ij obtains the cross spectrum K ij (f) of the signal input to the pair of microphones
Mi and the microphone M j, and further, using the phase angle information Ψ (rad) of the target
frequency f, It is calculated by equation 1]. In this example, the arrival time difference P13
between the microphone M1 and the microphone M3 and the arrival time difference P24
between the microphone M2 and the microphone M4 are measured for each incident elevation
angle φ in of the sound incident on the flat plate 13, and the acoustic center P1 is obtained. Set
virtual coordinates with the plane including 'P4' as the virtual horizontal plane. The virtual
coordinates are set for each incident elevation angle of sound φ in. In order to distinguish the
03-05-2019
8
arrival time difference for setting the virtual coordinates from the arrival time difference M13
and the arrival time difference M24 actually measured, they were P13 and P24, respectively.
[0017]
Next, a method of setting virtual coordinates will be described. First, as shown in FIG. 5, the
speaker 30 is installed on the flat plate 13 as a sound source, and the sound pressure signal of
the sound generated by the speaker 30 is collected by the microphones M1 to M4 to obtain the
space between the microphones M1 and M3. The arrival time difference P13 and the arrival time
difference P24 between the microphones M2 and M4 are measured in advance. Specifically, the
incident horizontal angle θ in is fixed at 0 °, the incident elevation angle φ in is changed from
0 ° to 180 ° at 10 ° pitch, the arrival time difference P13 is measured, and the incident
horizontal angle θ in is 90 The arrival time difference P24 is measured while changing the
incident elevation angle φ in from 0 ° to 180 ° at a pitch of 10 ° while fixing to °. The
incident horizontal angle θ in is limited to 0 ° and 90 ° because the acoustic center of the
microphone Mi is found not to depend on the incident horizontal angle θ in but to depend only
on the incident elevation angle φ in. Also, when the direction in which the microphone pair is
connected is orthogonal to the incident horizontal angle θ in, it is difficult to specify the phase
difference or the arrival time difference. However, in order to obtain more detailed information,
other incident horizontal angles θ in may also be measured.
[0018]
Here, let (x1, y1, z1) be a component of a first axis vector P13 'which is a vector starting from the
sound center P1' of the microphone M1 and ending at the sound center P3 'of the microphone
M3. A component of a second axis vector P24 'which is a vector starting from P2' and ending at
the acoustic center P4 'of the microphone M4 is set as (x2, y2, z2). As shown in FIG. 6, each
component (x1, y1, z1) of the first axis vector P13 'is between the microphone M1 and the
microphone M3 when the incident angle of sound is (0 °, φ in). It can be expressed as the
following formula [Equation 2] using the arrival time difference P13 ′ and the measured value
RP13 ′ of arrival time difference when the incident angle is (0 °, 180 ° −φ in) and the
incident elevation angle φ in. . By the way, in the arrival time difference P13 between the
microphone M1 and the microphone M3 actually measured, the arrival time difference P13 ′
between the acoustic center P1 ′ and the acoustic center P3 ′ and the characteristics of the
microphone, amplifier, filter, etc. And an error Δ13 that occurs in That is, P13 = P13 '+ Δ13.
Similarly, RP13 = RP13 '+ [Delta] 13. Therefore, since P13'-RP13 '= P13-RP13, in this example,
when calculating each component (x1, y1, z1) of the first axis vector P13', the arrival time
03-05-2019
9
difference of the above equation [Equation 1] Instead of P13 ′ and the arrival time difference
RP13 ′, actual arrival time differences P13 and RP13 are used.
[0019]
Further, each component (x2, y2, z2) of the second axis vector P24 'is also incident on the actual
value P24 of the arrival time difference at (90 °, φ in) as in the case of the first axis vector P13,
Using an actual measurement value RP24 of arrival time difference when the angle is (90 °, 180
° −φ in) and the incident elevation angle φ in, the following equation [Equation 3] can be
expressed. The component of the first axis vector P13 'and the component of the second axis
vector P24' are determined for each of the incident elevation angles φ in. Hereinafter, the
incident elevation angle φ in is referred to as a “condition angle”.
[0020]
Next, as shown in FIG. 7, assuming that the acoustic centers P1 ′ to P4 ′ of the microphones
M1 to M4 are on the same plane, this is taken as a virtual horizontal plane (x′y ′ plane), and a
direction perpendicular to this plane Consider virtual coordinates with the virtual vertical axis (z
'axis) as. In the present example, the origin O 'of the imaginary coordinates is the intersection
when the first axis vector P13' and the second axis vector P24 'intersect at the midpoint of each
vector. The direction from the acoustic center P3 'of the microphone M3 to the acoustic center
P1' of the microphone M1 is y 'axis, and the direction perpendicular to the y' axis in the x 'y'
plane is x 'axis. By performing such an operation for each condition angle (φ ink), virtual
coordinates can be obtained for all the condition angles φ ink. In this example, since the incident
elevation angle φ in is changed from 0 ° to 180 ° at a pitch of 10 °, k = 1 to 19. The axial
angle γ, which is the angle between the direction from the acoustic center P4 ′ of the
microphone M4 to the microphone M2, and the y ′ axis, uses the component of the vector P13
′ and the component of the vector P24 ′ to obtain the following equation It can be expressed
as [Equation 4].
[0021]
As described above, the virtual coordinate storage unit 24a changes the incident elevation angle
φ in to the flat plate 13 for each predetermined angle (here, 10 °) while forming the
microphones M1 and M3 constituting the microphone pair (M1 and M3). Of the arrival time
03-05-2019
10
difference P13, RP13 of the sound pressure signal input to the input signal and the arrival time
difference P24, RP24 of the sound pressure signal input to the microphones M2, M4 constituting
the microphone pair (M2, M4) From the arrival time difference, a first axis vector P13
'connecting the acoustic centers P1' and P3 'of the microphone pair (M1, M3) and a second axis
vector connecting the acoustic centers P2' and P4 'of the microphone pair (M2, M4) P24 'is
calculated, and the acoustic centers P1'-P of the microphones M1-M4 are calculated from these
first and second axis vectors P13' and P24 '. The plane containing the 'a virtual horizontal plane,
and stores the γ virtual coordinates and the axial angle of each "Condition angle phi in". The
setting of the virtual coordinates is performed in advance before the sound pressure signal of the
sound from the sound source is acquired. The set virtual coordinates are stored in the virtual
coordinate storage unit 24a.
[0022]
In the virtual sound source direction calculation unit 24b, the arrival time difference M13 of the
sound between the microphone M1 and the microphone M3 is the arrival time difference
between the acoustic centers P1 ′ and P3 ′ on virtual coordinates stored in the virtual
coordinate storage unit 24a. Assuming that the arrival time difference M24 of the sound between
M2 and the microphone M4 is the arrival time difference between the acoustic centers P2 ′ and
P4 ′ on virtual coordinates, the horizontal angle θ ′ and elevation angle φ ′ on the virtual
coordinates are calculated. The arrival time difference M13 and the arrival time difference M24
can be obtained by frequency analysis of the sound pressure waveform data of the microphones
M1 to M4 stored in the measurement data storage unit 23 by FFT. The arrival time difference
M13 and the arrival time difference M24 are obtained for each frequency f. That is, as shown in
FIG. 8, assuming that the sound is incident from the horizontal angle θ ′ and the elevation
angle φ ′, the relationship between the arrival time difference M13 of the sound between the
microphone M1 and the microphone M3 and the vector P13 ′ and the microphone M2 The
relationship between the difference in arrival time M24 of the sound of the microphone and the
microphone M4 and the vector P24 'can be expressed as the following equation [Equation 5].
Here, the horizontal angle θ ′ and the elevation angle φ ′, which are the incident angles of
the sound in virtual coordinates, are the arrival time difference M13 of the sound between the
microphone M1 and the microphone M3, and the sound between the microphone M2 and the
microphone M4. It is calculated from the arrival time difference M23, the component of the
vector P13 'when the conditional angle obtained in advance is φ in, and the component of the
vector P24'. Here, in the equation [6], c 0 is the velocity of sound measured when virtual
coordinates are set, and c is the velocity of sound at the time of sound source estimation (at the
time of arrival time difference measurement). In place of the speed of sound, the temperature
may be measured, and the speed of sound may be determined from the measured temperature.
03-05-2019
11
[0023]
The sound source direction conversion unit 24 c converts the horizontal angle θ ′ and the
elevation angle φ ′ at virtual coordinates into the horizontal angle θ p and the elevation angle
φ p at physical coordinates. Note that the origin of virtual coordinates and the origin of physical
coordinates are assumed to be the same point. If the incident direction of the sound in the virtual
coordinates is represented by the coordinates (x <'> 3, y <'> 3, 3, z <'> 3) of a unit vector having a
length of 1, it is as follows. x <'> 3 = sin .theta.' cos .phi. 'y <'> 3 = cos .theta. 'cos .phi.'z <'> 3 = sin
.phi. Next, convert the incident direction vector (x <'> 3, y <'> 3, z <'> 3) of the sound in the virtual
coordinates into the sound incident vector (x3, y3, z3) at the physical coordinates Describe the
procedure to First, the coordinates (0, L13, 0) of the first axis vector P13 'in virtual coordinates
and the coordinates (L24 · sin γ, L24 · cos γ, 0) of the second axis vector P24 ′ are
respectively calculated in the first The coordinates (x1, y1, z1) of the axis vector P13 'and the
coordinates (x2, y2, z2) of the second axis vector P24' are converted. In order to convert (x1, y1,
z1) to (0, L13, 0), as shown in FIG. 9A, the x and y axes of the physical coordinates are rotated
around the z axis by η. , (X1, y1, z1) are converted to (0, (x1 <2> + y1 <2>) <1/2>, z1). Here, η =
tan <−1> (x1 / y1). At this time, (x2, y2, z2) is converted to (xa2, ya2, za2). Next, as shown in
FIG. 9B, the x-axis is rotated by ε, and (0, (x1 <2> + y1 <2>) <1/2>, z1) is (0, L13, Convert to 0).
Here, ε = tan <−1> {z1 / (x1 <2> + y1 <2>) <1/2>}. At this time, (xa2, ya2, za2) is converted to
(xb2, yb2, zb2). Next, as shown in FIG. 9C, the y-axis is rotated by δ to convert (xb2, yb2, zb2)
into (L24 sin γ, L24 cos γ, 0).
Here, δ = tan <−1> (zb2 / xb2). In order to convert the incident direction vector of sound in
virtual coordinates (x <'> 3, y <'> 3, z <'> 3) into the sound incident vector (x3, y3, z3) in physical
coordinates, You can do the opposite conversion. That is, first, the y <'> axis is rotated by-[delta]
to convert (x'> 3, y <'> 3 and z <'> 3) into (xa3, ya3, za3). (Xa3, ya3, za3) becomes like the
following equation [Equation 7]. Next, the x <'> axis is rotated by -ε to convert (xa3, ya3, za3)
into (xb3, yb3, zb3). (Xb3, yb3, zb3) becomes like the following equation [Equation 8]. Finally, the
z <'> axis is rotated by-[eta] to convert (xb3, yb3, zb3) into sound incident vector (x3, y3, z3) at
physical coordinates. (X3, y3, z3) becomes like the following equation [Equation 9]. Therefore, as
shown in FIG. 9D, the horizontal angle θ k and the elevation angle φ k in physical coordinates at
the condition angle φ ink can be obtained by the following equation [Equation 10]. By
performing such an operation for each condition angle (φ in), the horizontal angle θ k and the
elevation angle φ k in physical coordinates can be determined for all the condition angles φ ink.
[0024]
03-05-2019
12
The sound source direction estimation unit 24 compares the elevation angle φ ink obtained for
each incident elevation angle φ in with the incident elevation angle φ in as a condition angle. As
the value of the calculated elevation angle φ ink is closer to the conditional angle φ ink, the
horizontal angle θ ink and the elevation angle φ ink should be closer to the actual horizontal
angle θ and the elevation angle φ. Therefore, the absolute value | φ ink −φ in | of the
difference between the calculated elevation angle φ ink and the condition angle φ in is
determined for each condition angle φ in, and the determined absolute value | φ ink −φ in | is
the smallest. The sound source direction can be accurately estimated by setting the horizontal
angle θ ink and the elevation angle φ ink as the direction of arrival of the sound (horizontal
angle θ p, elevation angle φ p).
[0025]
In this way, after measuring the arrival time differences M13 and M24 of the sound pressure
signals input to the microphones M1 to M4 and the sound speed c at the time of measurement,
in virtual coordinates set for each of a plurality of condition angles (incident elevation angle)
φink The virtual horizontal angle θ ′ and the virtual elevation angle φ ′, which are the
incident directions of sound in virtual coordinates, are calculated using the measured arrival time
differences M13 and M24 and the sound velocity c, respectively, and the virtual angle is
calculated for each condition angle φ ink. By rotating the coordinates and returning the virtual
coordinates to the physical coordinates that are the actual coordinate system, the virtual
horizontal angle θ ′ and the virtual elevation angle φ ′ are converted into the horizontal
angle θ k and the elevation angle φ k of the physical coordinates, respectively. Since the set of
the horizontal angle θ k and the elevation angle φ k having the elevation angle φ k close to the
conditional angle φ ink is the sound source direction, the sound source direction is determined.
It can be estimated accurately.
[0026]
EXAMPLE FIG. 10 (a) shows the difference in arrival time M13 of the sound between the
microphone M1 and the microphone M3 and the arrival time difference M24 between the sound
of the microphone M2 and the microphone M4 using the following equation [Equation 11]. It is
an image for sound source estimation which drew the calculated sound source direction.
FIG. 10 (b) is an image for sound source estimation according to the present invention, and FIG.
10 (c) is a conventional four microphones M1 to M4 arranged at predetermined intervals on two
03-05-2019
13
straight lines orthogonal to each other. It is an image for sound source estimation in which the
sound source direction is drawn by the sound collecting means in which the fifth microphone M5
is arranged at the apex of a quadrangular pyramid whose bottom surface is a square formed by
the microphones M1 to M4. As apparent from FIG. 10, the sound source direction estimated by
the sound source direction estimation method of the present invention is substantially the same
as the sound source direction estimated using the five conventional microphones M1 to M5, and
The distribution is also equivalent. Thereby, if the sound source direction estimation method of
the present invention is used, the sound source direction can be estimated in consideration of the
deviation of the sound center due to the sound incident direction. It was confirmed that it could
be estimated.
[0027]
As mentioned above, although this invention was demonstrated using embodiment, the technical
scope of this invention is not limited to the range as described in the said embodiment. It is
obvious to those skilled in the art that various changes or modifications can be added to the
above embodiment. It is also apparent from the scope of the claims that the embodiments added
with such alterations or improvements can be included in the technical scope of the present
invention.
[0028]
For example, in the above embodiment, a material having high rigidity and totally reflecting
sound is used as the flat plate 13. However, as shown in FIG. 11A, the windproof screen 31 is
disposed in front of the flat plate 13. Alternatively, as shown in FIG. 11B, if the sound absorbing
process such as disposing the sound absorbing material 32 around the flat plate 13 in addition to
the windproof screen 31, the influence of the reflection can be further reduced. As the windproof
screen 31, for example, a mesh material such as urethane foam can be used. Further, as the
sound absorbing material 32, a non-woven fabric in a ring shape or the like may be used. The
windproof screen 31 may be made of non-woven fabric. In the above embodiment, the sound
source direction is estimated using the sound collecting means 11 provided with the four
microphones M1 to M4 arranged on the flat plate 13. However, on the two straight lines crossing
the sound collecting means The microphones M1 to M3 may be arranged at predetermined
intervals. The microphones M1, M2, and M3 may be disposed at each vertex of an equilateral
triangle having a length of L as shown in FIG. 12 (a), or two as shown in FIG. 12 (b). It may be
arranged at each vertex of the equilateral triangle.
03-05-2019
14
[0029]
DESCRIPTION OF SYMBOLS 1 Image display device for sound source direction estimation, 10
sound / image collecting unit, 11 sound collecting means, 12 photographing means (camera), 13
flat plate, 13 a flat plate front surface, 13 b flat plate rear surface, 14 support stand, 15 support
Leg, 21 sound data input / output means, 21a amplifier, 21b A / D converter, 22 image input /
output means, 23 measurement data storage means, 24 sound source direction estimation
means, 24a virtual coordinate storage unit, 24b virtual sound source direction calculation unit,
24c sound source direction conversion unit, 24d sound source direction estimation unit, 25
sound source estimation image creation means, 26 display means, 26M display screen, M1 to M4
microphones.
03-05-2019
15
Документ
Категория
Без категории
Просмотров
0
Размер файла
29 Кб
Теги
jp2016114426
1/--страниц
Пожаловаться на содержимое документа