close

Вход

Забыли?

вход по аккаунту

?

JP2006246310

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 JP2006246310
PROBLEM TO BE SOLVED: To move a virtual sound source smoothly while reducing the number
of filter coefficients. SOLUTION: A speaker array 10 is configured by arranging a plurality of
speakers SP1 to SPm. An audio signal in which a predetermined transfer function is convoluted is
supplied to the speakers SP1 to SPm, respectively, to perform wavefront synthesis of the sound
output from the speakers SP1 to SPm. The virtual sound source VSS is formed in the area ABC
including the speaker array 10 by this wavefront synthesis. At this time, the audio signal is
controlled such that the virtual sound source VSS is dense at the vicinity of the speaker array 10
and is located at the intersection of unequally spaced meshes that are rough at a distance.
[Selected figure] Figure 2
Method and apparatus for reproducing audio signal
[0001]
The present invention relates to an audio signal reproduction method and apparatus.
[0002]
In a general stereo reproduction apparatus, a virtual sound source is formed on a line connecting
left and right speakers, and it is perceived as if sound is being output from this virtual sound
source (see, for example, Patent Document 1).
[0003]
03-05-2019
1
On the other hand, in the reproduction apparatus using the wavefront synthesis technique, the
virtual sound source can be disposed at a free position.
Hereinafter, this wavefront synthesis will be described in order.
[0004]
[1] Reproduction of sound field Now, as shown in FIG. 6, it is assumed that a closed surface S that
encloses a space of an arbitrary shape is assumed, and that no sound source is included inside
this closed surface S. .
Then, with respect to the inner space and the outer space of the closed surface S, p (ri): sound
pressure at an arbitrary point ri in the inner space p (rj): sound pressure at an arbitrary point rj
on the closed surface S ds: a point rj Small area including n n: Normal to small area ds at point rj
un (rj): Particle velocity in normal direction n at point rj ω: Angular frequency of audio signal::
Density of air v: Speed of sound (= 340 m / s ) Assuming that k: ω / v, the Kirchhoff's integral
formula is shown by equation (1) in FIG.
[0005]
If the sound pressure p (rj) of the point rj on the closed surface S and the particle velocity un (rj)
in the direction of the normal n at the point rj can be appropriately controlled, the inside of the
closed surface S It means that the sound field of space can be reproduced.
[0006]
Therefore, for example, as shown in FIG. 8A, it is assumed that a sound source SS is disposed on
the left side and a closed curved surface SR (shown by a broken line) covering a spherical space
of radius R is disposed on the right.
Then, the sound field generated in the internal space of the closed surface SR by the sound
source SS can be reproduced without the sound source SS if the sound pressure and particle
velocity un (rj) on the closed surface SR are controlled as described above is there.
03-05-2019
2
[0007]
At this time, a virtual sound source VSS is generated at the position of the sound source SS. That
is, if the sound pressure and particle velocity on the closed surface SR are appropriately
controlled, the listener inside the closed surface SR perceives the sound as if the virtual sound
source VSS exists at the position of the sound source SS.
[0008]
Next, when the radius R of the closed surface SR is made infinite, as shown by the solid line in
FIG. 8A, the closed surface SR becomes a plane SSR. Also in this case, the sound field generated in
the inner space of the closed surface SR by the sound source SS, that is, on the right side of the
plane SSR is reproduced even without the sound source SS by controlling the sound pressure and
particle velocity on the plane SSR. Is possible. Also at this time, the virtual sound source VSS is
generated at the position of the sound source SS.
[0009]
That is, by appropriately controlling the sound pressure and particle velocity at all points on the
plane SSR, the virtual sound source VSS can be disposed on the left side of the plane SSR, and the
sound field can be disposed on the right side. It can be a listening area.
[0010]
In practice, as shown in FIG. 8B, the plane SSR may have a finite width, and the sound pressure
and particle velocity of the finite points CP1 to CPx on the plane SSR may be controlled.
In the following, points CP1 to CPx at which sound pressure and particle velocity are controlled
on the plane SSR will be referred to as "control points".
[0011]
03-05-2019
3
[2] Control of sound pressure and particle velocity at control points CP1 to CPx To control the
sound pressure and particle velocity at control points CP1 to CPx, as also shown in FIG. 9, (A)
Sound source side of plane SSR The plural speakers SP1 to SPm are arranged, for example, in
parallel with the plane SSR. The speakers SP <b> 1 to SPm constitute the speaker array 10. (B)
Control the sound pressure and the particle velocity at the control points CP1 to CPx by
controlling the audio signals supplied to the speakers SP1 to SPm. とすればよい。
[0012]
In this way, the sound waves output from the speakers SP1 to SPm are wave-field synthesized
and act as if sound waves are output from the virtual sound source VSS, and a desired sound field
can be formed. In addition, since the position where the sound wave output from the speakers
SP1 to SPm is subjected to the wavefront synthesis is the plane SSR, the plane SSR is hereinafter
referred to as a "wavefront synthesis plane".
[0013]
[3] State of Wavefront Synthesis FIG. 10 shows an example of the state of wavefront synthesis by
computer simulation. The processing content and processing method of the audio signal supplied
to the speakers SP1 to SPm will be described later, but in this example, each value is set as
follows. Number of loudspeakers: 16 Speaker spacing: 10 cm diameter of loudspeaker: 8 cm φ
position of control point: 10 cm from the speaker to the number of listeners Number of control
points: 116 points in a row at 1.3 cm spacing Virtual sound source position: listening 1m in front
of the area (in the case of FIG. 10A) 3m in front of the listening area (in the case of FIG. 10B)
Width of the listening area: 2.9m (front and back direction) x 4m (left and right direction) : Sound
velocity (= 340 m / s) fhi: reproduction upper limit frequency [Hz] If fhi = v / (2w) (2) Therefore,
it is preferable to narrow the interval w of the speakers SP1 to SPm (m = 16), and for that
purpose, it is necessary to reduce the diameter of the speakers SP1 to SPm.
[0014]
When the audio signals supplied to the speakers SP1 to SPm are digitally processed, the interval
between the control points CP1 to CPx is 1/4 of the wavelength corresponding to the sampling
frequency in order to remove the influence of the sampling. It is preferable to use less than or
equal to 1/5. In the above numerical example, since the sampling frequency is 8 kHz, the
03-05-2019
4
distance between the control points CP1 to CPx is 1.3 cm as described above.
[0015]
Then, according to FIG. 10, the sound waves output from the speakers SP1 to SPm are wave-field
synthesized as if they were the sound waves output from the virtual sound source VSS, and clear
ripples are drawn in the listening area. That is, it can be seen that wavefront synthesis is
appropriately performed to form the target virtual sound source VSS and the sound field.
[0016]
From another viewpoint, in the case of FIG. 10A, since the curvature of the ripples is small, the
position of the virtual sound source VSS is relatively close to the wavefront synthesis surface SSR.
However, in the case of FIG. 10B, since the curvature of the ripples is larger than in the case of
FIG. 10A, the position of the virtual sound source VSS is farther from the wavefront synthesis
surface SSR than in the case of FIG. 10A. Therefore, it can be understood from this also that the
position of the virtual sound source VSS can be controlled.
[0017]
In FIG. 11, an area ABC theoretically indicates a range in which the virtual sound source VSS can
be arranged or moved. The arrangement possible area ABC is an inverted triangle in which each
side is slightly bulged outward, but in the drawings and the following description, it is an
inverted triangle for the sake of simplicity.
[0018]
And this arrangeable area ABC is located on the front (listener side) of the speakers SP1 to SPm
while the vertex PA is located on the line orthogonal to the arrangement direction at the center
of the arrangement of the speakers SP1 to SPm. Further, the vertex PB is located on the
extension of the straight line passing from the vertex PA to the speaker SP1, and the vertex PC is
located on the extension of the straight line passing from the vertex PA to the speaker SPm.
Furthermore, the side connecting the vertex PB and the vertex PC is parallel to the arrangement
03-05-2019
5
direction of the speakers SP1 to SPm.
[0019]
Then, within the locatable area ABC, the virtual sound source VSS can be arranged at an arbitrary
position, or the position of the virtual sound source VSS can be freely moved as shown as a locus
CSS.
[0020]
[4] Algorithm of Wavefront Synthesis As shown in FIG. 12, u (ω): output signal of virtual sound
source VSS, that is, original audio signal H (ω): signal u (ω) to realize appropriate wavefront
synthesis. Transfer function C (ω): transfer function from the speakers SP1 to SPm to the control
points CP1 to CPm q (ω): assuming that the signal is actually reproduced at the control points
CP1 to CPx by wavefront synthesis, the original audio signal u Since a signal obtained by
convolving transfer functions C (ω) and H (ω) with (ω) is the reproduced audio signal q (ω), q
(ω) = C (ω) · H (ω) · u It becomes (ω).
In this case, the transfer function C (ω) can be defined by obtaining the transfer characteristics
from the speakers SP1 to SPm to the control points CP1 to CPx.
[0021]
Then, if the transfer function H (ω) is controlled, appropriate wave-field synthesis is realized by
the reproduced audio signal q (ω) at this time, and the position or movement of the virtual sound
source VSS is realized as shown for example in FIG. Can be controlled arbitrarily.
[0022]
[5] Signal Conversion When the original audio signal u (ω) is converted to the reproduced audio
signal q (ω) according to the above [4] to control the position of the virtual sound source VSS,
the reproducing apparatus It can be configured as shown.
That is, the digital filters DF1 to DFm are configured by FIR filters, and filter coefficients (tap
coefficients) of the digital filters DF1 to DFm are prepared in the coefficient memory 21 of the
03-05-2019
6
control circuit 20. Then, the filter coefficients are read from the coefficient memory 21 and
loaded into the digital filters DF1 to DFm, and transfer functions H (ω) and C (ω) required for
the digital filters DF1 to DFm are set.
[0023]
Then, the original audio signal (digital signal) u (ω) is supplied from the input terminal Tin to the
digital low pass filter DF0, and in the principle of wavefront synthesis, an unnecessary high
frequency component, ie, the frequency fhi shown by equation (2) The above frequency
components are removed.
[0024]
This filter output is supplied to the digital filters DF1 to DFm, and the transfer function H (ω) and
the transfer function C (ω) are convoluted, and reproduced audio signals q (ω) to q (ω) from the
digital filters DF1 to DFm. Is taken out.
Then, the signals q (ω) to q (ω) are D / A converted by the D / A converter circuits DA1 to DAm
and then supplied to the speakers SP1 to SPm through the output amplifiers PA1 to PAm.
[0025]
Therefore, the virtual sound source VSS is formed by the outputs of the speakers SP1 to SPm,
and at this time, the transfer functions C (ω) and H (ω) of the digital filters DF1 to DFm are set
to predetermined values. You can change the position.
[0026]
[6] Arrangement or Movement Position of Virtual Sound Source VSS In the case where the virtual
sound source VSS is arranged in the arrangeable area ABC or in the case where the virtual sound
source VSS is moved in the arrangeable area ABC, a method as shown in FIG. It is done.
[0027]
That is, as shown by a broken line in FIG. 14, the arrangeable area ABC of the virtual sound
source VSS is divided in a mesh shape by a grid parallel to the arrangement direction of the
03-05-2019
7
speakers SP1 to SPm and a grid orthogonal thereto.
Then, the filter coefficient when the virtual sound source VSS is located at the intersection of the
mesh is calculated in advance and stored in the coefficient memory 21.
Then, when moving the virtual sound source VSS, the filter coefficient of the intersection closest
to the movement position is read out from the coefficient memory 21 and loaded to the digital
filters DF1 to DFm.
[0028]
In this way, the virtual sound source VSS can be arranged or moved at each intersection point.
Then, at this time, the filter coefficients at the respective intersections can be used for the
plurality of virtual sound sources VSS.
[0029]
For example, the following are known as prior art documents. Japanese Patent Publication No.
2002-505058
[0030]
By the way, in order to realize the placement and movement of the virtual sound source VSS as
described above, the number of taps of about 1024 taps is usually required for each of the digital
filters DF1 to DFm. Therefore, for example, as shown in FIG. 14, when the locatable area ABC is
divided by a grid of 100 points × 100 points in the left and right direction and in the front and
back direction, the number of intersection points is 5000 points (= 100 × 100/2).
[0031]
As a result, assuming that the number of speakers SP1 to SPm is 16 (m = 16), the coefficient
memory 21 for 1024 taps × 16 × 5000 points = 81920000 taps is required for the filter
coefficient (tap coefficient). It will be necessary.
03-05-2019
8
[0032]
Then, assuming that each of the filter coefficients is a 32-bit (4-byte) float type, a coefficient
memory 21 of 81920000 taps × 4 bytes = 327680000 bytes = 312.5 M bytes is required.
[0033]
That is, a considerable capacity is required for the coefficient memory 21.
Then, the capacity of the coefficient memory 21 increases in proportion to the number m of the
speakers SP1 to SPm and the number of intersection points of the mesh.
Also, since it is necessary to load the next filter coefficient to the digital filters DF1 to DFm as the
virtual sound source VSS moves, when the interval of the mesh intersection points is narrow, the
frequency of loading of the filter coefficient becomes high, and the control circuit 20 The burden
on the
[0034]
Furthermore, when changing the position of the virtual sound source VSS, reading processing of
a large amount of filter coefficients occurs in the digital filters DF1 to DFm, so the data line of the
filter coefficients between the control circuit 21 and the digital filters DF1 to DFm In addition,
considerable transfer capability is required.
[0035]
However, if the number of speakers SP1 to SPm is reduced or the distance between mesh
intersections is increased, the virtual sound source VSS can not be properly formed or the
arrangement and movement of the virtual sound source VSS become unnatural. I will.
[0036]
Also, in practice, the digital filters DF1 to DFm are usually realized by one DSP or CPU.
03-05-2019
9
Then, in that case, (A) The tap coefficient of the a-th tap is fetched for a digital filter among the
digital filters DF1 to DFm.
(B) Perform product-sum operation. (C) Let the result be the output of the ath tap. (D) The
processes of (A) to (C) are sequentially performed for each tap. Therefore, the processes of (A) to
(C) are repeated by the number of taps. (E) The process of (D) is performed for each of the digital
filters DF1 to DFm. Therefore, the process (D) is performed 16 times (= m times). To perform
processing like
[0037]
Therefore, assuming that the number of taps of each of the digital filters DF1 to DFm is 1024, the
DSP or CPU processes the (A) to (C) by 1024 taps for one sample of the original audio signal u
(ω). It will be executed for 16 × 16 = 16 times.
[0038]
Since m digital filters DF1 to DFm are required for one virtual sound source VSS, N sets of digital
filters DF1 to DFm are required when moving a plurality of N virtual sound sources VSS
simultaneously.
[0039]
Therefore, the total number of taps of the digital filter is 1024 taps × 16 × N pieces = N ×
16384 taps.
That is, it is necessary to execute the processes of (A) to (C) in real time for the number of taps.
Note that this processing needs to be performed during one sample period of the original audio
signal u (ω), for example, a period of 1 / 44.1 kHz.
[0040]
03-05-2019
10
However, such processing requires the DSP or CPU to have several GFLOPS to several tens
GFLOPS of computing ability. For this reason, at the current technical level, when controlling the
position of the virtual sound source VSS in real time, the number of virtual sound sources VSS is
limited to several to several tens.
[0041]
In recent years, although the wave-field synthesis as described above has become feasible due to
the speeding up of personal computers and the rise of high-speed DSP, etc., computing power
and memory transfer capability are still required to be huge. However, there is a strong demand
for means of implementation in lower cost systems.
[0042]
The present invention is intended to solve the above problems and problems.
[0043]
In the present invention, a plurality of speakers are arranged to form a speaker array, and a
predetermined transfer function is convoluted in each of the audio signals supplied to the
plurality of speakers to generate an acoustic wavefront output from the plurality of speakers. The
wave source is combined to form a virtual sound source in the area including the speaker array,
and the virtual sound source is dense at the vicinity of the speaker array and rough at a distance,
and is located at the intersection of unequally spaced meshes. It is a method of reproducing an
audio signal in which the transfer function is controlled as described above.
[0044]
According to the present invention, the number of filter coefficients can be reduced, and as a
result, the capacity of the coefficient memory storing the filter coefficients can be reduced.
Also, loading of the filter coefficients into the digital filter is facilitated.
[0045]
[11] Auditory Characteristics of Human As described in FIG. 14, in the case where the locatable
03-05-2019
11
area ABC is divided by a mesh and the virtual sound source VSS is arranged at the intersection of
the mesh, the coarser the mesh, the more localized the virtual sound source VSS. The finer the
point is, the finer the point is.
However, since there is a limit to the resolution of human discrimination for localization of sound
image, the difference can not be seen even if the mesh is too fine.
[0046]
On the other hand, there is also an auditory characteristic that the resolution for discriminating
the localization is relatively high in the vicinity of the listener, but becomes lower the further
from the listener.
For example, even if the position of the sound source is changed by 10 cm in the vicinity of the
listener, the difference can be seen, but when the sound source moves away from the listener, the
difference of 10 cm can not be perceived.
[0047]
The present invention utilizes such audibility characteristics to enable the virtual sound source
VSS to be smoothly moved while suppressing the number of filter coefficients. The following will
be described in order.
[0048]
[12] Example of Reproducing Apparatus FIG. 1 shows an example of a reproducing apparatus
according to the present invention. In this example, the positions of the N virtual sound sources
VSS can be set independently. That is, the first to Nth original audio signals (digital signals) u (ω)
to u (ω) are supplied from the input terminals Tin1 to TinN to the digital low pass filters DF01 to
DF0N to remove unnecessary high frequency components. The filter output is supplied to digital
filters (DF11 to DFm1) to (DF1N to DFmN).
03-05-2019
12
[0049]
The digital filters (DF11 to DF1N) to (DFm1 to DFmN) correspond to the digital filters DF1 to
DFm in FIG. 13, and are configured by the FIR filter and loaded with predetermined filter
coefficients by the control circuit 20. The transfer functions H (ω) and C (ω) required for the
digital filters DF11 to DFmN are set.
[0050]
In this case, the filter coefficients of the digital filters (DF11 to DFm1) to (DF1N to DFmN) and the
loading method thereof will be described in detail by the following [13], but the filter coefficients
are prepared in the coefficient memory 21 of the control circuit 20. It is done.
Then, when the listener operates the external input means to specify the positions of the virtual
sound sources VSS to VSS, or when a program such as a game specifies by arithmetic processing,
the corresponding filter coefficient is read from the coefficient memory 21 and the digital filter
Loaded into DF11 to DFmN.
[0051]
Therefore, the reproduced audio signals q (ω) to q (ω) of the first to Nth original audio signals u
(ω) to u (ω) are extracted from the digital filters (DF11 to DFm1) to (DF1N to DFmN) Be Then,
the signals q (ω) to q (ω) are supplied to D / A converter circuits DA1 to DAm through addition
circuits AC1 to ACm to be D / A converted to analog audio signals, and the output signals are
output amplifier PA1. Are supplied to the speakers SP1 to SPm through .about.PAm. The
speakers SP <b> 1 to SPm are arranged as described with reference to FIG. 11 to constitute the
speaker array 10.
[0052]
Therefore, by the outputs of the speakers SP1 to SPm, the locatable area ABC is formed in a
substantially inverted triangle as described with reference to FIG. 11, and the first to Nth original
audio signals u (ω First to Nth virtual sound sources VSS to VSS corresponding to) to u (ω) are
arranged. Then, at this time, by setting transfer functions C (ω) and H (ω) of digital filters (DF11
to DFm1) to (DF1N to DFmN) to predetermined values, the first to Nth in arrangement range area
03-05-2019
13
ABC are obtained. The positions of the virtual sound sources VSS to VSS can be changed.
[0053]
[13] Arrangement position of virtual sound source in the present invention In the present
invention, together with dividing the arrangeable area ABC into a mesh, based on [11], for
example, as shown in FIG. Define the interval smaller and define the mesh interval wider as you
move away from the listener. And let the intersection of this mesh be the arrangement position of
virtual sound source VSS.
[0054]
[13-1] Setting Method of Arrangement Position (Part 1) In the example shown in FIG. 2, among
meshes dividing the arrangeable area ABC, the lattice GX in the direction parallel to the
arrangement direction of the speakers SP1 to SPm is from the listener Set the distance wider as
you go further. Further, the lattice GY intersecting with the lattice GX is separated from the
lattice at the center PA (radial center of the speakers SP1 to SPm and orthogonal to the
arrangement direction) GY0 radially from the vertex PA and among the lattices GY Set the
interval widely.
[0055]
For example, lattices GX are set logarithmically at equal intervals, and lattices GY are set
logarithmically at equal intervals on the same lattice GX. Then, when the virtual sound source
VSS is arranged (moved), it is arranged at any of the intersections of the lattice GX and the lattice
GY.
[0056]
In this way, the distance between the intersections of the grid GX and the grid GY in the locatable
area ABC becomes wider as it gets farther from the listener. Then, when the placement position
of the virtual sound source VSS is moved for each intersection point of the mesh, the distance
between the intersection points is narrow even if the discrimination ability of human localization
03-05-2019
14
is high near the listener, so the movement of the virtual sound source VSS is continuous It will be
smooth. On the other hand, although the distance between the points of intersection becomes
wide at a distant place, the movement ability of the virtual sound source VSS does not become
discontinuous or sense of incongruity does not occur because the discrimination ability of human
localization is low. That is, the virtual sound source VSS can be moved smoothly.
[0057]
And in that case, the arrangement possible area ABC of the virtual sound source VSS is an
inverted triangle as shown in FIGS. 11 and 2 in principle of wavefront synthesis, so the virtual
sound source VSS should be arranged near the listener The zones in which the can be done are
narrow, so even if the spacing between the intersections is narrow, the number of intersections is
small. On the other hand, the further away from the listener, the wider the zone in which the
virtual sound source VSS can be arranged, but the distance between the intersections is wider at
a distance, so the number of intersections is also small.
[0058]
Therefore, the number of intersections as a whole, that is, the number of placement positions of
the virtual sound source VSS can be reduced, so that the number of filter coefficients can be
reduced. As a result, the capacity of the coefficient memory 21 storing filter coefficients is
reduced. be able to. Also, this eliminates the need for high transfer capability for the data line
between the coefficient memory 21 and the digital filters DF1 to DFm.
[0059]
[13-2] Setting Method of Arrangement Position (Part 2) In the example shown in FIG. 3, the
density of the intersection that is the arrangement position of the virtual sound source VSS is two
stages of coarse and dense. That is, in this example, the gratings GX and GY are set in a direction
parallel to and orthogonal to the arrangement direction of the speakers SP1 to SPm. Then, in the
zone near the listener, the interval of the intersection points of the grids GX and GY is defined
narrowly, and in the zone far from the listener, the interval of the intersection points of the grids
GX and GY is widely defined.
03-05-2019
15
[0060]
Therefore, also in this case, when the arrangement position of the virtual sound source VSS is
moved for each mesh intersection, the movement of the virtual sound source VSS becomes
continuous and smooth, and the user does not feel discomfort. In addition, since the number of
intersection points can be reduced, the memory capacity for filter coefficients can be reduced,
and high load capability is not required when loading filter coefficients into digital filters DF1 to
DFm. .
[0061]
[13-3] Setting Method of Arrangement Position (Part 3) In the example shown in FIG. 2, the
density of the intersection points is varied continuously or in multiple steps. In the example
shown in FIG. In the case of making the coarse and dense two stages different, it may be
intermediate between them. That is, the allocable area ABC is divided into zones in a number
orthogonal to the arrangement direction of the loudspeakers SP1 to SPm so as not to feel
unnaturalness, for example, 3 zones or 4 zones, and the interval of intersections in each zone is
Define a zone wider as you move away from the listener.
[0062]
Also in this case, the same effects as [13-1] and [13-2] can be obtained.
[0063]
[13-4] Setting Method of the Arrangement Position (Part 4) As apparent from FIG. 2 and the like,
the arrangement position of the virtual sound source VSS is generally set in line symmetry with
respect to the center grid GY0 of the grids GY. It is meaningless to make this as asymmetric.
[0064]
Therefore, in the example shown in FIG. 4, the vertex PA of the settable area ABC is set to the
origin of the XY coordinates (orthogonal coordinates), and the central grid GY0 is set to the Y
axis, passing through the vertex PA to the grid GX. Set a parallel straight line on the X axis.
03-05-2019
16
Then, if the intersection point Q1 in the right half area (first quadrant) of the locatable area ABC
and the intersection point Q3 in the left half area (third quadrant) are in line symmetry with
respect to the grid GY0 (Y axis) Do.
[0065]
Then, when the filter coefficients in the digital filters DF1 to DFm are loaded, if the virtual sound
source VSS is localized at the intersection point Q1 by the speakers SP1 to SPm, the virtual sound
source is arranged when the arrangement direction of the speakers SP1 to SPm is left-right
reversed VSS will be localized at the intersection point Q3.
Then, in practice, the filter coefficients loaded to the digital filters DF1 to DFm are localized to
the intersection point Q1 by loading the digital filters DFm to DF1 in the reverse order without
reversing the arrangement direction of the speakers SP1 to SPm. The supposed virtual sound
source VSS is localized at the intersection point Q3.
[0066]
Therefore, the total number of filter coefficients can be halved by loading the filter coefficients in
the reverse order when the virtual sound source VSS is arranged in the left half area and in the
right half. The capacity of the coefficient memory 21 for filter coefficients can be reduced to 1⁄2.
At that time, the number of locatable positions of the virtual sound source VSS does not decrease.
[0067]
For example, if the original capacity of the coefficient memory 21 is 1 Gbyte, 500 Mbytes can be
reduced. Also, the algorithm is simple because the loading of the filter coefficients is simply
reversed. Furthermore, in the case of extending this method to a three-dimensional space, the
upper and lower symmetry can also be used, so the memory capacity can be reduced to 1⁄4.
[0068]
[13-5] Setting Method of Arrangement Position (Part 5) For example, as shown in FIG. 2, when
03-05-2019
17
the interval between the intersections in the possible arrangement area ABC is varied
continuously or in multiple steps, the filter coefficients of the digital filters DF1 to DFm When
reading out the coefficient memory 21 from the coefficient memory 21, the reading process
becomes troublesome as described below.
[0069]
That is, the position of the virtual sound source VSS is usually expressed in the XY coordinate
system, regardless of whether the listener operates the external input means to specify or the
program such as a game specifies by arithmetic processing.
Therefore, if the intersections which are the localization positions of the virtual sound source VSS
are regularly arranged on the XY coordinates, if the filter coefficient is regularly stored in the
coefficient memory 21, the position of the designated virtual sound source VSS is simply
coefficiented It is possible to convert to the address of the memory 21 and read out the filter
coefficient from the address.
[0070]
However, for example, as shown in FIG. 2, when there is no regularity in the arrangement of the
intersections in the XY coordinates, an ID number is regularly assigned to each of the
intersections, and one to one correspondence with the ID numbers is made. A separate map table
with XY coordinates will be prepared. However, in the case of this method, it is difficult to
intuitively grasp the relative positional relationship between the position of the virtual sound
source VSS and the position of the listener only by looking at the ID number. Furthermore, it is
necessary to search the map table for ID numbers corresponding to each of the XY coordinates,
which increases the operation cost.
[0071]
Therefore, in the example shown in FIG. 5, the arrangement position of the virtual sound source
VSS is defined by a polar coordinate system arranged on the same plane as the XY coordinate
system. That is, in FIG. 5, polar coordinates are set with the vertex PA of the locatable area ABC
as the origin. At this time, in the center of the arrangement of the speakers SP1 to SPm, a straight
03-05-2019
18
line orthogonal to the arrangement direction is made to correspond to the Y axis of the XY
coordinates, and a straight line passing the vertex PA is parallel to the arrangement direction of
the speakers SP1 to SPm as the X axis Make it correspond. In polar coordinates, a distance R is
taken about the origin PA, and an angle θ is taken clockwise from the Y axis.
[0072]
Then, a plurality of circles GR are set concentrically around the origin PA, and a plurality of
straight lines GA are set radially around the origin PA. Here, the radius of the circle GR is made
larger toward the outside, for example, logarithmically at equal intervals. Further, although the
straight line GA makes the angular interval Δθ equal, the resolution (a spread angle) of human
hearing for two sound sources is about 2 °, so Δθ = 2 °. When the virtual sound source VSS is
localized or moved, the virtual sound source VSS is disposed at any one of the intersections
included in the allocable area ABC among the intersections of the circle GR and the straight line
GA.
[0073]
In this way, as in the case of [13-1], the distance between the points of intersection of the circle
GR and the straight line GA in the allocable area ABC becomes narrower near the listener and
becomes wider as it gets farther from the listener. Therefore, also in this case, when the position
of the virtual sound source VSS is moved for each intersection, the movement of the virtual
sound source VSS becomes continuous and smooth, and the user does not feel discomfort.
[0074]
In particular, by setting Δθ = 2 ° or so, the movement of the virtual sound source VSS can be
realized with no noticeable appearance of roughness in movement, even when the movement is
made near or away, and movement without excess or deficiency can be realized. . In addition,
since the number of intersection points can be reduced, the memory capacity for filter
coefficients can be reduced, and high load capability is not required when loading filter
coefficients into digital filters DF1 to DFm. .
[0075]
03-05-2019
19
Furthermore, since the points of intersection are represented in the form of polar coordinates,
even if the points of intersection are irregularly spaced, they can be treated mathematically
regularly, and as a result, filter coefficients can be regularly stored in the coefficient memory 21.
Reading of the filter coefficient from the coefficient memory 21 is facilitated.
[0076]
Also, the position of the intersection point is expressed by two parameters R and θ in the polar
coordinate system, but since there is a relationship of X = R · sin θ Y = R · cos θ between the XY
coordinate system, it is easy to mutually It can be converted.
[0077]
Furthermore, only by increasing or decreasing the parameter R, the movement of the virtual
sound source VSS closer to or away from the listener can be realized, and control can be
facilitated.
Further, also in the case of expressing a motion in which the virtual sound source VSS rotates in
an arc shape left and right around the listener, it is only necessary to increase or decrease the
parameter θ.
By handling the coordinates mathematically in this manner, it becomes easy to control the
movement of the virtual sound source VSS.
[0078]
In addition, left-right symmetry as described in [13-4] is also established. That is, the intersection
point of the angle θ located in the right half area (first quadrant) of the locatable area ABC and
the intersection point of the angle -θ located in the left half area (third quadrant) with respect to
the Y axis It becomes line symmetry. Therefore, when the virtual sound source VSS is arranged in
the left half area, it is sufficient to load the filter coefficients to be loaded to the digital filters DF1
to DFm to the digital filters DFm to DF1 in reverse order when arranging in the right half area. .
[0079]
03-05-2019
20
[14] Summary According to the above-described reproducing apparatus, the arrangement
position of the virtual sound source VSS can be reduced without impairing the listener's aural
sense, so that the capacity of the coefficient memory 21 storing the filter coefficient can be
reduced. The upper memory space can be reduced, and the cost can be reduced.
[0080]
Further, when moving the virtual sound source VSS, the number of movable points (intersection
points) can be significantly reduced without impairing the listener's sense of hearing, so the
frequency of reading of the filter coefficient can be reduced and the load can be reduced.
In addition, the bandwidth of the data bus for transferring the filter coefficients can be reduced
in implementation. Further, by defining the position of the virtual sound source VSS in polar
coordinate format, the movement of the virtual sound source VSS can be easily controlled.
[0081]
[15] Others In the above description, although a plurality of m speakers SP1 to SPm are arranged
horizontally in one row to form a locatable area ABC of the virtual sound source VSS in a twodimensional manner The present invention can also be applied to the case where the speakers
are arranged in a matrix over rows and a plurality of columns, and the arrangeable area ABC of
the virtual sound source VSS is three-dimensionally formed. In that case, the speakers SP1 to
SPm can be arranged in a cross shape or an inverted T shape, and the sensitivity and
discrimination ability regarding the direction of hearing are high in the horizontal direction but
low in the vertical direction. The number of speakers can also be reduced.
[0082]
Furthermore, in the above description, the loudspeakers SP1 to SPm and the wavefront synthesis
surface SSR are parallel to each other, but they need not be parallel, and furthermore, the
loudspeakers SP1 to SPm need not be arranged linearly or planarly Good. Also, when integrating
with an AV system etc., the speakers SP1 to SPm are arranged in a frame shape at the top,
bottom, left and right of the display (screen), It can also be arranged.
03-05-2019
21
[0083]
The present invention can also be applied to a rear speaker, side speakers, and a speaker system
that outputs sound waves in the vertical direction. Alternatively, the present invention can be
combined with general 2-channel stereo or 5.1-channel audio.
[0084]
[List of abbreviations] AV: Audio and Visual CPU: Central Processing Unit D / A: Digital to Analog
DSP: Digital Signal Processor ID: IDentification
[0085]
It is a systematic diagram showing one form of this invention.
It is a figure which shows the acoustic space for demonstrating this invention. It is a figure which
shows the acoustic space for demonstrating this invention. It is a figure which shows the acoustic
space for demonstrating this invention. It is a figure which shows the acoustic space for
demonstrating this invention. It is a figure which shows the acoustic space for demonstrating this
invention. It is a figure which shows the numerical formula for demonstrating this invention. It is
a figure which shows the acoustic space for demonstrating this invention. It is a figure which
shows an example of the acoustic space which can apply this invention. It is a figure which shows
the mode of the wave front synthetic | combination which can apply this invention. It is a top
view for demonstrating the setting possible area of a virtual sound source. It is a figure which
shows the acoustic space for demonstrating this invention. It is a systematic diagram showing
one form of a circuit which can be used for this invention. It is a figure which shows the acoustic
space for demonstrating this invention.
Explanation of sign
[0086]
DESCRIPTION OF SYMBOLS 10 ... Speaker array, 20 ... Control circuit, 21 ... Coefficient memory,
AF1-AFm ... Addition circuit, DA1-'m ... D / A converter circuit, DF01-DF0N ... Low-pass filter,
DF11-DFmN ... Digital filter, SP1-SPm ... speaker
03-05-2019
22
03-05-2019
23
Документ
Категория
Без категории
Просмотров
0
Размер файла
36 Кб
Теги
jp2006246310
1/--страниц
Пожаловаться на содержимое документа