JP3172391B2 - Sound environment reproduction method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for reproducing a sound environment, and more specifically, to create a more realistic three-dimensional sound image by adding a reflection component of a virtual space in realizing a virtual reality. About the method. Further, the present invention relates to an acoustic analysis method capable of checking the acoustic characteristics of a design during the design of an indoor acoustic environment such as a concert hall, a concert theater, and a listening room.

[0002]

2. Description of the Related Art With the recent development of digital signal processing technology, it is possible to change a sound image in accordance with the movement of a sound source or the movement of a person (sound receiving point) in virtual reality. However, addition of information on acoustic characteristics such as reflection is limited. For example, a method of adding reflection from the floor as described in JP-A-6-30500 or a primary reflection by a mirror image method only in a virtual space of a rectangular parallelepiped such as a three-dimensional sound image device of Crystal River, USA The method of obtaining and adding is used.

On the other hand, when designing sound equipment such as a concert hall, it is necessary to determine the shape of the room, the conditions of reflection and sound absorption of the walls, and the like. 2. Description of the Related Art There is a method in which information about acoustic characteristics such as reflection is added to a sound taken in an anechoic room and output. Also, audio equipment for controlling the sound field of a pseudo concert hall is commercially available. Further, as a method of obtaining acoustic characteristics, there is a method of setting a transfer function in an actual environment or a model thereof and synthesizing the transfer function with a sound signal by a convolution operation.

[0004] On the other hand, there is also a method of performing simulation only by a computer without making a model. In this method, the state of reflection when a sound hits a wall is examined by a computer, and there are a mirror image method, a sound ray tracing method, and the like. In these methods, a transfer function is created by performing a simulation by paying attention to from which direction the initial reflection which is the most important in determining the sound quality of the sound field and from what time lapse.

FIG. 5 is a conceptual diagram of the mirror image method, where X is the position of the sound receiving point, Y is the position of the sound source, and ABCD is the virtual space (however,
For simplicity, the description will be made in two dimensions).
Y, C, a mirror image about the AB (0-order wall W ₀₎ of the D Y ', C', and D '. AD 'of Y', C ", B (0
Y mirror images about the next wall surface _{W 1) ", C",} B " and. The, Y", C "of AB" (2-order wall W ₂₎ the mirror image about the Y ''',C''' And Here, sound source Y
In the case of the example shown in the figure, the sound from the sound source Y directly reaches the sound receiving point X, the route from the AB surface (the 0th-order wall W ₀ ) to reach the sound receiving point X, and the AD surface After reflection at
There is a route that reaches the sound receiving point X after being reflected by the surface B. At the sound receiving point X, the sum of the sounds passing through these routes reaches, but this reaches Y, Y ′, and Y ″. It is equivalent to having a sound source.

FIG. 6 is a diagram for explaining an example of a simulation by the conventional mirror image method when the sound from the sound source Y arriving as described above is heard at the sound receiving point X.
First, in FIG. 6A, a virtual space is input (S
1) Input the position of the sound source / person (S2), and determine from what direction and how much time the reflection comes by mirror image method (S3). Next, a transfer function representing an acoustic characteristic is obtained from these pieces of information, and convolution with a sound or the like taken in an anechoic room is calculated (S4). If the position of the sound source / person changes, the process returns to the input of the position of the sound source / person in S2, and the above processing is repeated.

As shown in FIG. 6 (b), the flow for obtaining the reflection can be obtained recursively if the input sound source position is called the 0th-order sound source and the wall of the virtual space is called the 0th-order wall. FIG. 6 (b)
In, for obtaining the reflection of order N (S5), choose a wall W _N from the N-1 order wall (S6). If it cannot be selected (the processing has been performed on all the walls), the processing returns to the processing for obtaining the N-order reflection (S13). A mirror image of the (N−1) -order sound source by the wall W _N is obtained and set as an N-order sound source (S7).
A line segment connecting the position of the person and the N-th sound source has N walls W ₁ , W ₂ ,
.., W _N in this order, and check that they do not cross other walls (S8), and if they pass the check, register for reflection (S9). If it is necessary to determine the reflection of the (N + 1) th order (S10), a mirror image of the (N−1) th order wall by W _N is obtained and set as the Nth order wall (S11).
First-order reflection is obtained (S12).

[0008]

As described above, in virtual reality, a sound image is changed in accordance with the movement of a sound source or a person. Naturally, if the sound source or the person (sound receiving point) moves, the reflected sound changes accordingly. The method of measuring the actual transfer function is not suitable for realizing virtual reality, and the method of using a model requires that a model be created for each virtual space. It is necessary to set the transfer function accordingly, which is not practical. On the other hand, the method of obtaining and adding the primary reflection by the mirror image method only for the shape of the rectangular parallelepiped is too simplistic, and is far from the acoustic characteristics of the virtual space.

[0009] In addition, the computer simulation method is difficult to use in virtual reality because the current method requires a long time for calculation. Therefore, it is necessary to calculate by simulation how the initial reflection, which is the most important in determining the sound quality of the sound field, arrives at high speed in accordance with the movement of the sound source and sound receiving point, and from what direction over time. is there.

[0010]

According to the present invention, in order to solve the above problems, (1) a reflected wave of a sound is calculated from a wall of a virtual space and a sound source / sound receiving point position using a mirror image method. In a sound environment reproduction method for obtaining a transfer function and synthesizing a sound, a process of creating a table by partially calculating the reflected wave based on information of a wall in a virtual space, and generating a table and a sound source / sound receiving point position And a process of calculating a reflected wave from the calculation. Further, (2) a search space for reflection, or (3) a determination condition for reflection is reduced to reduce The acoustic characteristics can be calculated at high speed.

[0011]

The process of calculating a reflected wave using the mirror image method is a process of creating a table by performing a partial calculation based on information on walls in a virtual space, and calculating the reflected wave from this table and the sound source / sound receiving point position. By dividing it into processes, reflection and transfer functions calculated repeatedly are calculated in advance and stored in a table, and when calculating reflection, refer to this table to reflect and transfer. Eliminate redundant calculations when finding functions and reduce the amount of calculations.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS As described above, the wall W _N
The process of obtaining a mirror image by using the Nth order wall does not depend on the position of the sound source / sound receiving point, and it is sufficient to calculate once. Therefore, it may be calculated in advance, saved in a table, and referred to when calculating reflection. As a result, redundant calculations can be omitted. Further, optimization depending on how to select the walls W ₁ , W ₂ ,..., W _N can be performed. Wall W ₁ ,
Depending on the selection of W ₂ ,..., W _N, no matter how the sound source / sound receiving point is selected, the line connecting the position of the sound receiving point and the N-th sound source has N walls W ₁ , W ₂ ,. , W _N may not intersect. For example, if W ₁ and W ₂ are the same,
When the N ≧ 3, wall W _1, W also choose any two points on the _N, is if it is not always intersect with the line segment that connects W _i that (1 <i <N) is present (W _i intersection of Are not intersected if they are points on W ₁ , W _N ). By checking these in advance and deleting them from the search space for which reflection is to be obtained, useless calculations can be omitted, and the processing can be speeded up.

FIG. 1 is a diagram showing an example of reduction of a search space.
As described with reference to FIG. 5, X is a sound receiving point position, Y is a sound source position, ABCD is a virtual space, Y, C, and D mirror images of AB are Y ', C', and D ', and Y', C ', and D' are Y ', C', and D '. A of C ', B
Mirror images of D ′ are Y ″, C ″, B ″, and A ′ of Y ″
The mirror image of B "is Y""andC"". At this time, AD 'does not intersect (except point A) regardless of the point of line segment AB or line segment AB". . Therefore, this space (the shaded portion) is deleted from the search space for obtaining reflection. In fact, in this example, XY ""
Is, AB, AD ', does not intersect with AB ", not a reflection. Also, the wall of checks that do not intersect W _1,
It is not necessary to do for all walls except W ₂ ,..., W _N. For example, if the virtual space is convex (any 2 in the space)
If it is the same as a wall even if the points are connected by a straight line), there is no need to check itself. Even in the case of a concave space, the walls W ₁ and W _N
There is no need to check for walls that do not intersect even if you select any two points above and draw a line.

FIG. 2 is a diagram showing an example of the reduction of the intersection check (however, for the sake of simplicity, the description will be made in two dimensions).
ABCDEF is a virtual space. AB of C, D, E, F
Are mirror images C ', D', E ', and F'. B,
The mirror images of C ′, D ′, E ′ with respect to AF ′ are B ″, C ″,
D "and E". At this time, the range of the line connecting the point on AB and the point on AF 'is only the shaded portion in FIG. Therefore, there is no need to check E'F ', E'D', D'C ', B'C'. Also, BC, CD, AF,
Obviously, there is no need to check for AB ", B" C ", C" D ", so that DE, EF, E"
It is sufficient to check only F 'and E "D".

Next, a description will be given of a method for speeding up the calculation based on the information held in the table. Although various input methods of the virtual space are conceivable, it is assumed that the shape of the wall is divided into a parallelogram or a triangle in order to easily determine the reflection of the wall. At this time, the (N-1) -th sound source is a mirror image of the wall W _N (△ ABC) of Z, that is, N
The next sound source is Z ', O is the source point,

[0016]

(Equation 1)

## EQU1 ## In addition, ABC and 2 not on it
When there are points X and Z, the condition that the intersection R of XZ and △ ABC exists

[0018]

(Equation 2)

## EQU1 ## further,

[0020]

(Equation 3)

Therefore, by storing such information in a table, iterative calculation can be speeded up. Further, the transfer function Tr, T obtained from one reflection
l is, Tr = Hr * Td * T W1 * T w2 * ... * T WN Tl = Hl * Td * T W1 * T w2 * ... * T WN here, Hr, Hl is a head-related transfer function, the head of the It is determined from the direction and the direction of reflection. Td is a distance-dependent transfer function and is determined from the position of the mirror image and the position of the head. T _W1 , T _w2 ,…,
T _WN is a transfer function determined by the reflectance of each wall.
The reflectivity varies depending on the incident angle, frequency, and material of the sound. Here, it is assumed that the reflectance is determined for each wall as a transfer function for each incident angle of the sound. However, since the positional relationship between the walls is irrelevant to the positional relationship between the sound image and the sound receiving point, if the incident angle to T _W1 is given, the incident angle of T _w2 ,..., T _WN is uniquely determined, and T _W1 * T _w2 * ... * T _WN can be calculated. That is, T
_{W1, T w2, ..., T} WN can have a table as a transfer function for each angle of incidence on the T _W1, thereby making it possible to reduce the calculation for determining the transfer function. As a result of the above,
The search space for reflection can be reduced based on the information derived from the mutual positional relationship of the reflecting walls, the decision formula can be simplified, and the amount of calculation for iteratively calculating reflection and transfer functions can be reduced. Can be measured.

FIG. 3 is a flowchart for explaining an embodiment of the present invention. First, a virtual space is input (S21).
A table is created (S22). Next, the position of the sound source / sound receiving point is input (S23), and it is determined from what direction and how much time the reflection comes (S24). From these information and the transfer function of the table, a transfer function representing acoustic characteristics is obtained, and convolution with a sound or the like taken in an anechoic room is calculated (S25), and the positions of sound sources and sound receiving points are input (S25).
Returning to 23), the above processing is repeated. One flow for obtaining the reflection is selected from the table (S27). If it cannot be selected (processing has been performed on all the walls), return (S3
1). Obtains a mirror image of the sound source (S28), the line segment are N wall W ₁ connecting the position and N-order sound of the human, W _2, ..., W _N and intersect in this order, or the like does not intersect with the other wall Is checked (S29), and if the check passes, registration is performed for reflection (S30).

FIG. 4 is a diagram showing an example of the structure of a table. The table contains information on the (N-1) th wall, a pointer to a block in which the next candidate is stored, the order of reflection, V, α, and the like. For obtaining a mirror image of the sound image of, and for calculating conditions with intersections, a list of walls that must not have intersections,
Has a transfer function. These are created at the time of table creation and do not change thereafter. In addition, it has an area for holding a mirror image of a sound image in order to calculate a mirror image. This is because the mirror image of the N-th order sound image is calculated from the mirror image of the N-1 order sound image and V and α. Therefore, the calculation of the mirror image of the (N−1) -th order sound image is N
This must be done before the mirror image of the next sound image. vice versa,
If it is satisfied, the order of other calculations may be changed.

The sound reproducing method according to the present invention is not limited to the above-described embodiment, but can be applied in various ways. For example,

[0025]

(Equation 4)

If Dmin (W ₁ , W _i )> Dmax (W ₁ ,
When W _{i + 1} ), it can be deleted from the search space. Further, when only the initial reflection is required, if Dmin (W ₁ , W _N ) is a certain value or more, it is deleted from the search space.

[0027]

As described above, according to the present invention, the process of calculating the reflected wave by using the mirror image method includes the steps of performing a partial calculation based on the information on the wall in the virtual space and creating a table. By dividing into the process of calculating the reflected wave from the sound source and sound receiving point positions, it is possible to reduce the amount of reflection and transfer function calculations that are repeatedly calculated, and to reduce the search space for reflection when creating the table. This makes it possible to simplify the determination formula. Therefore, it is possible to calculate at which direction the initial reflection, which is the most important in determining the sound quality of the sound field, comes at a high speed in accordance with the movement of the sound source and the sound receiving point from what direction.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating an example of reduction of a search space.

FIG. 2 is a diagram illustrating an example of a decrease in intersection check;

FIG. 3 is a flowchart for explaining an embodiment of the present invention.

FIG. 4 is a diagram illustrating a configuration example of a table.

FIG. 5 is a diagram for explaining the principle of the mirror image method.

FIG. 6 is a simple flowchart of a simulation by the mirror image method in the embodiment of the prior art.

Continuation of the front page (58) Field surveyed (Int. Cl. ⁷ , DB name) G10K 15/00-15/12 H04S 1/00-7/00 JICST file (JOIS)

Claims (3)

(57) [Claims] 1. From the wall of a virtual space and the position of a sound source / sound receiving point,
Calculate the reflected wave of the sound using the mirror image method to obtain the transfer function,
In the sound environment reproducing method for synthesizing sound, the calculation of the reflected wave is performed by partially calculating the information of the wall in the virtual space to create a table, and the calculation of the reflected wave from the table and the sound source / sound receiving position A sound environment reproduction method characterized in that the sound environment reproduction method is performed by dividing the sound environment into two steps. 2. A method according to claim 1, wherein the line segment connecting the sound receiving point and the N-th sound source is at least one of N walls.
2. The sound environment reproducing method according to claim 1, wherein, if the two do not intersect, the N-order sound source space is deleted from the search space for obtaining reflection to reduce the search space for reflected waves. 3. The sound environment reproducing method according to claim 1, wherein when a straight line connecting any two points in the virtual space does not intersect with the wall, the intersection check for the wall is not performed. .