JP7583638B2 - Object-based audio rendering device and program - Google Patents

Description

The present invention relates to an object-based audio playback device, and in particular to an object-based audio rendering device and program that calculates speaker drive signals based on the concept of WFS (Wave Field Synthesis).

In recent years, object-based audio technology has been gaining attention, particularly in the field of movie sound. Previously, channel-based audio was the mainstream, in which multiple audio materials were mixed down to a specified channel format and then recorded.

In object-based audio, each piece of audio material is recorded separately as an audio object, and rendering is performed based on audio metadata that describes the level and coordinates of the audio object.

Hereinafter, such an object-based audio playback device will be referred to as an object-based audio rendering device. The object-based audio rendering device renders audio objects based on audio metadata and the positions of multiple speakers arranged for the audio objects, and generates drive signals according to the positions of each speaker, thereby playing audio according to the listening environment. Therefore, even if the content is played back with a speaker arrangement different from that used when it was created, playback that is adapted to the listening environment is possible.

[VBAP]
Conventionally, an algorithm called Vector Based Amplitude Panning (VBAP) is often used for rendering object-based audio (see, for example, Non-Patent Document 1).

In VBAP, the playback space is divided into a triangular region consisting of three speakers, and weights are calculated for the speakers located at each vertex of the triangle that contains the sound source coordinates. Amplitude panning is then performed by distributing the sound source signal using the calculated weights.

However, VBAP processes the sound as if the speakers are arranged on a sphere of a specific radius, and the sound source is placed on that sphere. Therefore, it is not possible to express the sense of perspective based on the distance of the sound source.

When using VBAP to represent a sound field that includes both near and far, distance attenuation and reverberation may be added (see, for example, Non-Patent Document 2), but these are methods based on psychoacoustics and are not intended to physically reproduce the sound field.

In addition, VBAP's weight calculation algorithm calculates weights assuming listening in the sweet spot, so playback quality is not guaranteed when listening from a position away from that sweet spot.

[WFS]
On the other hand, a method called WFS is known as a method for expressing a sound field based on wave acoustics. WFS is a sound field reproduction method supported by wave acoustics using the Kirchhoff-Helmholtz integral equation expressed by the following formula (see, for example, Non-Patent Document 3):

Here, p(r) denotes the sound pressure at any point r in a region V enclosed by the boundary ∂V, and G(r|r _o ) denotes the Green's function from point r _o on the boundary ∂V to any point r within the boundary ∂V. n(r _o ) denotes the unit vector inward in the normal direction of the boundary ∂V at point r _o . p(r _o ) denotes the sound pressure at point r _o on the boundary ∂V, and ∂p(r _o )/∂n denotes the sound pressure gradient vector in the direction of unit vector n(r _o ) at point r _o on the boundary ∂V. dS _o is an infinitesimal area on the boundary ∂V.

In addition, when the boundary has an arbitrary shape, a Green's function assuming the Neumann condition of ∂G(r|r _o )/∂n=0 is set as the boundary condition at the boundary ∂V. Then, the following window function a(r _o ) is introduced, which is determined by the average acoustic intensity I(r _o ) at the point r _o and the unit vector n(r _o ):

This results in the following equality:

G _o (r|r _o ) denotes the free-field Green's function from point r _o to point r. The Kirchhoff-Helmholtz integral equation expresses the fact that when a sound source exists outside a certain region, the sound field inside the region is determined by the sound pressure gradient distribution in the boundary normal direction of the boundary surface.

In WFS, speakers are densely arranged on the boundary surfaces of the area in which the sound field is to be reproduced, and the coordinates of the sound source (assumed to be a point source) are given, and the driving signals for the speakers are determined based on the Kirchhoff-Helmholtz integral equation.

VlLLE PULKKI, “Virtual Sound Source Positioning Using Vector Base Amplitude Panning”, J. Audio Eng. Soc., Vol. 45, No.6, 1997 June Recommendation ITU-R.BS.2127-0 Ahrens Jens, Rabenstein Rudolph, Spors Sascha, “The Theory of wave field synthesis revisited”, in 124th Conv. Audio Eng. Soc., 2008

In this way, by using WFS, the wavefront can be resynthesized inside the playback area, so unlike VBAP, there is no problem with the playback quality deteriorating at points away from the sweet spot.

However, in order to precisely synthesize wavefronts using WFS, it is necessary to place speakers closely on the boundary surfaces, and it is also necessary to add elements such as filters with specific frequency characteristics and delays. For this reason, when using WFS, there is an issue that the system size is larger than that of VBAP.

The present invention has been made to solve the above problems, and its purpose is to provide an object-based audio rendering device and program that can generate drive signals that represent the distance of a sound source based on WFS without expanding the scale of the system.

In order to solve the above problem, an object-based audio rendering device according to claim 1 is an object-based audio rendering device that generates a drive signal for each of a plurality of speakers arranged in a listening environment by performing rendering on a sound source object arranged at a position described in audio metadata, the object-based audio rendering device generating a drive signal for each of a plurality of speakers arranged in a listening environment, the object-based audio rendering _device generating a drive signal for each of a plurality of speakers arranged in a listening environment by performing rendering on a sound source object arranged at a position described in audio _metadata , the object-based audio rendering device generating a drive signal for each of a plurality of speakers by performing rendering on a sound source object arranged at a position described in audio _metadata , the object- _based audio rendering device generating a drive signal for each of a plurality of speakers by performing _rendering on a sound source object arranged at a position described in audio _metadata , the object- _based audio rendering device generating a drive signal for each of _a plurality of speakers by performing rendering on ), as well as signal amplification units, HPFs (high pass filters), attenuation application units, area element application units and first angle weighting units corresponding to each of the plurality of speakers, wherein the signal amplification units amplify the sound source signal by _a constant, the HPF sets HPFh(r l , r S ) based on the virtual sound source coordinates r _S and the speaker coordinates _{r l} of the first speaker, and performs filtering on the signal amplified by the signal amplification units using the HPFh(r _l , r _S ), the attenuation application units obtain an absolute value of a vector from a position of the virtual sound source _{coordinates r S to} a position of the speaker coordinates r _l of the first speaker, set the reciprocal of the result of squaring the absolute value as an attenuation coefficient 1/||r _l -r _S || ² , and apply the attenuation coefficient 1/||r _l -r _S || to the signal that has been subjected to the filtering processing. ² , the area element adding unit multiplies the signal multiplied by the attenuation adding unit by the area element ΔS _l , the first angle weighting adding unit sets an angle weight w C (θ l ) which, in a range from 0 to 1 based on the angle θ _l of the first speaker calculated by the angle calculation unit, takes a value closer to 1 as the absolute value of the angle θ _l approaches 0°, takes a value closer to 0 as the absolute value of the angle θ _l approaches 90°, and takes a value of 0 when the absolute value of the angle _{θ l} is 0° or less or 90° or more, and generates the drive signal d _{(r l )} for the first speaker by multiplying the signal multiplied by the area element adding unit by the angle weight w _C (θ _l ).

The object-based acoustic rendering device of claim 2 is the object-based acoustic rendering device of claim 1, further comprising: identifying three speakers including the virtual sound source therein as three speakers to be subjected to Vector Based Amplitude Panning ( _VBAP ) based on a unit vector of a virtual sound source direction toward the virtual sound source coordinate _rS and a unit vector of a speaker direction toward the speaker coordinate r1 of all speakers; obtaining unit vectors _rn1 , _rn2 , _rn3 of the three speakers to be subjected to VBAP; determining coefficients gl = _gn1 , _gn2 , _gn3 corresponding to the three speakers to be subjected to VBAP based on the unit vector of the virtual sound source direction toward _{the virtual} sound source coordinate rS and the unit vectors _rn1 , _rn2 , _rn3 of the three speakers to be subjected to VBAP; setting coefficient _gl = 0 corresponding to speakers other than the three speakers to be subjected to _VBAP among all the speakers; the VBAP coefficient multiplication unit multiplies the signal multiplied by the area element addition unit by the coefficient _{g output by the VBAP coefficient calculation unit, the second angle weighting unit sets an angle weight wS(θl) which, in a range from 0 to 1 ,} takes a value closer to 0 as the absolute value of the angle θl approaches 0°, takes a value closer to 1 as the absolute value of the angle _θl approaches 90°, and takes a value of 0 when the absolute value of the angle _θl is 0° or less or 90° or more, and generates a second drive signal by multiplying the signal multiplied by the VBAP coefficient multiplication unit, and the first angle weighting unit sets the angle weight _wS ( _θl ) which takes a value closer to 0 as the absolute value of the angle _θl approaches 0°, takes a value closer to 1 as the absolute value of the angle _θl approaches 90°, and takes a value of 0 when the absolute value of the angle _θl is 0° or less or 90° or more, based on the angle _θl of the first speaker calculated by the angle _calculation unit _. ), and multiplying the signal multiplied by the area element adding unit by the angle weight w _C (θ _l ) to generate a first drive signal, and the adder adds the first drive signal and the second drive signal to generate the drive signal d(r _l ) for the l-th speaker.

The object-based acoustic rendering apparatus of claim 3 is characterized in that, in the object-based acoustic rendering apparatus of claim 2, the VBAP coefficient multiplication unit inputs the sound source signal instead of the signal multiplied by the area element addition unit, and multiplies the sound source signal by the coefficient g _l .

にて、ＨＰＦ係数ｈ_n（ｒ_l，ｒ_S）を設定し、
ｘ［ｔ］及びｙ［ｔ］を時刻ｔにおける当該ＨＰＦの入力及び出力として、入出力特性が以下の式：

となるように、前記信号増幅部により増幅された信号に対し、前記ＨＰＦ係数ｈ_n（ｒ_l，ｒ_S）を用いてフィルタ処理を施す、ことを特徴とする。 The object-based acoustic rendering apparatus of claim 4 is the object-based acoustic rendering apparatus of any one of claims 1 to 3, wherein the HPF is expressed by the following formula: where α is a preset parameter, N is a preset filter order, and n=0, 1, . . . , N:

Set the HPF coefficients h _n (r _l , r _S ) in
Let x[t] and y[t] be the input and output of the HPF at time t, and the input/output characteristics are expressed by the following formula:

The signal amplified by the signal amplifier is filtered using the HPF coefficients h _n (r ₁ , r _s ) so that:

にて減衰係数ｇ（ｒ_l，ｒ_S）を求め、前記フィルタ処理が施された信号に対し前記減衰係数ｇ（ｒ_l，ｒ_S）を乗算する、ことを特徴とする。 According to a fifth aspect of the present invention, there is provided an object-based sound rendering apparatus according to any one of the first to fourth aspects, wherein the attenuation adding section calculates a value of β ₁ and β ₂ by the following equation:

and multiplying the filtered signal _by the attenuation coefficient g(r _l , r _{S )} .

Furthermore, the program of claim 6 is characterized in that it causes a computer to function as an object-based audio rendering device according to any one of claims 1 to 5.

As described above, according to the present invention, it is possible to generate a drive signal that expresses the distance of a sound source based on WFS without expanding the scale of the system.

1 is a block diagram showing an example of the configuration of an object-based acoustic rendering device according to a first embodiment. 4 is a pseudo code showing an example of processing of the object-based audio rendering device of the first embodiment. 4 is a block diagram showing an example of the configuration of an angle calculation unit; FIG. FIG. 13 is a diagram illustrating a case where speakers are sparsely arranged. FIG. 11 is a block diagram showing an example of the configuration of an object-based acoustic rendering device according to a second embodiment. 11 is a pseudo code showing an example of processing of the object-based audio rendering device of the second embodiment. 11 is a block diagram showing an example of the configuration of a VBAP coefficient calculation unit. FIG. FIG. 11 is a block diagram showing an example of the configuration of an object-based acoustic rendering device according to a third embodiment. 13 is a pseudo code showing an example of processing of the object-based audio rendering device of the third embodiment. FIG. 1 is a conceptual diagram for explaining VBAP. FIG. 1 is a conceptual diagram for explaining WFS.

The following describes in detail the embodiments of the present invention with reference to the drawings. The object-based acoustic rendering device of the first embodiment of the present invention generates a drive signal that expresses the distance of a sound source based on the WFS without expanding the scale of the system by simplifying the WFS configuration through approximation. The object-based acoustic rendering device of the second and third embodiments incorporates a VBAP configuration into the configuration of the first embodiment, thereby solving the problem of the drive signal becoming small when speakers are sparsely arranged.

Before explaining the object-based acoustic rendering devices of the first, second, and third embodiments of the present invention, we will provide an overview of the sound field reproduction methods using VBAP and WFS.

[Outline of VBAP]
First, an overview of VBAP will be described. As described above, VBAP is an amplitude panning technique that divides a reproduction space into triangular regions each consisting of three speakers, calculates weights for the speakers located at each vertex of a triangle including the sound source coordinates, and distributes the sound source signal to perform amplitude panning.

10 is a conceptual diagram for explaining VBAP. In the xyz space, the sound receiving point 100 is set as the origin, and the unit vector in the direction of the virtual sound source 101 is set as r _nS = (x _nS , _ynS , z _nS ) ^T. In addition, among the divided triangular regions, the unit vectors in the directions of the three speakers 102-1, 102-2, and 102-3 that include the virtual sound source 101 inside are set as r _n1 = (x _n1 , _yn1 , z _n1 ) ^T , r _n2 = (x _n2 , _yn2 , z _n2 ) ^T , and r _n3 = (x _n3 , _yn3 , z _n3 ) ^T , respectively.

ここで、Ｒ＝[ｒ_n1 ｒ_n2 ｒ_n3]^Tとする。 In this case, the unit vector r _nS in the direction of the virtual sound source 101 is expressed as follows using the unit vectors r _n1 , r _n2 , and r _n3 in the directions of the speakers 102-1, 102-2, and 102-3 and the coefficient vector g=(g _n1 , g _n2 , g _n3 ) ^T :

Here, let R ₌ [ _{rn1rn2rn3 ]} ^T .

Ｒ^-1はＲの逆行列である。 The coefficient vector g can be calculated by the following formula.

R ⁻¹ is the inverse matrix of R.

Weighting is performed using the coefficient vector g = (g _n1 , g _n2 , g _n3 ) ^T calculated by the above formula (5). Then, by playing back the weighted signals from the three speakers 102-1, 102-2, and 102-3, the listener can localize the sound image in the direction of the unit vector r _nS . Note that the weights of all speakers forming a triangle that does not include the virtual sound source 101 inside are set to 0.

[Outline of WFS]
Next, an overview of WFS will be described. As described above, WFS is a sound field reproduction method in which speakers are densely arranged on the boundary surface of an area in which a sound field is to be reproduced, and the sound pressure gradient in the normal direction of the boundary surface in the original sound field, which is the sound field to be reproduced, is reproduced at the speaker positions based on the Kirchhoff-Helmholtz integral equation of the above formula (1), thereby reproducing the sound field inside the boundary surface.

Figure 11 is a conceptual diagram to explain WFS, where (1) shows the original sound field and (2) shows the reproduced sound field.

ここで、ｋは波数である。尚、図１１の座標ｒ₁は、ｌ＝１の場合を示している。 The original sound field of (1) is formed in the xyz space by sound waves radiated from a sound source 103 located at a point with coordinate r _S = (x _S , y _S , z _S ) ^T outside the area C1. The sound pressure p at coordinate r _l = (x _l , y _l , z _l ) ^T on the boundary surface of the area C1 in this case is expressed by the following formula: l (el) indicates the number of the point with coordinate r _l , and corresponds to the number (system number) of the speaker 106-l described later.

Here, k is the wave number. Note that the coordinate _r1 in FIG. 11 indicates the case where l=1.

Furthermore, the unit vector inward in the boundary normal direction (boundary surface inward normal vector) at the point of coordinate r _l = (x _l , y _l , z _l ) ^T on the boundary surface of area C1 is defined as n(r _l ). Also, the angle between the vector (r _l -r _S ) connecting the point of coordinate r _S to the point of coordinate r _l and the boundary surface inward normal vector n(r _l ) is defined as θ _l . The sound pressure gradient vector in the normal direction at the point of coordinate r _l is expressed by the following formula.

Therefore, according to the Kirchhoff-Helmholtz integral equation, by driving the speaker array 105, which is discretely arranged in the reproduced sound field of (2), with a drive signal d(r _l ) of the following equation according to the coordinate r _l of each speaker element, it is possible to reproduce a sound field in which a point sound source of the virtual sound source 104 is arranged at the point of coordinate r _S within the area C2.

Here, ΔS _l is the area (area element) of the lth element when the boundary surface of area C2 is divided into area elements centered on the speaker position, and the more densely the speaker elements are arranged, the smaller the value, and the more sparsely the speaker elements are arranged, the larger the value.

In the xyz space, r _S is the coordinate of the virtual sound source 104 (virtual sound source coordinates), and r _l is the coordinate of the l-th speaker element that constitutes the speaker array 105 (speaker coordinates).

Furthermore, w _C (θ _l ) is a function defined by the following equation.

[Object-based audio rendering device]
Next, the object-based audio rendering apparatuses according to the first, second and third embodiments will be described.

In the first embodiment, in the above equation (8) showing the speaker drive signal d(r _l ) determined by the WFS sound field reproduction method, the term e -jk||r _{l -rs||, which represents the time delay due to the distance between the virtual sound source 104 located at the virtual sound source coordinate r S} and the speaker located at the ^speaker coordinate r _l, is ignored, thereby simplifying the system at the expense of accurate wavefront reproduction.

In addition, in the first embodiment, the term jk∥r _l -r _S ∥+1 in the above equation (8) has a high-pass characteristic that changes depending on the distance between the virtual sound source 104 and the speaker, and this characteristic is approximated by a low-order high-pass filter h(r _l , r _S ) (HPF) that depends on the distance.

As a result, in Example 1, by simplifying the WFS configuration through approximation, it is possible to generate a drive signal that expresses the distance of a sound source based on the WFS without expanding the scale of the system.

In addition, in Examples 2 and 3, by incorporating a VBAP configuration into the configuration of Example 1, in addition to the effects of Example 1, the problem of the drive signal becoming small when the speakers are sparsely placed is resolved.

Example 1
First, an object-based acoustic rendering device according to Example 1 will be described. As described above, in Example 1, in the above-mentioned formula (8) for generating the speaker drive signal d(r _l ), the term e ^{-jk||r l -rs||} representing a time delay is ignored, and the term jk||r _l -r _S ||+1 is treated as an HPF, and the high-pass characteristics that change depending on the distance between the virtual sound source 104 and the speaker are approximated by a low-order high-pass filter h(r _l , r _S ) (HPF) that depends on the distance.

Figure 1 is a block diagram showing an example of the configuration of an object-based acoustic rendering device of the first embodiment, and Figure 2 is pseudocode showing an example of the processing of the object-based acoustic rendering device of the first embodiment.

This object-based acoustic rendering device 1 includes signal amplifiers 9-1 to 9-L, HPFs (high pass filters) 10-1 to 10-L, attenuation adding units 11-1 to 11-L, area element adding units 12-1 to 12-L, angle weighting adding units 13-1 to 13-L, and an angle calculation unit 20. L is the number of speakers 106-1 to 106-L, and is also the number of systems.

Hereinafter, in the object-based acoustic rendering device 1, the components corresponding to the lth speaker 106-l (lth system) are respectively the signal amplifier 9-l, HPF 10-l, attenuation adding unit 11-l, surface area element adding unit 12-l, and angle weighting adding unit 13-l. In other words, the object-based acoustic rendering device 1 comprises L system signal amplifiers 9-l, HPF 10-l, attenuation adding unit 11-l, surface area element adding unit 12-l, and angle weighting adding unit 13-l, and further comprises an angle calculation unit 20. l = 1, ..., L.

The overall processing of the object-based sound rendering apparatus 1 will be described with reference to Fig. 2. The object-based sound rendering apparatus 1 receives a sound source signal S and virtual sound source coordinates _rS (step S201).

The object-based acoustic rendering device 1 inputs, for each of the speakers 106-1 to 106-L (for the speaker 106-l system), the speaker coordinates r _l of the speaker 106-l, the area element ΔS _l of the speaker 106-l, and the inward normal vector n(r _l ) of the boundary surface at the position of the speaker 106-l (step S202).

Here, the subscript "l" in the speaker coordinate r _l , the surface area element ΔS _l and the boundary surface inward normal vector n(r _l ) is the number (system number) of the speakers 106-1 to 106-L.

The sound source signal S is, for example, a signal broadcast from a broadcast station, and the virtual sound source coordinates r _S are, for example, data included in the audio metadata broadcast from the broadcast station. The speaker coordinates r _l , the area element ΔS _l and the boundary surface inward normal vector n(r _l ) are, for example, data included in speaker layout information preset by a user who operates the object-based audio rendering device 1.

In this case, the object-based acoustic rendering device 1 receives a sound source signal S and virtual sound source coordinates r _S transmitted from a broadcasting station, and inputs speaker coordinates r _l , area element ΔS _{l ,} and boundary surface inward normal vector n(r _l ) preset by the user.

The object-based acoustic rendering device 1 sets HPFh(r l , r _{S )} of the HPF 10-l having coefficients of a predetermined order for the speaker 106-l based on the virtual sound source coordinates r _S input in step S201 and the speaker coordinates r _l input in step _S202 (step S203).

The object-based acoustic rendering device 1 sets the attenuation coefficient 1/||r l -r S ∥ 2 of the attenuation adding unit 11-l for the _speaker 106-l based on the virtual sound source coordinates r _S input in step S201 and the speaker coordinates _{r l} input in step ^S202 , in accordance with the distance between the virtual sound source 104 and the speaker 106-l (step S204).

In other words, the object-based acoustic rendering device 1 calculates the absolute value ||r _l -r _S || of the vector from the position of the virtual sound source 104 to the position of the speaker 106-l, and sets the inverse of the result, ||r _l -r _S || ² , of squaring the absolute value as the attenuation coefficient 1/||r _l -r _S || ² .

The object-based acoustic rendering device 1 calculates an angle θ _l between a vector (r _l -r S ) connecting the position of the virtual sound source 104 to the position of the speaker 106-l and the boundary surface inward normal vector n(r _l ) at the position of the speaker 106-l, based on the virtual sound source coordinate r _S input in step S201, _and the speaker _coordinate r l and boundary surface inward normal vector n(r _l ) input in step S202 (step S205). This angle θ _l is calculated by the angle calculation unit 20.

The object-based acoustic rendering device 1 determines whether the absolute value |θ _l | of the angle θ _l calculated for the speaker 106-l in step S205 is greater than 0° and less than 90°, and sets the angle weight w _C (θ _l ) in accordance with equation (9) (step S206).

Specifically, if the object-based sound rendering device 1 determines that the formed angle |θ _l | is greater than 0° and less than 90° (if cos θ _l > 0 in the above formula (9)), it sets the angle weight w _C (θ _l ) = cos θ _l . On the other hand, if the object-based sound rendering device 1 determines that the formed angle |θ _l | is less than 0° or greater than 90° (if other than cos θ _l > 0 in the above formula (9)), it sets the angle weight w _C (θ _l ) = 0. This angle weight w _C (θ _l ) is used by angle weight adding section 13-l.

The angle weight _wC ( _θl ) takes a value in the range from 0 to 1. When the angle | _θl | is in the range of 0°<| _θl |<90°, the closer the angle | _θl | is to 0°, the closer the value is to 1. When the angle |θl| is closer to 90°, the closer the value is to 0. When the angle | _θl | is in the range of | _θl |≦0° or 90°≦| _θl |, the angle weight takes a value of 0.

＊は畳み込みを表す。この駆動信号ｄ（ｒ_l）は、信号増幅部９－ｌ、ＨＰＦ１０－ｌ、減衰付加部１１－ｌ、面積要素付加部１２－ｌ及び角度重み付加部１３－ｌにより算出される。 The object-based acoustic rendering device 1 calculates and outputs a drive signal d(r l ) for the speaker 106-l using the constant u (for example, u = 1/2π, which may be a different value for each system), the sound source signal S and virtual sound source coordinates r _S input in step S201, the speaker coordinates r _l and area element ΔS _l input in step S202, the HPFh(r _l , r _S ) set in step S203, the attenuation coefficient 1/||r _{l -r S} || ² set in step S204, and the angle weight w _C (θ _l ) set in step S206, using the following equation (step S207).

The symbol * denotes convolution. This drive signal d(r _l ) is calculated by a signal amplifier 9-l, a HPF 10-l, an attenuation adding section 11-l, an area element adding section 12-l and an angle weighting adding section 13-l.

Next, the processing of the components of the object-based acoustic rendering device 1 will be described with reference to Fig. 1. The signal amplifier 9-l (9-1 to 9-L) inputs the sound source signal S, amplifies the sound source signal S by multiplying it by a constant u (e.g., 1/2π), and outputs the amplified signal to the HPF 10-l.

The HPF 10-l (10-1 to 10-L) receives the virtual sound source coordinates r _S and the speaker coordinates r _L. Based on the virtual sound source coordinates r _S and the speaker coordinates r _L , the HPF 10-l sets an HPFh (r _L , r _S ) having a predetermined order of coefficients as in step S203. The HPF 10-l receives the amplified signal from the signal amplifier 9-l, filters the signal using the HPFh (r _L , r _S ), and outputs the filtered signal to the attenuation adder 11-l.

The attenuation adding unit 11-l (11-1 to 11-L) receives the virtual sound source coordinates r _S and the speaker coordinates r _L , and sets the attenuation coefficient 1/||r _L -r _S ^∥2 based on the virtual sound source coordinates r _S and the speaker coordinates r _L as in step S204. The attenuation adding unit 11-l then receives the filtered signal from the HPF 10-l, and multiplies the signal by the attenuation coefficient 1/||r _L -r _S ^∥2 to attenuate the signal in accordance with the distance between the virtual sound source 104 and the speaker 106-l. The attenuation adding unit 11-l outputs the attenuated signal (the signal with attenuation added) to the surface element adding unit 12-l.

The area element adding unit 12-l (12-1 to 12-L) inputs an area element ΔS _l . The area element ΔS _l takes a small value when the speakers 106-l are densely arranged, and takes a large value when the speakers 106-l are sparsely arranged.

The area element adding unit 12-l receives the signal to which attenuation has been added from the attenuation adding unit 11-l and multiplies the signal by an area element ΔS _l to generate a signal to which an area element has been added according to the degree of discreteness of the speaker 106-l. The area element adding unit 12-l outputs the signal to which the area element has been added to the angle weighting unit 13-l.

The angle calculation unit 20 inputs the virtual sound source coordinates r _S , the speaker coordinates r ₁ to r _L and the boundary surface inward normal vectors n (r ₁ ) to n (r _L ). Then, as in step S205, the angle calculation unit 20 obtains a vector (r _l -r S ₎ connecting the virtual sound source 104 position to the speaker 106-l position for each of the speakers 106-1 to 106-L (speaker 106-l), and calculates the angle θ _l between the vector (r _l -r _S ) and the boundary surface inward normal vector n (r _l ) at the speaker 106-l position. The angle calculation unit 20 outputs the formed angle θ _l (θ ₁ to θ _L ) to the corresponding angle weighting unit 13-l (13-1 to 13-L).

3 is a block diagram showing an example of the configuration of the angle calculation unit 20. The angle calculation unit 20 includes calculation units 30-1 to 30-L. The calculation unit 30-1 is a component that calculates the angle θ ₁ for the speaker 106-1, and inputs the virtual sound source coordinate r _S , the l=1st speaker coordinate r ₁ , and the boundary surface inward normal vector n (r ₁ ). The calculation unit 30-1 then obtains a vector (r ₁ -r _S ) that connects the virtual sound source 104 position to the speaker 106-l position, and calculates the angle θ ₁ between the vector (r ₁ -r _S ) and the boundary surface inward normal vector n (r ₁ ) at the speaker 106-1 position. The calculation unit 30-1 outputs the angle θ ₁ to the angle weighting unit 13-1.

Similarly, the calculation unit 30-L is a component that calculates the angle θ _L for the speaker 106-L, and receives as input the virtual sound source coordinates r _S , the l=Lth speaker coordinates r _L and the boundary surface inward normal vector n(r _L ). The calculation unit 30-L then obtains a vector (r _L -r _S ) connecting the position of the virtual sound source 104 to the position of the speaker 106-l, and calculates the angle θ _L between the vector (r _L -r _S ) and the boundary surface inward normal vector n(r _L ) at the speaker 106-L position. The calculation unit 30-L outputs the angle θ _L to the angle weighting unit 13-L.

Similar processing is performed for calculation units 30-2 to 30-(L-1), and angles θ 2 to θ L-1 between the vector (r ₂ to r _L-1 -r _S ) connecting the position of the virtual sound source 104 to the positions of the speakers 106-2 to 106-(L-1) and the inward normal vectors n(r ₂ ) to n(r _L-1 ) of the boundary surface at the positions of the speakers _106-2 to 106-( _L-1 ) are calculated, respectively.

1, angle weighting unit 13-l (13-1 to 13-L) receives angle θ _l (θ ₁ to θ _L ) from angle calculation unit 20, and sets angle weight w _C (θ _l ) using equation (9) based on angle θ _l as in step S206. Then, angle weighting unit 13-l receives a signal to which an area element has been added from area element adding unit 12-l, and multiplies the signal by angle weight w _C (θ _l ) to generate a signal reflecting angle θ _l .

The signal reflecting the angle _θl approaches a signal to which an area element has been added as the angle _θl approaches 0°. On the other hand, the signal reflecting the angle _θl approaches a signal of 0 as the angle _θl approaches 90°.

The angle weighting unit 13-l outputs a signal reflecting the formed angle θ _l as a drive signal d(r _l ) to the speaker 106-l.

1 and 2 show the configuration and processing of object-based audio rendering for the sound source signal S of one audio object, but in reality, processing is performed for the sound source signals S of a plurality of audio objects. In this case, the drive signals d(r _l ) generated for each audio object are added for each speaker 106-l, and the drive signal resulting from the addition is output to the speaker 106-l. The same applies to Figs. 5 and 6 (Example 2) and Figs. 8 and 9 (Example 3) described later.

As described above, according to the object-based acoustic rendering device 1 of the first embodiment, in the above equation (8) for generating the drive signal d(r _l ) for the speaker 106-l, the term e ^{-jk||r l -rs||} representing a time delay is ignored, and the term jk||r _l -r _S || + 1 is approximated as a low-order high-pass filter h(r _l , r _S ) (HPF) that depends on the distance between the virtual sound source 104 and the speaker 106-l.

Specifically, the signal amplifier 9-l amplifies the sound source signal S by multiplying the sound source signal S by a constant u (which may have a different value for each system). The HPF 10-l sets an HPFh(r _l , r _S ) having a coefficient of a predetermined order based on the virtual sound source coordinates r _S and the speaker coordinates r _l . The HPF 10-l then performs filtering on the signal amplified by the signal amplifier 9-l using the HPFh(r _l , r _S ).

The attenuation adding unit 11-l sets an attenuation coefficient 1/∥r _l -r _S ∥ ² based on the virtual sound source coordinates r _S and the speaker coordinates r _l . Then, the attenuation adding unit 11-l multiplies the filtered signal by the attenuation coefficient 1/∥r _l -r _S ∥ ² to generate a signal attenuated according to the distance between the virtual sound source 104 and the speaker 106-l.

The area element adding unit 12-l multiplies the attenuated signal by an area element ΔS _l of the speaker 106-l to generate a signal to which an area element has been added.

The angle calculation unit 20 calculates the angle θ l between the vector (r _l -r _s ) connecting the position of the virtual sound source ₁₀₄ to the position of the speaker 106-l and the inward normal vector n (r _l ) of the boundary surface at the position of the speaker 106-l.

The angle weighting unit 13-l sets the angle weight w _C (θ _l ) in accordance with the equation (9) based on the angle θ _l . Then, the angle weighting unit 13-l multiplies the signal to which the area element has been added by the angle weight w _C (θ _l ) to output a signal reflecting the angle θ _l to the speaker 106-l as a drive signal d(r _l ).

This allows the system to be simplified because the term e ^-jk||rl-rs|| , which indicates the time delay in equation (8), is ignored. Also, the system can be simplified by using the HPF 10-l of HPFh(r _l , r _S ), which approximates the term jk||r _l -r _S ||+1.

Therefore, by simplifying the configuration of the WFS through approximation, it is possible to generate a driving signal d(r _l ) that expresses the distance of the virtual sound source 104 based on the WFS without increasing the scale of the system.

(Another example of HPF10-1)
The HPF 10-l (10-1 to 10-L) of the object-based acoustic rendering device 1 shown in FIG. 1 performs filtering on the signal amplified by the signal amplifier 9-l (9-1 to 9-L) using HPFh (r _l , r _S ) which approximates the term jk∥r _l -r _S ∥+1 in the above equation (8).

For example, the term jk∥r _l -r _s ∥+1 in the above equation (8) is replaced with a high-pass filter expressed by the following equation.

αは、フィルタの特性を決定するためのパラメータであり、正の実数である。Ｎはフィルタ次数である。これらのパラメータα及びフィルタ次数Ｎの値は、ユーザが任意に決定するものとする。 x[t] and y[t] are the input and output of the HPF 10-l at time t. The HPF coefficients h _n (r _l , r _s ) are expressed by the following equation, where n=0, 1, . . . , N.

Here, α is a parameter for determining the characteristics of the filter and is a positive real number, and N is the filter order. The values of the parameter α and the filter order N are determined arbitrarily by the user.

That is, the HPF 10-l receives the virtual sound source coordinates r _S and the speaker coordinates r _l , as well as the parameter α and the filter order N preset by the user. Then, the HPF 10-l sets HPF coefficients having the above-mentioned formula (12) as coefficients, based on the virtual sound source coordinates r _S , the speaker coordinates r _l , the parameter α, and the filter order N.

The HPF 10-l receives the signal amplified by the signal amplifier 9-l (9-1 to 9-L) and performs filtering on the signal using the HPF coefficient h _n (r _l , r _S ) of the above formula (12) so that the input/output characteristics of the HPF 10-l satisfy the above formula (11). In other words, the HPF 10-l is configured with an HPF that has the input/output characteristics of the above formula (11).

This allows the object-based acoustic rendering device 1 as a whole to be simplified without increasing the system scale.

(Another Example of the Attenuation Addition Unit 11-l)
In addition, the attenuation adding unit 11-l (11-1 to 11-L) of the object-based acoustic rendering device 1 shown in Figure 1 sets an attenuation coefficient 1/||r _l -r _S || ² based on the virtual sound source coordinates r _S and the speaker coordinates r _l , and multiplies the filtered signal by the attenuation coefficient 1/||r _l -r _S || ² to generate a signal that is attenuated in accordance with the distance between the virtual sound source 104 and the speaker 106-l.

However, when the virtual sound source coordinate r _S and the speaker coordinate r _l coincide with each other, the attenuation coefficient 1/||r _l -r _S || ² diverges to infinity, and as a result, the signal cannot be reproduced. Therefore, the attenuation adding unit 11-l uses the attenuation coefficient g(r _l , r _S ) expressed by the following formula that approximates the attenuation coefficient 1/||r _l -r _S || ² :

_β1 is a positive real number for adjusting the signal amplitude, and _β2 is a positive real number for suppressing the divergence of the attenuation coefficient g(r _l , r _S ). The values of these parameters _β1 and _β2 are determined arbitrarily by the user. In particular, the value of the parameter _β2 is set by the user to a small value so as to suppress the divergence of the attenuation coefficient g(r _l , r _S ).

In other words, the attenuation adding unit 11-l finds the absolute value ||r _l -r _S || of the vector from the position of the virtual sound source 104 to the position of the speaker 106-l, squares the absolute value, adds the parameter β ₂ to the result to find the addition result (||r _l -r _S || ² + β ₂ ), and divides the parameter β ₁ by the addition result (||r _l -r _S || ² + β ₂ ) to find the attenuation coefficient g(r _l , r _S ).

Then, the attenuation adding unit 11-l multiplies the filtered signal by g(r _l , r _s ) to generate a signal attenuated according to the distance between the virtual sound source 104 and the speaker 106-l.

As the value of parameter _β2 becomes smaller, the behavior of the attenuation coefficient g(r _l , r _S ) approaches the attenuation coefficient 1/||r _l -r _S || ^2. On the other hand, as the virtual sound source coordinate r _S approaches the speaker coordinate r _l , the signal amplitude increases drastically.

This makes it possible to suppress signal divergence, and the user can adjust the fluctuation in signal amplitude when the position of the sound object changes according to the values of parameters β ₁ and β ₂ that the user sets.

Furthermore, if the user sets the values of the parameters _β1 and _β2 to _β1 = _β2 , when the virtual sound source coordinate _rS and the speaker coordinate _r1 coincide ( _rS = _r1 ), the attenuation coefficient g( _r1 , _r1 ) = 1.

Therefore, the user can use the values of these parameters β ₁ and β ₂ as a guide for adjustment so as not to unnecessarily increase or attenuate the signal.

Furthermore, if the user sets the values of parameters _β1 and _β2 to _β1 = _β2 / _ΔS1 , when the virtual sound source coordinate _rS and the speaker coordinate _r1 coincide ( _rS = _r1 ), the attenuation coefficient g( _r1 , _rS ) _ΔS1 = 1, and furthermore h( _r1 , rS ₎ g( _r1 , _rS ) ΔS1 _wC ( _θl ) = _1. This is because h( _r1 , _rS ) = 1 and _wC ( _θl ) = 1.

Therefore, when the virtual sound source coordinate r _S and the speaker coordinate r _l coincide with each other, the drive signal d(r _l ) output from the angle weighting unit 13-l to the speaker 106-l becomes the same as the signal amplified by the signal amplifier 9-l. In other words, the speaker 106-l located at the speaker coordinate r _l reproduces an amplified signal of the sound source signal S input to the object-based acoustic rendering device 1.

In this way, by using the attenuation coefficient g(r _l , r _S ) expressed by the above formula (13) in the attenuation adding unit 11-l, even if the virtual sound source coordinate r _S and the speaker coordinate r _l match, the signal does not diverge to infinity. This makes it possible to eliminate the problem of being unable to reproduce the signal.

Example 2
Next, an object-based acoustic rendering device according to a second embodiment will be described. As described above, the second embodiment incorporates the configuration of VBAP into the configuration of the first embodiment, thereby achieving the effects of the first embodiment and solving the problem that the drive signal d(r _l ) becomes small when the speakers 106-l are sparsely arranged. More specifically, the second embodiment generates the drive signal d(r _l ) by seamlessly switching between the configuration of a part of the simplified WFS shown in the first embodiment and the configuration of VBAP according to the angle weight.

In general, when using the WFS sound field reproduction method, it is assumed that many speakers 106-l are arranged closely together. However, in home audio, it is difficult to arrange many speakers 106-l closely together, and it is common to arrange the speakers 106-l sparsely.

4 is a diagram for explaining a case where the speakers 106-l are sparsely arranged. As shown in FIG. 4, it is assumed that the speakers 106-l are sparsely arranged in a reproduced sound field, and the position of the virtual sound source 104 (virtual sound source coordinates r _S ) is close to the boundary surface of the area where the sound field is to be reproduced.

In this case, in all speakers 106-l (106-1 to 106-L), the angle θ l between the vector (r _l -r _S ) connecting the position of the virtual sound source 104 to the position of the speaker 106-l and the inward normal vector n (r _l ) of the boundary surface at the position of the speaker 106- _l becomes a value close to or exceeding 90°.

For this reason, the angle weight w _C (θ _l ) used in the angle weighting unit 13-l becomes a value close to 0 or 0 according to the above formula (9). Then, the values of the drive signals d(r _l ) for all the speakers 106-l become small or 0, which causes a problem that the reproduced signal becomes small even though the virtual sound source 104 is close to the boundary surface.

In order to solve this problem, in the second embodiment, a VBAP configuration is added to the object-based acoustic rendering device 1 of the first embodiment shown in FIG. 1 so that the signal distribution law of the WFS is continuously and seamlessly switched to the VBAP signal distribution law in accordance with the angle _θl .

Here, in the object-based acoustic rendering device 1 of the first embodiment, which is the simplified WFS shown in Fig. 1, the HPFh(r _l , r _S ) of the HPF 10-l is a term related to frequency correction. Also, the attenuation coefficient 1/||r _l -r _S || ² of the attenuation adding unit 11-l is interpreted as a term related to distance attenuation due to the position of the virtual sound source 104, and the area element ΔS _l of the area element adding unit 12-l is interpreted as a term related to correction by discretizing the speaker 106-l. Furthermore, the angle weight w _C (θ _l ) of the angle weight adding unit 13-l is interpreted as a term related to the distribution side of the signal to each speaker 106-l.

In the second embodiment, a term relating to the distribution law of signals for each speaker 106-l determined by VBAP is connected in parallel to a term relating to the distribution law of signals for each speaker 106- _l based on the angle weight w _C (θ l ) of the angle weighting unit 13-l in WFS.

Figure 5 is a block diagram showing an example of the configuration of an object-based acoustic rendering device of the second embodiment, and Figure 6 is pseudocode showing an example of the processing of the object-based acoustic rendering device of the second embodiment.

This object-based acoustic rendering device 2 includes signal amplifiers 9-1 to 9-L, HPFs 10-1 to 10-L, attenuation adding units 11-1 to 11-L, surface area element adding units 12-1 to 12-L, angle weighting adding units 13-1 to 13-L, VBAP coefficient multipliers 14-1 to 14-L, angle weighting adding units 15-1 to 15-L, adders 16-1 to 16-L, an angle calculation unit 20, and a VBAP coefficient calculation unit 21.

Hereinafter, in the object-based acoustic rendering device 2, the components corresponding to the lth speaker 106-l (lth system) are the signal amplifier 9-l, HPF 10-l, attenuation adding unit 11-l, area element adding unit 12-l, angle weight adding unit 13-l, VBAP coefficient multiplier 14-l, angle weight adding unit 15-l, and adder 16-l. In other words, the object-based acoustic rendering device 2 comprises signal amplifiers 9-l, HPF 10-l, attenuation adding unit 11-l, area element adding unit 12-l, angle weight adding unit 13-l, VBAP coefficient multiplier 14-l, angle weight adding unit 15-l, and adder 16-l for L systems, and further comprises an angle calculation unit 20 and a VBAP coefficient calculation unit 21.

The signal amplifier 9-l, HPF 10-l, attenuation adding unit 11-l, surface area element adding unit 12-l, and angle weight adding unit 13-l are components of the WFS. The VBAP coefficient multiplier 14-l and angle weight adding unit 15-l are components of the VBAP, and are connected in parallel to the angle weight adding unit 13-l of the WFS.

Comparing the object-based acoustic rendering device 1 of the first embodiment shown in FIG. 1 with the object-based acoustic rendering device 2 of the second embodiment, both object-based acoustic rendering devices 1 and 2 have in common the fact that they are equipped with a signal amplifier 9-l, HPF 10-l, attenuation adding unit 11-l, surface element adding unit 12-l, angle weighting adding unit 13-l, and angle calculation unit 20.

On the other hand, the object-based acoustic rendering device 2 differs from the object-based acoustic rendering device 1, which does not have these components, in that it has a VBAP coefficient multiplication unit 14-l, an angle weighting unit 15-l, an addition unit 16-l, and a VBAP coefficient calculation unit 21.

The overall processing of the object-based acoustic rendering apparatus 2 will be described with reference to Fig. 6. As in step S201 of Fig. 2, the object-based acoustic rendering apparatus 2 inputs the sound source signal S and the virtual sound source coordinates _rS of the virtual sound source 104 (step S601).

As in step S202 in FIG. 2, the object-based acoustic rendering apparatus 2 inputs the speaker coordinates r _l , the area element ΔS _l and the boundary surface inward normal vector n(r _l ) for the speaker 106-l (step S602).

The object-based acoustic rendering apparatus 2 sets HPFh(r _l , r _S ) for the speaker 106-l based on the virtual sound source coordinates r _S and the speaker coordinates r _l (step S603), similarly to step S203 in FIG.

The object-based acoustic rendering apparatus 2 sets an attenuation coefficient 1/∥r _l -r _S ∥ ² for the speaker 106-l based on the virtual sound source coordinates r _S and the speaker coordinates r _l (step S604), similarly to step S204 in FIG.

Similar to step S205 in FIG. 2, the object-based acoustic rendering apparatus 2 calculates an angle θ l for the speaker 106- _l based on the virtual sound source coordinates r _s , the speaker coordinates r _l and the boundary surface inward normal vector n(r _l ) (step S605).

The object-based acoustic rendering device 2 determines whether the absolute value |θ _l | of the angle θ _l formed by the speaker 106-l is greater than 0° and less than 90°, and sets the angle weight w _C (θ _l ) of the WFS using the above equation (9) and the angle weight w _S (θ _l ) of the VBAP using the following equation (14) (step S606).

Specifically, when the object-based sound rendering device 2 determines that the formed angle |θ _l | is greater than 0° and less than 90° (when cos θ _l > 0 in the above formulas (9) and (14)), it sets the angle weight w _C (θ _l ) = cos θ _l and the angle weight w _S (θ _l ) = |1 - cos θ _l |. On the other hand, when the object-based sound rendering device 2 determines that the formed angle |θ _l | is less than or equal to 0° or greater than or equal to 90° (when other than cos θ _l > 0 in the above formulas (9) and (14)), it sets the angle weight w _C (θ _l ) = 0 and the angle weight w _S (θ _l ) = 0. This angle weight w _C (θ _l ) is used by the angle weight adding unit 13-l, and the angle weight w _S (θ _l ) is used by the angle weight adding unit 15-l.

The angle weight _wC ( _θl ) takes a value in the range from 0 to 1. When the angle | _θl | is in the range of 0°<| _θl |<90°, the closer the angle | _θl | is to 0°, the closer it is to 1 (cos _θl ), and the closer it is to 90°, the closer it is to 0. When the angle | _θl | is in the range of | _θl |≦0° or 90°≦| _θl |, the angle weight wC(θl) takes a value of 0.

The angle weight w _S (θ _l ) takes a value closer to 0 as the angle |θ _l | is closer to 0° when the angle |θ _l | is in the range of 0°<|θ l |<90°, and takes a value closer to 1 (|1-cos θ _l |) as the angle |θ _l | is closer to 90°, and takes a value of 0 when the angle |θ _l | is in the range of |θ _l |≦0° or 90°≦|θ _l |.

The object-based acoustic rendering device 2 identifies three speakers that include the virtual sound source 104 therein as three VBAP target speakers 106- _n1 , 106-n2, 106-n3 based on the unit vector of the virtual sound source direction from a specific sound receiving point (listener's position) to the virtual sound source coordinate r S and the unit vector of the speaker direction from a specific sound receiving point in all speakers 106-l to the speaker coordinate r _l .

The object-based acoustic rendering device 2 then obtains unit vectors r _n1 , r _n2 , and r _{n3 in the speaker directions of the three speakers 106-n1, 106-n2, and 106-n3 targeted for VBAP (step S607). The unit vectors r n1 , r n2 , and r n3 can be obtained} based on the coordinates (speaker coordinates) of the predetermined sound receiving point and the speakers 106-n1, 106-n2, and 106-n3, respectively. n1, n2, and n3 are indexes of the three speakers 106-n1, 106-n2, and 106-n3 targeted for VBAP.

前記式（１５）のｒ_nSは、仮想音源座標ｒ_Sへの仮想音源方向の単位ベクトルである。 The object-based acoustic rendering device 2 calculates coefficients g _l = g _n1 , g _n2 , g _n3 corresponding to speakers 106-n1, 106-n2, 106-n3 using the following formula based on the unit vector of the virtual sound source direction to the virtual sound source coordinate r _S and the unit vectors r _n1 , r _n2 , r _n3 of the speaker directions (step S608).

In the above equation (15), r _nS is a unit vector of the virtual sound source direction to the virtual sound source coordinate r _S .

Furthermore, the object-based acoustic rendering device 2 sets the coefficient g _l of the speaker 106-l other than the speakers 106-n1, 106-n2, and 106-n3 among the speakers 106-1 to 106-L to 0. These coefficients g _l =g _n1 , g _n2 , g _n3 , and 0 are calculated and set by the VBAP coefficient calculation unit 21.

For the speakers 106-n1, 106-n2, and 106-n3 with indexes l=n1, n2, and n3, the object-based acoustic rendering device 2 calculates and outputs a drive signal d(r _{l ) using the following equation based on a constant u (for example, u=1/2π), the sound source signal S and virtual sound source coordinates r S input in step S601, the speaker coordinates r l and} area element ΔS _l input in step S602, the HPFh(r _l , r _S ) set in step S603, the attenuation coefficient 1/||r _l -r _S || ² set in step S604, and the angle weights w _C (θ _l ) _and w _S (θ _l ) set in step S606 (step S609).

In this formula, the coefficients g _l = g _n1 , g _n2 , g _n3 . * indicates convolution. The drive signal d(r _l ) is calculated by a signal amplifier 9-l, HPF 10-l, attenuation adding unit 11-l, surface area element adding unit 12-l, angle weighting adding unit 13-l, VBAP coefficient multiplier 14-l, angle weighting adding unit 15-l, and adder 16-l.

この式（１７）は、前記式（１０）と同じである。 On the other hand, for indexes other than l=n1, n2, n3, i.e., for speakers 106-l other than speakers 106-n1, 106-n2, 106-n3, the object-based acoustic rendering device 2 calculates and outputs a drive signal d(r l ) using the following formula based on the constant u, sound source signal S, virtual sound source coordinates r _S , speaker coordinates r _l and area element ΔS _l , HPFh(r _l , r _S ), attenuation coefficient 1/||r _{l -r S} || ² and angle weighting w _C (θ _l ) (step S610).

This formula (17) is the same as the above formula (10).

This drive signal d(r _l ) is calculated by a signal amplifier 9-l, HPF 10-l, attenuation adding section 11-l, surface area element adding section 12-l and angle weighting section 13-l.

Next, the processing of the components of the object-based acoustic rendering device 2 will be described with reference to Figure 5. The signal amplifier 9-l (9-1 to 9-L), HPF 10-l (10-1 to 10-L), attenuation adding unit 11-l (11-1 to 11-L), area element adding unit 12-l (12-1 to 12-L), angle weight adding unit 13-l (13-1 to 13-L) and angle calculation unit 20 are the same as the components shown in Figure 1, so their description will be omitted.

Here, the angle weighting unit 13-l outputs the WFS drive signal reflecting the formed angle θ _l as a first drive signal to the corresponding adding unit 16-l. The angle calculation unit 20 outputs the formed angle θ _l to the corresponding angle weighting unit 13-l and angle weighting unit 15-l.

The VBAP coefficient calculation unit 21 inputs the virtual sound source coordinate r _S and the speaker coordinates r ₁ to r _L. Then, as in steps S607 and S608, the VBAP coefficient calculation unit 21 calculates coefficients g n1 , g _n2 , and g n3 corresponding to the three speakers 106- _n1 , 106-n2, and 106-n3 that are VBAP targets and include the virtual sound source 104 therein, based on the unit vector of the virtual sound source direction from a predetermined sound receiving point to the virtual sound source coordinate r _S and the unit vectors of the speaker directions from predetermined sound receiving points in all the speakers 106-1 to 106-L to the speaker coordinates r _{1 to} r _L.

The VBAP coefficient calculation unit 21 outputs the coefficients g _l =g _n1 , g _n2 , g _n3 to the VBAP coefficient multiplication units 14-n1, 14-n2, 14-n3 corresponding to the speakers 106-n1, 106-n2, 106-n3, respectively. Also, the VBAP coefficient calculation unit 21 outputs the coefficient g _l =0 to the VBAP coefficient multiplication unit 14-l corresponding to the speaker 106-l other than the speakers 106-n1, 106-n2, 106-n3.

7 is a block diagram showing an example of the configuration of the VBAP coefficient calculation unit 21. This VBAP coefficient calculation unit 21 includes a speaker determination unit 31 and a calculation unit 32. The speaker determination unit 31 inputs the virtual sound source coordinate r _S and the speaker coordinates r ₁ to r _L of all the speakers 106-1 to 106-L. Then, the speaker determination unit 31 specifies three speakers 106-n1, 106-n2, and 106-n3 that are VBAP targets and include the virtual sound source 104 therein, based on the virtual sound source coordinates r _S and the speaker coordinates r ₁ to r _L.

The speaker determination unit 31 acquires unit vectors r _n1 , r _n2 , r _n3 of the speaker directions for the three speakers 106-n1, 106-n2, 106-n3 that are VBAP targets, and outputs the virtual sound source coordinates r _S and the unit vectors r _n1 , r _n2 , r _n3 of the speaker directions to the calculation unit 32.

The calculation unit 32 inputs the virtual sound source coordinate r _S and the unit vectors r _n1 , r _n2 , r _n3 of the speaker direction from the speaker determination unit _31. Then, the calculation unit 32 calculates the coefficients g n1 , g _n2 , g n3 corresponding to the speakers 106-n1, 106-n2, 106- _n3 using the above formula (15) based on the unit vector of the virtual sound source direction to the virtual sound source coordinate r _S and the unit vectors r _n1 , r _n2 , r _n3 of the speaker direction. Then, the calculation unit 32 outputs the coefficients g _n1 , g _n2 , g _n3 to the corresponding VBAP coefficient multiplication units 14-n1, 14-n2, 14-n3.

Furthermore, the calculation unit 32 sets the coefficient g _l =0 corresponding to the speaker 106-l other than the speakers 106-n1, 106-n2, and 106-n3, and outputs it to the corresponding VBAP coefficient multiplication unit 14-l.

5, VBAP coefficient multiplication unit 14-l (14-1 to 14-L) receives coefficient g _l (any value of g _n1 , g _n2 , g _n3 , or 0) from VBAP coefficient calculation unit 21. Then, VBAP coefficient multiplication unit 14-l receives a signal to which an area element has been added from area element addition unit 12-l, and multiplies the signal by coefficient g _l to generate a signal that reflects coefficient g _l .

The VBAP coefficient multiplier 14-l outputs a signal reflecting the coefficient g _l to the angle weighting unit 15-l.

Angle weighting unit 15-l (15-1 to 15-L) receives angle θ _l (θ ₁ to θ _L ) from angle calculation unit 20, and sets angle weight w _S (θ _l ) in equation (14) based on the angle θ _l as in step S606. Then, angle weighting unit 15-l receives a signal reflecting coefficient g _l from VBAP coefficient multiplication unit 14-l, and multiplies the signal by angle weight w _S (θ _l ) to generate a signal reflecting angle θ _l .

In the angle weighting units 15-n1, 15-n2, and 15-n3 corresponding to the three speakers 106-n1, 106-n2, and 106-n3 including the virtual sound source 104, the signal reflecting the angle θ _l becomes closer to the signal reflecting the coefficient g _l as the angle θ _l becomes closer to 90°. On the other hand, the sound source signal S reflecting the angle θ _l becomes closer to 0 as the angle θ _l becomes closer to 0°. Also, in the angle weighting units 15-l corresponding to the speakers 106-l other than the three speakers 106-n1, 106-n2, and 106-n3 including the virtual sound source 104, the signal reflecting the angle θ _l is 0.

The angle weighting section 15-l outputs a signal reflecting the formed angle θ _l to the adding section 16-l as a second drive signal.

The adder 16-l (16-1 to 16-L) receives the first drive signal reflecting the angle θ l from the angle weighting unit 13-l, and receives the second drive signal reflecting the angle θ _{l from the angle weighting unit 15-l. The adder 16-l then adds the first drive signal reflecting the angle θ l and} the second drive signal reflecting the angle θ _l , and outputs the addition result as a drive signal d( _{r l} ) to the speaker 106-l.

As described above, according to the object-based acoustic rendering device 2 of the second embodiment, the VBAP configuration is incorporated into the WFS configuration of the first embodiment, and a drive signal d(r _l ) is generated by seamlessly switching between a part of the WFS configuration and the VBAP configuration according to the angle weighting.

Specifically, the VBAP coefficient calculation unit 21 calculates coefficients g _n1 , g n2 , and g _{n3 corresponding to the three speakers 106-n1, 106-n2, and 106-n3 that include the virtual sound source 104 therein, based on the unit vector of the virtual sound source direction from a specified sound receiving point to the virtual sound source coordinate r S and} the unit vectors of the speaker directions to the speaker coordinates r ₁ to _{r L.}

The VBAP coefficient multiplication unit 14-l inputs a coefficient g _l (any value of g _n1 , g _n2 , g _n3 , or 0) from the VBAP coefficient calculation unit 21. Then, the VBAP coefficient multiplication unit 14-l multiplies the signal obtained by the signal amplification unit 9-l, HPF 10-l, attenuation addition unit 11-l, and area element addition unit 12-l by the coefficient g _l to generate a signal reflecting the coefficient g _l .

The angle weighting unit 15-l sets the angle weight w _S (θ _l ) in the above formula (14) based on the formed angle θ _l . Then, the angle weighting unit 15-l generates a second drive signal that reflects the formed angle θ _l by multiplying the signal that reflects the coefficient g _l by the angle weight w _S (θ _l ).

The adder 16-l adds together a first drive signal reflecting the angle θ l and a second drive signal reflecting the angle θ _l obtained by processing the WFS consisting of the signal amplifier 9-l, the HPF 10-l, the attenuation adding unit 11-l, the surface element adding unit 12- _l and the angle weighting adding unit 13-l, and outputs the addition result as a drive signal d(r _l ) to the speaker 106-l.

In this way, the angle weighting unit 15-l using the angle weight w _S (θ _l ) is connected in series to the VBAP coefficient multiplier 14-l using the coefficient g _l , and the VBAP coefficient multiplier 14-l and angle weighting unit 15-l of the VBAP are connected in parallel to the angle weighting unit 13-l of the WFS. The angle weight w _C (θ _l ) used in the angle weighting unit 13-l and the angle weight w _S (θ _l ) used in the angle weighting unit 15-l are in a trade-off relationship in the range of 0 to 1.

For this reason, when the position of the virtual sound source 104 (virtual sound source coordinate r _S ) is away from the boundary surface of the speaker 106-l and the angle θ _l formed therewith is close to 0°, the angle weight w _C (θ _l ) becomes a large value close to 1, but the angle weight w _S (θ _l ) becomes a small value close to 0. In this case, the distribution side by the WFS becomes dominant in the drive signal d(r _l ).

On the other hand, when the position of the virtual sound source 104 (virtual sound source coordinate r _S ) is close to the boundary surface of the speaker 106-l and the formed angle θ _l is close to 90°, the angle weight w _C (θ _l ) is a small value close to 0, but the angle weight w _S (θ _l ) is a large value close to 1. In this case, the distribution side by VBAP becomes dominant in the drive signal d (r _l ).

In other words, depending on the position of the virtual sound source 104 relative to the speaker 106-l, the distribution side based on WFS and the distribution side based on VBAP are switched continuously and seamlessly.

Therefore, the object-based sound rendering device 2 of the second embodiment can achieve the same effect as that of the first embodiment. In an environment in which the speakers 106-l are sparsely arranged, as the position of the virtual sound source 104 (virtual sound source coordinate r _S ) approaches the boundary surface of the speakers 106-l and the angle θ _l formed increases, the angle weight of the WFS decreases and the angle weight of the VBAP increases, and the rendering rule is switched continuously. Therefore, even if the playback signal becomes small due to the decrease in the angle weight of the WFS, the problem of the playback signal becoming extremely small can be solved by compensating for it with the angle weight of the VBAP.

The HPF 10-l (10-1 to 10-L) of the object-based acoustic rendering device 2 of the second embodiment may be used by replacing jk∥r _l -r _S ∥+1 with the high-pass filter of the above formula (11) using the HPF coefficient h _n (r _l , r _S ) of the above formula (12), as in the first embodiment shown in FIG.

Also, the attenuation adding unit 11-l (11-1 to 11-L) of the object-based acoustic rendering device 2 of the second embodiment may use the attenuation coefficient g(r _l , r _S ) of the above formula (13) instead of the attenuation coefficient 1/∥r _l -r _S ∥ ² , as in the first embodiment shown in Fig. 1. In other words, the attenuation adding unit 11-l uses the attenuation coefficient g(r _l , r _S ) to generate a signal attenuated according to the distance between the virtual sound source 104 and the speaker 106-l.

Example 3
Next, a third embodiment will be described. As described above, in the third embodiment, similar to the second embodiment, the VBAP configuration is incorporated into the configuration of the first embodiment, thereby achieving the effects of the first embodiment and solving the problem that the drive signal d(r _l ) becomes small when the speakers 106-l are sparsely arranged. More specifically, the third embodiment generates the drive signal d(r _l ) by seamlessly switching between the entire simplified WFS configuration shown in the first embodiment and the VBAP configuration according to the angle weighting.

Figure 8 is a block diagram showing an example of the configuration of an object-based acoustic rendering device of the third embodiment, and Figure 9 is pseudocode showing an example of the processing of the object-based acoustic rendering device of the third embodiment.

This object-based acoustic rendering device 3 includes signal amplifiers 9-1 to 9-L, HPFs 10-1 to 10-L, attenuation adding units 11-1 to 11-L, surface area element adding units 12-1 to 12-L, angle weighting adding units 13-1 to 13-L, VBAP coefficient multipliers 14-1 to 14-L, angle weighting adding units 15-1 to 15-L, adders 16-1 to 16-L, an angle calculation unit 20, and a VBAP coefficient calculation unit 21.

The signal amplifier 9-l, HPF 10-l, attenuation adding unit 11-l, surface element adding unit 12-l and angle weight adding unit 13-l are components of the WFS, and the VBAP coefficient multiplier 14-l and angle weight adding unit 15-l are components of the VBAP. The components of the WFS and the components of the VBAP are connected in parallel as the object-based acoustic rendering device 3 as a whole.

When comparing the object-based acoustic rendering device 2 of the second embodiment shown in FIG. 2 with the object-based acoustic rendering device 3 of the third embodiment, both object-based acoustic rendering devices 2 and 3 have the same components in common.

On the other hand, the object-based acoustic rendering device 3 differs from the object-based acoustic rendering device 2 in that the VBAP coefficient multiplication unit 14-l and the angle weighting unit 15-l, which are components of the VBAP, are connected in parallel to the entire components of the WFS, whereas the object-based acoustic rendering device 3 is connected in parallel to only a part of the WFS, the angle weighting unit 13-l.

The overall processing of the object-based acoustic rendering device 3 will be described with reference to FIG. 9. The processing of steps S901 to S908 shown in FIG. 9 is the same as the processing of steps S601 to S608 by the object-based acoustic rendering device 2 of the second embodiment shown in FIG. 6, so a description thereof will be omitted here.

When the index of the speaker coordinate r _l obtained in step S907 is l = n1, n2, n3, the object-based acoustic rendering device 3 calculates and outputs a drive signal d(r l ) for the speakers 106-n1, 106-n2, 106-n3 using the following equation based on the constant u (e.g., u = 1/2π), the sound source signal S and virtual sound source coordinate r _S input in step S901, the speaker coordinate r _l and area element ΔS _l input in step S902, the HPFh(r _l , r _S ) set in step S903, the attenuation coefficient 1/|| _{r l -r S} || ² set in step S904, and the angle weight w _C (θ _l ) and angle weight w _S (θ _l ) set in step S906 (step S909).

In this formula, the coefficients g _l = g _n1 , g _n2 , g _n3 . * indicates convolution. The drive signal d(r _l ) is calculated by a signal amplifier 9-l, HPF 10-l, attenuation adding unit 11-l, surface area element adding unit 12-l, angle weighting adding unit 13-l, VBAP coefficient multiplier 14-l, angle weighting adding unit 15-l, and adder 16-l.

この式（１９）は、前記式（１０）及び前記式（１７）と同じである。 On the other hand, when the index is other than l=n1, n2, n3, the object-based acoustic rendering device 3 calculates and outputs a drive signal d(r l ) for each speaker 106-l other than speakers 106-n1, 106-n2, 106-n3 based on the constant u, sound source signal S, virtual sound source coordinate r _S , speaker coordinate r _l and area element ΔS _l , HPFh(r _l , r _S ), attenuation coefficient 1/||r _l -r _S || ² and angle weight w _C ( _{θ l} ) using the following formula (step S910).

This formula (19) is the same as the above formulas (10) and (17).

This drive signal d(r _l ) is calculated by a signal amplifier 9-l, HPF 10-l, attenuation adding section 11-l, surface area element adding section 12-l and angle weighting section 13-l.

In the calculation of equation (18), a delay occurs due to filtering between the calculation result of the WFS configuration, which is the first term on the right side, and the calculation result of the VBAP configuration, which is the second term. For this reason, it is necessary to add a delay to the VBAP side in order to adjust the phase in the angle weighting units 13-l and 15-l, but this will not be shown here.

In addition, in the formulas (18) and (19), the convolution * may be calculated by inversely transforming the product in the frequency domain using FFT (Fast Fourier Transform). The same applies to the formulas (16) and (17).

Referring to FIG. 8, the processing of each component of the object-based acoustic rendering device 3 is the same as the processing of each component of the object-based acoustic rendering device 2 of the second embodiment shown in FIG. 5, so the description of the processing of the components is omitted.

The input data of the VBAP coefficient multiplier 14-l (14-1 to 14-L) differs between the object-based acoustic rendering devices 2 and 3. The VBAP coefficient multiplier 14-l of the object-based acoustic rendering device 2 shown in FIG. 5 inputs a signal to which an area element has been added from the area element adding unit 12-l. In contrast, the VBAP coefficient multiplier 14-l of the object-based acoustic rendering device 3 shown in FIG. 8 inputs the sound source signal S.

8 receives a coefficient g _l (any value of g _n1 , g _n2 , g _n3 , or 0) from a VBAP coefficient calculation unit 21. The VBAP coefficient multiplication unit 14-l receives a sound source signal S and multiplies the sound source signal S by the coefficient g _l to generate a signal reflecting the coefficient g _l .

As described above, according to the object-based acoustic rendering device 3 of Example 3, the VBAP configuration is incorporated into the WFS configuration of Example 1, and the drive signal d(r _l ) is generated by seamlessly switching between the overall WFS configuration and the VBAP configuration according to the angle weighting.

Specifically, the VBAP coefficient multiplication unit 14-l inputs a coefficient g _l (any value of g _n1 , g _n2 , g _n3 , or 0) from the VBAP coefficient calculation unit 21. Then, the VBAP coefficient multiplication unit 14-l multiplies the sound source signal S by the coefficient g _l to generate a signal that reflects the coefficient g _l .

The angle weighting unit 15-l sets the angle weight w _S (θ _l ) in the above formula (14) based on the formed angle θ _l . Then, the angle weighting unit 15-l generates a second drive signal that reflects the formed angle θ _l by multiplying the signal that reflects the coefficient g _l by the angle weight w _S (θ _l ).

The adder 16-l adds together a first drive signal reflecting the angle θ l and a second drive signal reflecting the angle θ _l obtained by processing the WFS consisting of the signal amplifier 9-l, the HPF 10-l, the attenuation adding unit 11-l, the surface element adding unit 12- _l and the angle weighting adding unit 13-l, and outputs the addition result as a drive signal d(r _l ) to the speaker 106-l.

In this way, the angle weighting unit 15-l using the angle weight w _S (θ _l ) is connected in series to the VBAP coefficient multiplier 14-l using the coefficient g _l . The VBAP coefficient multiplier 14-l and angle weighting unit 15-l of the VBAP are connected in parallel to the signal amplifier 9-l, HPF 10-l, attenuation unit 11-l, surface element adding unit 12-l and angle weighting unit 13-l of the WFS. The angle weight w _C (θ _l ) used in the angle weighting unit 13-l and the angle weight w _S (θ _l ) used in the angle weighting unit 15-l are in a trade-off relationship in the range of 0 to 1.

For this reason, similarly to the second embodiment, when θ _l approaches 0°, the distribution side by WFS becomes dominant in the drive signal d(r _l ), and when θ _l approaches 90°, the distribution side by VBAP becomes dominant in the drive signal d(r _l ).

In other words, the distribution side between WFS and VBAP is switched continuously and seamlessly depending on the position of the virtual sound source 104 relative to the speaker 106-l.

Therefore, the object-based acoustic rendering device 3 of the third embodiment can achieve the same effect as that of the first embodiment. Also, as in the second embodiment, when rendering by WFS is being performed in an environment in which the speakers 106-l are sparsely arranged, if the virtual sound source 104 is placed close to a boundary surface, rendering by VBAP is continuously and seamlessly switched over. This solves the problem of the playback signal becoming smaller.

The HPF 10-l (10-1 to 10-L) of the object-based acoustic rendering device 3 of the third embodiment may be configured to use a high-pass filter of the above formula (11) using the HPF coefficient h _n (r _l , r _S ) of the above formula (12), replacing jk∥r _l -r _S ∥+1, as in the first embodiment shown in FIG. 1 and the second embodiment shown in FIG.

Furthermore, the attenuation adding unit 11-l (11-1 to 11-L) of the object-based acoustic rendering device 3 of the third embodiment may use the attenuation coefficient g(r l , r S ) of the above formula (13) instead of the attenuation coefficient 1/∥r _l -r _S ∥ ² , as in the _first embodiment shown in Fig. 1 and the second embodiment shown in Fig. 5. In other words, the attenuation adding unit 11-l uses the attenuation coefficient g(r _l , r _{S )} to generate a signal attenuated according to the distance between the virtual sound source 104 and the speaker 106-l.

However, when the virtual sound source 104 is placed close to the speaker 106-l, the signal amplitude may be significantly different between the simplified WFS configuration (HPF 10-l to angle weighting unit 13-l) and the VBAP configuration (VBAP coefficient multiplier 14-l and angle weighting unit 15-l). Therefore, the parameters β1 and β2 used to calculate the attenuation coefficient g(r _l , r _S ) in the above formula (13) need to be appropriately adjusted so that the signal amplitude does not differ significantly.

The present invention has been described above using Examples 1, 2, and 3, but the present invention is not limited to the above Examples 1, 2, and 3, and various modifications are possible without departing from the technical concept thereof.

In addition, a normal computer can be used as the hardware configuration of the object-based acoustic rendering devices 1, 2, and 3 according to the first, second, and third embodiments of the present invention. The object-based acoustic rendering devices 1, 2, and 3 are configured by a computer equipped with a CPU, a volatile storage medium such as a RAM, a non-volatile storage medium such as a ROM, and an interface, etc.

The functions of the HPFs 10-1 to 10-L, attenuation adding units 11-1 to 11-L, surface area element adding units 12-1 to 12-L, angle weighting adding units 13-1 to 13-L, and angle calculation unit 20 provided in the object-based acoustic rendering device 1 are each realized by having the CPU execute a program that describes these functions.

Furthermore, the functions of the HPFs 10-1 to 10-L, attenuation adding units 11-1 to 11-L, surface area element adding units 12-1 to 12-L, angle weighting adding units 13-1 to 13-L, VBAP coefficient multipliers 14-1 to 14-L, angle weighting adding units 15-1 to 15-L, adders 16-1 to 16-L, angle calculation unit 20, and VBAP coefficient calculation unit 21 provided in the object-based acoustic rendering devices 2 and 3 are each realized by having the CPU execute a program that describes these functions.

These programs are stored in the storage medium and are read and executed by the CPU. In addition, these programs can be distributed by storing them on storage media such as magnetic disks (floppy disks, hard disks, etc.), optical disks (CD-ROMs, DVDs, etc.), and semiconductor memories, and can also be transmitted and received via a network.

1, 2, 3 Object-based acoustic rendering device 9 Signal amplifier 10 HPF (high pass filter)
REFERENCE SIGNS LIST 11 attenuation adding section 12 area element adding section 13, 15 angle weight adding section 14 VBAP coefficient multiplying section 16 addition section 20 angle calculation section 21 VBAP coefficient calculation section 30, 32 calculation section 31 speaker determination section 100 sound receiving point 101, 104 virtual sound source 102, 106 speaker 103 sound source 105 speaker array r _nS unit vector r _n1 , r _n2 , r _n3 in virtual sound source direction unit vector g _l , g _n1 , g _n2 , g _n3 in speaker direction coefficient r _S virtual sound source coordinate r _l speaker coordinates C1, C2 area S sound source signal L number of speakers l speaker number (system number)
Angle formed by ΔS _l, surface area element n (r _l ) of speaker, inward normal vector k of speaker's boundary surface, and wave number θ _l (angle formed by vector (r _l -r _S ) connecting the position of virtual sound source 104 to the position of speaker 106-l and inward normal vector n (r _l ) of the boundary surface)
h(r _l , r _s ) HPF
w _C (θ _l ), w _S (θ _l ) Angular weight 1/||r _l -r _S || ² , g(r _l , r _S ) Damping coefficient d(r _l ) Drive signal

Claims (6)

1. An object-based audio rendering device that generates drive signals for a plurality of speakers arranged in a listening environment by performing rendering on a sound source object arranged at a position described in audio metadata,
Let L be the number of the plurality of speakers, l (=1, ..., L) be the number of the speakers, the lth speaker be the lth speaker, the coordinates of the virtual sound source from which a sound source signal is output be virtual sound source coordinates r _S , the coordinates of the lth speaker be speaker coordinates r _l , the inward normal vector of the boundary surface at the position of the lth speaker be n(r _l ), the area element centered on the position of the lth speaker be ΔS _l , and the drive signal for the lth speaker be d(r _l ),
an angle calculation unit that calculates an angle θ l between a vector from the position of the virtual sound source coordinate r _S to the position of the speaker coordinate r _l of the first speaker and the boundary surface inward normal vector n(r _{l )} of the first speaker, for each of the plurality of speakers;
a signal amplifier, a high pass filter, an attenuation adding unit, an area element adding unit, and a first angle weighting unit, each of which corresponds to one of the plurality of speakers;
The signal amplifier unit includes:
amplifying the sound source signal by a constant;
The HPF is
setting an HPFh(r _l , r _S ) based on the virtual sound source coordinates r _S and the speaker coordinates r _l of the first speaker, and performing a filtering process on the signal amplified by the signal amplifier using the HPFh(r _l , r _S );
The attenuation adding unit is
an absolute value of a vector from the position of the virtual sound source coordinate r _S to the position of the speaker coordinate r _l of the first speaker is obtained, the reciprocal of the result of squaring the absolute value is set as an attenuation coefficient 1/||r _l -r _S || ² , and the signal that has been subjected to the filtering process is multiplied by the attenuation coefficient 1/||r _l -r _S || ² ;
The surface area element adding unit
Multiplying the signal multiplied by the attenuation adding unit by the area element ΔS _l ;
The first angle weighting unit
an angle weight wC(θl) is set based on the angle _θl of the first speaker calculated by the angle calculation unit, the angle weight wC(θl) taking a value closer to 1 as the absolute value of the angle _θl approaches 0°, the value closer to 0 as the absolute value of the angle _θl approaches 90°, and the value 0 when the absolute value of the angle _θl is less than 0° or greater than 90°, in the range from 0 to 1; and the drive signal d( _rl ) for the first speaker is generated by multiplying the signal multiplied by the area element addition unit by _{the angle weight wC ( θl )} . 2. The object-based audio rendering apparatus of claim 1,
Furthermore, a VBAP coefficient calculation unit that identifies three speakers including the virtual sound source as three speakers that are targets of VBAP (Vector Based Amplitude Panning ₎ based on a unit vector of a virtual sound source direction toward the virtual sound source coordinate _rS and a unit vector of a speaker direction toward the speaker coordinate _r1 for all speakers, obtains unit vectors _rn1 , _rn2 , _rn3 of the three speakers that are targets of VBAP, calculates coefficients _g1 = _gn1 , _gn2 , _gn3 corresponding to the three speakers that are targets of VBAP based on the unit vector of the virtual sound source direction toward the virtual sound source coordinate _rS and the unit vectors rn1, _rn2 , _rn3 of the three speakers that are targets of VBAP, sets coefficient _g1 = 0 corresponding to speakers other than the three speakers that are targets of VBAP among all the speakers, and outputs the coefficient _g1 ;
a VBAP coefficient multiplier, a second angle weighting unit, and an adder, each corresponding to one of the plurality of speakers;
The VBAP coefficient multiplication unit
multiplying the signal multiplied by the area element adding unit by the coefficient _g output by the VBAP coefficient calculation unit;
The second angle weighting unit is
based on the angle _θl of the first speaker calculated by the angle calculation unit, an angle weight ws(θl) is set, which takes a value closer to 0 as the absolute value of the angle _θl approaches 0°, takes a value closer to 1 as the absolute value of the angle _θl approaches 90°, and takes a value of 0 when the absolute value of the angle _θl is less than 0° or greater than 90°, in a range from ₀ to 1; and generates a second drive signal by multiplying the signal multiplied by the VBAP coefficient multiplication unit by the angle weight _ws ( _θl ) _;
The first angle weighting unit
an angle weight _wC ( _θl ) is set, and a first drive signal is generated by multiplying the signal multiplied by the area element adding unit by the angle weight _wC ( _θl ), and the adder unit generates the drive signal d( _rl ) for the first speaker by adding the first drive signal and the second drive signal. 3. The object-based audio rendering apparatus of claim 2,
The VBAP coefficient multiplication unit
2. An object-based sound rendering apparatus comprising: an input of the sound source signal instead of the signal multiplied by the area element adding unit; and a multiplication unit for multiplying the sound source signal by the coefficient g _l . An object-based acoustic rendering apparatus according to any one of claims 1 to 3,
The HPF is
With a preset parameter as α, a preset filter order as N, and n=0, 1, . . . , N, the following equation is used:

Set the HPF coefficients h _n (r _l , r _S ) in
Let x[t] and y[t] be the input and output of the HPF at time t, and the input/output characteristics are expressed by the following equation:

and filtering the signal amplified by the signal amplifier using the HPF coefficients h _n (r ₁ , r _s ) so that: An object-based acoustic rendering apparatus according to any one of claims 1 to 4,
The attenuation adding unit is
With preset parameters as β ₁ and β ₂ , the following equation is used:

and multiplying the filtered _signal by the attenuation coefficient g(r _l , r _{S )} . A program for causing a computer to function as an object-based acoustic rendering device according to any one of claims 1 to 5.

Images