JP2003061200A - Sound processing apparatus and sound processing method, and control program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To use an impulse response measured in a real space so as to calculate a sound waveform in more detail with a calculation amount smaller than that of a conventional method. SOLUTION: The sound processing apparatus is provided with a space configuration data storage section 12 for storing data with respect to a configuration of a virtual space and with an impulse response segment data storage section 14 stored with segment data of an impulse response of a sound wave actually measured in a real space equivalent to the virtual space, an amplitude/delay calculation section 13 calculates respective amplitude/delay amounts for a direct response and a reflection response on the basis of the space configuration data, an impulse response composite section 15 generates a composite impulse response on the basis of the amplitude delay data produced by the amplitude/delay calculation section 13 and a convolution section 16 convolutes the composite impulse response to sound source sound data.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice processing device, a voice processing method, and a control program, and particularly to amplitude response data obtained by simulation and impulse response fragment data obtained by actually measuring real space. The present invention relates to a voice processing device and a voice processing method capable of expressing a realistic virtual space by using, and a control program for executing voice processing based on the voice processing method.

[0002]

2. Description of the Related Art A method of realizing a transfer function of impulse response between two points by actually modeling the space in an acoustic space having a particular sound is often used especially in the field of architectural acoustics. ing.

The impulse response is the acoustic transfer characteristic from the sound source to the sound receiving point. If there is a reflective object such as a wall in the acoustic space, in addition to the direct sound transfer response, the reflected sound transfer characteristic is included. Also, if there is an obstacle in the sound wave transmission path, the sound wave will be diffracted according to the size of the object.
It is also affected by the frequency response. Especially, if the sound receiving point is the position of the listener's ear, the head related transfer function (HRTF: He
ad Related Transfer Function) is applied.

For example, when building a concert hall or studio, software for simulating the acoustic effect at the time of completion at the design stage by using information such as the shape of the inside of the building and the material of the wall is already known. There is. Such software is designed by, for example, preparing a database of impulse responses created according to various situations actually recorded in an anechoic room, and convolving the impulse responses according to the situations with sound source audio data. It is possible to simulate how to hear a sound when it emits a sound at a specific position in a space.

[0005]

However, in general,
Since such software for acoustic simulation performs rigorous calculations using the “sound ray method”, the “virtual image method”, etc., it is difficult to generate an impulse response in real time.

On the other hand, there are many "digital reverb methods" as software that can generate a simulation of an acoustic space in real time by greatly simplifying the model. Some of them use the information of the wall material and the shape of obstacles to calculate the absorption rate and reflectance of sound waves, but none of them is a sampling of the actual sound, so the realism of the sound , Lacks in three-dimensional effect and realism.

Therefore, the present invention has been proposed in view of such a conventional situation, and by effectively using the fragment data of the impulse response obtained by actually measuring the real space, A voice processing device and a voice processing method capable of performing more detailed model calculation with a smaller amount of calculation than before and synthesizing an impulse response capable of expressing a realistic virtual space, and a control program for executing the voice processing method. The purpose is to provide.

[0008]

In order to achieve the above-mentioned object, a voice processing apparatus according to the present invention is provided with a sound source between a sound source in a virtual space and a sound receiving point for receiving a sound from the virtual sound source. In a voice processing device that generates a voice according to a positional relationship in a virtual space, a space configuration data storage unit that stores space configuration data regarding elements that configure the virtual space, and propagates from a virtual sound source to a sound receiving point. Amplitude delay calculation means for calculating amplitude delay data consisting of at least amplitude and delay of direct sound response of sound wave and main reflected sound response based on the data for space configuration, and virtual measured in real space corresponding to virtual space. From the impulse response of the sound wave propagating from the sound source to the sound receiving point, impulse response fragment data storage means in which impulse response fragment data extracted from a predetermined section is stored,
An impulse response synthesizing unit that generates a synthetic impulse response based on the amplitude delay data and the impulse response fragment data, and a convolution unit that convolves the synthetic impulse response synthesized by the impulse response synthesizing unit with the sound data of the sound source.

In such a voice processing apparatus, the amplitude delay calculation means spatially configures the amplitude delay data including at least the amplitude and delay of the direct sound response of the sound wave propagating from the virtual sound source to the sound receiving point and the main reflected sound response. Calculated based on the use data, the impulse response synthesizing means generates a synthetic impulse response based on the amplitude delay data and the impulse response fragment data, and the convolution means generates the synthetic impulse response synthesized by the impulse response synthesizing means. Fold it into the sound data of the sound source.

In such a voice processing device, the impulse response fragment data is obtained by extracting a predetermined section of the impulse response measured in the real space corresponding to the virtual space. Further, the reflected sound response includes at least a primary reflection or diffraction response to a reflector or an obstacle based on the spatial configuration of the virtual space.

Further, the voice processing device according to the present invention generates a voice corresponding to a positional relationship in the virtual space between a sound source in the virtual space and a sound receiving point for receiving the voice from the virtual sound source. In a voice processing device, impulse response fragment data obtained by extracting a predetermined section from an impulse response of a sound wave propagating from a virtual sound source measured in a real space corresponding to a virtual space to a sound receiving point is superimposed on the sound data of the sound source in advance. Sound data storage means for storing the superposed sound data, space configuration data storage means for storing space configuration data relating to elements forming the virtual space, and direct transmission of sound waves propagating from the virtual sound source to the sound receiving point. Amplitude delay calculation means for calculating amplitude delay data consisting of at least the amplitude and delay of the sound response and the main reflected sound response, based on the spatial configuration data,
And a synthesizing means for synthesizing the superimposed voice data based on the amplitude delay data.

In such a voice processing apparatus, the amplitude delay calculation means spatially configures the amplitude delay data including at least the amplitude and delay of the direct sound response of the sound wave propagating from the virtual sound source to the sound receiving point and the main reflection response. Calculated based on the data for the sound source, by the synthesizing means, the impulse response fragment data representing each of the direct sound response and the reflected sound response extracted from the impulse response of the sound wave propagating to the sound receiving point from the virtual sound source is converted into the sound data of the sound source. On the other hand, the superimposed voice data that has been superimposed in advance is synthesized based on the amplitude delay data.

Further, the voice processing method according to the present invention generates a voice corresponding to a positional relationship in the virtual space between a sound source in the virtual space and a sound receiving point for receiving the voice from the virtual sound source. In the voice processing method, the amplitude delay data consisting of at least the amplitude and delay of the direct sound response of the sound wave propagating from the virtual sound source to the sound receiving point and the main reflected sound response is based on the data for space configuration related to the elements forming the virtual space. From the impulse response of the sound wave propagating from the virtual sound source measured in the real space corresponding to the virtual space to the sound receiving point, the impulse response fragment data and the amplitude delay data extracted from the predetermined section are calculated. Based on the impulse response synthesizing step that generates a synthetic impulse response based on the And a Komu only convolution process.

According to such a voice processing method, in the amplitude delay calculation step, the amplitude delay data including at least the amplitude and delay of the direct sound response of the sound wave propagating from the virtual sound source to the sound receiving point and the main reflected sound response is obtained. Calculated based on the data for spatial configuration, in the impulse response synthesis step, the impulse response and the impulse response fragment data are synthesized to generate a synthetic impulse response, and in the convolution step,
The synthesized impulse response synthesized in the impulse response synthesis step is convoluted with the sound data of the sound source.

Further, the voice processing method according to the present invention generates a voice according to a positional relationship in the virtual space between a sound source in the virtual space and a sound receiving point for receiving the voice from the virtual sound source. In the voice processing method, the amplitude delay data consisting of at least the amplitude and delay of the direct sound response of the sound wave propagating from the virtual sound source to the sound receiving point and the main reflected sound response is based on the data for space configuration related to the elements forming the virtual space. From the impulse response of the sound wave propagating from the virtual sound source measured in the real space corresponding to the virtual space to the sound receiving point, the impulse response fragment data obtained by extracting the predetermined section is converted into the sound data of the sound source. A synthesizing step of synthesizing the superposed voice data which has been superposed in advance based on the amplitude delay data.

According to such a voice processing method, in the amplitude delay calculation step, the amplitude delay data consisting of the amplitude and delay of the direct sound response of the sound wave propagating from the virtual sound source to the sound receiving point and the main reflected sound response is stored in the space. Calculated based on the configuration data, in the synthesis process, from the impulse response of the sound wave propagating from the virtual sound source measured in the real space corresponding to the virtual space to the sound receiving point, impulse response fragment data to the sound data of the sound source Then, the superimposed voice data that has been superimposed in advance is synthesized based on the amplitude delay data.

In this voice processing step, it is preferable that the impulse response fragment data is obtained by extracting a predetermined section of the impulse response measured in the real space corresponding to the virtual space. The reflected sound response includes at least a primary reflection or diffraction response to a reflector or an obstacle based on the spatial configuration of the virtual space.

Further, the control program according to the present invention is a computer for generating a sound according to a positional relationship in the virtual space between a sound source in the virtual space and a sound receiving point for receiving the sound from the virtual sound source. In a control program of a controllable voice processing device, amplitude delay data including at least amplitude and delay of a direct sound response of a sound wave propagating from a virtual sound source to a sound receiving point and a main reflected sound response is related to an element forming a virtual space. Amplitude-delay calculation processing that is calculated based on the spatial configuration data, and impulse response fragment data that extract a predetermined section from the impulse response of the sound wave that propagates from the virtual sound source measured in the real space corresponding to the virtual space to the sound receiving point And the impulse delay data to generate a combined impulse response, and combined in the impulse fragment combination process. A convolution process of convoluting synthetic impulse response to a sound source of the audio data to be executed by the sound processing apparatus.

Further, the control program according to the present invention is a computer for generating a sound according to a positional relationship in the virtual space between a sound source in the virtual space and a sound receiving point for receiving the sound from the virtual sound source. In a control program of a controllable voice processing device, amplitude delay data including at least amplitude and delay of a direct sound response of a sound wave propagating from a virtual sound source to a sound receiving point and a main reflected sound response is related to an element forming a virtual space. Amplitude-delay calculation processing that is calculated based on the spatial configuration data, and impulse response fragment data that extract a predetermined section from the impulse response of the sound wave that propagates from the virtual sound source measured in the real space corresponding to the virtual space to the sound receiving point Performs on the voice processing device a synthesis process of synthesizing the superimposed voice data, which has been superposed on the voice data of the sound source, based on the amplitude delay data. To.

[0020]

BEST MODE FOR CARRYING OUT THE INVENTION An audio processing apparatus shown as a specific example of the present invention uses a multi-channel audio reproduction processing technology as described in Japanese Patent Application No. 2000-354530, for example, and a sound source and a virtual sound source in a virtual space. It is a voice processing device that generates voice according to a positional relationship in the virtual space with a sound receiving point that receives a voice from a sound source.

That is, the transfer characteristic (impulse response) from the virtual sound source is obtained for a plurality of sound receiving points arranged so as to surround the listener. By arranging the same number of sound reproducing means (speakers) at the positions corresponding to these sound receiving points and reproducing the sound source sound convolving the respective impulse responses from these speakers, it is as if a virtual sound source for the listener. It can be heard as if the sound image was localized at the position where was.

In particular, this speech processing apparatus comprises impulse response fragment data storage means for storing impulse response fragment data in which a predetermined section of impulse response measured in advance in a real space corresponding to a virtual space is stored, and the amplitude / delay The calculation unit calculates the amplitude delay data consisting of the amplitude and delay of the direct sound response of the sound wave propagating from the virtual sound source to the sound receiving point and the main reflected sound response, and the amplitude response delay unit calculates the amplitude delay. By generating a synthetic impulse response based on the data and the impulse response fragment data, and convolving the synthetic impulse response with the sound source speech data in the convolution unit, a more detailed model is calculated with a smaller amount of calculation than before, It realizes voice processing that can express a feeling of virtual space.

This sound processing device is designed to show how sound waves from a sound source existing in a certain place in a virtual space, particularly direct sound from the sound source and reflected sound from a wall / floor, are observed in other places in the same space. Can be simulated in real time. Therefore, especially 3D (3-Dimensional)
In a game, CG (Computer Graphics), or the like, when a sound source moves in a virtual space, if the way of hearing changes due to a change in the positional relationship with respect to the sound source, it can be expected to give a more realistic feeling.

A voice processing apparatus as a specific example of the present invention is
For example, it is premised that a real space corresponding to a virtual space and a real space prepared for each considered situation are actually set and the impulse response is acquired in this real space. Ideally, the impulse response at the time of this measurement should be able to separate the direct sound and the reflected sound. The measured impulse response is separated into direct sound and reflected sound as much as possible, and stored as a database for each fragment.

On the other hand, at the stage of modeling the space, the amplitude and delay of the direct sound and the reflected sound from the sound source to the sound receiving position (or microphone position) are calculated to create amplitude delay data used in the simulation. The impulse fragment in the database is combined with the part represented by the delay according to the amplitude. By including other information (how many times the sound has reached the listening point after being reflected, etc.) as a database, it is possible to make a voice expression that enhances the realism and presence.

Specific examples of the present invention will be described below in detail with reference to the drawings. Here, a case where the audio processing device as a specific example of the present invention is applied to, for example, a 3D game in which a sound source or a listener moves in a space represented by CG is considered.

The audio processing apparatus 1 shown in FIG. 1 has speakers 2a, 2b, 2c, 2d, 2e, 2f, 2f, 2b, 2c, 2d so that audio can be reproduced in a multi-channel system, for example, 8 channels.
It is equipped with 2g and 2h. Each speaker is arranged so as to form a closed curved surface 3 shown by a dotted line in FIG. 1, and is surrounded by each speaker by outputting a sound based on a sound signal from a sound signal control generator 4 whose details will be described later. Listeners (hereinafter referred to as listeners) located within the inner region of the closed curved surface 3
It will be referred to as a player if necessary. ) 100 can obtain a realistic sound effect according to the positional relationship between the virtual sound source position and each speaker arrangement position.

The audio processing apparatus 1 is equipped with a display 5 as a display unit for displaying an image, and the display 5
A video synchronized with the voice is displayed on the. More specifically, the display 5 is, for example, a so-called head mount display (Head Mount Displa).
y: HMD), and the video is displayed based on the video signal from the video control generator 6. Listener 100 is this H
It is possible to obtain a sound corresponding to the virtual space represented as a video while viewing the video in the 3D space displayed on the MD.

The sound processing apparatus 1 further changes the positional relationship between the virtual sound source and each speaker position in the virtual space displayed on the display 5, that is, the operation for changing the positional relationship between the listener 100 and the virtual sound source. It has a section 7. The operation unit 7 may be controlled by a program, may be operated by a third party other than the listener 100, or may be operated and input via the controller 7a that can be input by the listener 100. May be

The audio signal control generation unit 4, the video control generation unit 6 and the operation unit 7 are centrally controlled by a CPU (not shown).

In the audio system having the audio signal control generator 4 and the output of a plurality of channels like the audio processing device 1, the listener 100 located inside the closed curved surface 3 has a sound source as if the virtual space were at a certain point. You can create a sound field like that. Further, by smoothly exchanging a plurality of impulse responses from the sound source to each speaker position, the sound source and the listener 1 can be moved when the sound source moves or when the listener 100 moves in the virtual space.
The relative change in the positional relationship with 00 can be reflected in the voice.

FIG. 2 shows an audio processing device 1 in which the concrete configurations of the audio signal control generator 4 and the video control generator 6 are clarified.
Shown in.

The voice signal control generator 4 is voice data such as voice and sound effects depending on a scene and a situation, and is voice data of a sound source (hereinafter, referred to as unprocessed original voice data).
It is referred to as sound source voice data. ) Is stored, a spatial configuration data storage unit 12 that stores data relating to the configuration of the virtual space, an attenuation amount and time of the amplitude of the direct sound and the reflected sound based on the spatial configuration data. An amplitude / delay calculation unit 13 for calculating a delay, and an impulse response fragment data storage unit in which an impulse response of a sound wave actually measured in a real space corresponding to a virtual space is fragmented and stored into a direct sound portion and a reflected sound portion. 14, an impulse response synthesizing unit 15 for synthesizing a synthetic impulse response based on the child clothing delay data and the impulse response fragment data generated by the amplitude / delay calculating unit 13, and the impulse response synthesizing unit 15 The convolution unit 16 that convolves the synthesized impulse response with the sound source speech data, and the speech processing when outputting the speech data with the synthesized impulse response superimposed from each speaker. And a sound processing unit 20 for performing, which are connected by an internal bus 19. Further, a D / A (Digital to Analog) unit 17 and an amplifier 18 for outputting a sound signal from the sound processing unit 20 are provided.

The video control generator 6 is, as shown in FIG.
For example, an object display control unit 21 for controlling a display positional relationship between an object which is a virtual sound source in the virtual space and the listener 100 in the virtual space, and an image signal processing unit 22 are provided.

In this specific example, a virtual space as shown in FIG. 3 is assumed, and a case where a sound from a virtual sound source 41 in this virtual space 40 is expressed is shown. A wall 42 indicated by a wavy line is assumed in the virtual space 40. Therefore, listener 10
When 0 generates a sound wave output from each speaker so that a sound image localization feeling is obtained at the position of the virtual sound source 41, in this specific example, the direct wave 51 from the virtual sound source 41 and the primary wave with respect to the wall 42 of the virtual sound source 41 are generated. Considering the influence of the reflected waves 52a and 52b and the secondary reflected wave 53 on the wall 42 of the virtual sound source 41, first, the generation of the voice waveform output from the speaker 2g will be described.

More specifically, the spatial configuration data storage unit 12 in the audio signal control generation unit 4 stores data representing the spatial configuration of the virtual space 40. In the present specific example, the spatial configuration data is a component in which a sound wave from the virtual sound source to the sound receiving point in the virtual space 40 is affected,
A data structure such as position and shape dimensions. For example, the wall surface that reflects sound waves is represented by several three-dimensional coordinate values (polygon data). This data structure may include data such as reflectance (absorption rate) and transmittance. In this specific example, the virtual space is configured with simplified or simplified elements, and only the components that can obtain only the direct sound from the sound source and the main reflected sound are considered.

The amplitude / delay calculation unit 13 attenuates the direct sound response of the sound wave propagating to the speaker 2g as the transmission response from the virtual sound source 41 and the time amplitude and delay of the reflected wave with respect to the direct wave, based on the spatial configuration data. The degree is calculated as an amplitude / delay model (hereinafter referred to as amplitude delay data). The amplitude / delay calculation unit 13 includes a virtual sound source 41, a wall 42,
Based on the spatial configuration data indicating the positional relationship between the speakers 2g, the sound wave is linearly regarded as a path from the virtual sound source 41 to the speaker 2g, and the sound wave from the virtual sound source 41 is a wall 42.
Then, the path to the speaker 2g and the degree of attenuation of the amplitude of the sound wave when reflected by are calculated.

In the present invention, the amplitude delay data includes
In addition to the data of the amplitude and delay of the direct sound response and the amplitude and delay of the reflected sound response described above, it is desirable to have information indicating the paths of the respective direct sounds and reflected sounds. This information can include, for example, information that the wall 42a is incident at an incident angle of 60 ° and that it is diffracted by an obstacle X (not shown), and is appropriate from the impulse response fragment data storage unit 14 described later. It can be used as a search key when selecting various impulse response fragment data.

The amplitude / delay calculation unit 13 calculates the amplitude / delay by using the so-called "virtual image method". Here, for the sake of simplicity, it is assumed that the order, which means how many times the sound ray is reflected, is cut off at a low order such as the second order. Whether or not a higher order calculation needs to be performed can be freely selected in consideration of the load of hardware such as a CPU and the magnitude of a higher-order reflected wave. For example, if the load of image processing presented to the user is large, it is possible to terminate the calculation at a lower order.

FIG. 4 shows a general process of calculating the ray of the reflected sound using the virtual image method. FIG. 4A shows a case where the path of the primary reflected wave is obtained, and FIG. 4B shows a case where the path of the secondary reflected wave is obtained. In the virtual image method, for walls W ₁ , W ₂ , and W _{3 on} which sound rays from sound waves from a virtual sound source are reflected, by obtaining a “virtual image” at a symmetrical position of the walls, all reflected sound paths are obtained. Find L.

That is, in the case of the primary reflected wave,
Wall W in₁Against virtual sound source P₁Virtual image source P₁’
I assume. In the case of the secondary reflected wave, in the virtual space
Wall W _TwoAnd W_ThreeAgainst virtual sound source P_TwoVirtual image source
P_Two’、 P_Two"Assuming that you can split the path
For example, you can know the distance from the route length, and
You can know the extra time. Also, at the same time, there is an attenuation factor
It is possible to calculate the attenuation depending on the distance by assuming
You can Furthermore, the reflectance and sound absorption of each wall, etc.
The attenuation amount may be made more precise by including.

The method of calculating the amount of attenuation of each reflected wave and the time delay is not particularly limited. In addition to the method described above, for example, a parameter indicating the degree of attenuation according to the reflection angle with respect to the wall may be used.

In this way, the amplitude / delay calculation unit 13
Generates amplitude delay data as shown in FIG. Here, only the direct wave 51, the primary reflected wave 52a for the wall surface 42a, and the secondary reflected wave 53 for the wall surface 42a and the wall surface 42b in FIG. 3 are considered.

In FIG. 5, 0 indicates the time when a certain sound (impulse response measurement signal) is generated from the virtual sound source 41. Direct wave 5 with the passage of time along the time axis T
1, the signals corresponding to the primary reflected wave 52a and the secondary reflected wave 52 are sequentially displayed. t ₁ indicates a period until the direct wave 51 is detected at the position of the speaker 2g in the virtual space 40, and A ₁ indicates the amplitude of the direct wave 51. Similarly, the signal S ₂ indicates the time delay t ₂ and the amplitude A ₂ of the primary reflected wave 52a detected at the position of the speaker 2g, and the signal S ₃ is the secondary reflected wave 53 detected at the position of the speaker 2g. The time delay t ₃ and the amplitude A ₃ are shown.

In the impulse response fragment data storage unit 14, in the impulse response from the virtual sound source to each speaker arrangement position, the portion corresponding to the direct sound and the portion corresponding to the reflected sound correspond to the real space corresponding to the virtual space. It is stored as impulse response fragment data obtained by extracting a predetermined section of the measured value of the impulse response measured in. In this specific example, it is premised that the impulse response is acquired in advance in the real space corresponding to the virtual space and the real space prepared for each situation to be considered as described above.

For example, the impulse response fragment data of the reflected sound of the wall of a certain virtual space is obtained by measuring the actual impulse response in the real space corresponding to this virtual space, as shown in FIG. Real space 60 shown in FIG.
At, the sound wave generated by the sound source 61 existing at the position P is acquired by the microphone 62.

The impulse response acquired here is shown in FIG. From the impulse response shown in FIG. 7, a portion corresponding to the direct sound is extracted as impulse response fragment data D ₁ and a portion corresponding to the reflected sound is extracted as impulse response fragment data D ₂ . Here, each impulse response fragment data is divided by a time window in which the amplitude of the voice waveform gradually attenuates in the latter half of the extracted section, for example. Since voice processing is performed based on the impulse response in the real space,
The voice processing device 1 can reproduce a more natural voice waveform.

The number of sound wave samples may be the lowest score that can express the reverberation characteristic. In some cases, the time window may be such that the amplitude of the voice waveform gradually attenuates in the first half of the extracted section.

The impulse response synthesizing unit 15 generates a synthetic impulse response based on the amplitude delay data and the impulse response fragment data stored in the impulse response fragment data storage unit 14. FIG. 8 shows a combined impulse response combined by the impulse response combining unit 15. The impulse response synthesis unit 15 synthesizes the impulse response fragment data of the direct sound, the first-order reflected sound, and the second-order reflected sound while maintaining the relationship between the amplitude and the time delay of the amplitude delay data shown in FIG. Amplitude delay data, that is, a synthetic impulse response is generated. In the example of FIG. 8, the impulse response fragment data corresponding to the secondary reflection also uses the impulse fragment data D ₂ similarly to that of the primary reflected sound.

The synthesized impulse response synthesized by the impulse response synthesis section 15 is convolved by the convolution section 16 with the sound source speech data stored in the speech data storage section 11.

The voice processing unit 20 sends to the listener 100 based on the above-mentioned synthesized impulse response according to the change in the positional relationship between the virtual sound source and each speaker in the virtual space input from the operation unit 7 (controller 7a). Crossfade processing is performed to smoothly change the sound that can be heard. Details regarding the crossfade processing will be described later. In the audio processing device 1, since the crossfading process can be performed in the audio processing unit 20, if some representative points in the virtual space 40 are selected and the spatial configuration data and the impulse response fragment data regarding the position are prepared. Well, it is not necessary to prepare each data for every position.

In the above example, the sound output from the speaker 2g has been described, but the sound processing device 1
Since the 8-channel multi-channel method is adopted, the transfer characteristic from the virtual sound source 41 to each of the speakers 2a to 2h is formed by combining the combined impulse response with each speaker in the same manner as described above. is doing. At the time of reproduction, the synthesized impulse responses thus formed are convoluted with the sound source sound data and sound is emitted from the speakers 2a to 2h, so that the listener 100 has a sense of localization as if the virtual sound source 4 is located. Obtainable. Therefore, the amplitude / delay calculation unit 13 obtains the amplitude and time delay of each response part for all channels.

As described above, the voice processing device 1 has a positional relationship between the listener 100 and the virtual sound source 41 in the virtual space 40.
Alternatively, when the positional relationship between the virtual sound source 41 and each speaker is changed by the operation of the controller 7a or the like, for example, even when the virtual sound source is moving, a sound wave from the position of the virtual sound source that changes moment by moment is generated. The virtual sound source 41 can be smoothly switched and expressed as if it were moving. When the listener 100 moves in the virtual space 40, or when the virtual sound source 41
A process in which the voice processing device 1 generates a sound field when the player moves will be specifically described with reference to FIG.

In step S1, the voice processing apparatus 1 determines whether or not the position movement of the player or the virtual sound source 41 has been input via the operation unit 7 (controller 7a). When the position movement is input, the process proceeds to step S2.

In step S2, the voice processing device 1
In the amplitude / delay calculation unit 13, the spatial configuration data of the position information of the moved virtual sound source 41 is read from the spatial configuration data storage unit 12, and the direct sound and the reflected sound from the virtual sound source 41 to each speaker arrangement position are generated. A sound ray is calculated and amplitude delay data is generated. The sound ray is a straight line representing the path of the direct sound and the reflected sound from the virtual sound source 41 to each speaker arrangement position, and the reflected sound includes at least the primary reflected sound.

Next, in step S3, the impulse response synthesizer 15 of the voice processing apparatus 1 uses the information that the sound ray is reflected or not reflected by the wall 42 in the virtual space 40 as a key, and the impulse response fragment data storage unit is used. The impulse response fragment data is extracted from 14.

In step S4, the impulse response synthesizing unit 15 of the voice processing device 1 calculates the attenuation degree and the time delay of the amplitude of the direct wave, the primary reflected sound and the secondary reflected sound generated from the sound wave path in step S2. The impulse response fragment data extracted in step S3 is combined with the indicated amplitude delay data. The combined impulse response obtained by combining is set as the final impulse response in step S5.

In step S6, the voice processing device 1 convolves the current synthesized impulse response with the voice source voice data.

If the position movement is not input in step S1, the process proceeds to step S6, and the synthesized impulse response at that time point is convoluted with the sound source voice data.

As described above, by the series of processing shown in FIG.
The sound processing device 1 can smoothly switch the sound waves from the virtual sound source 41, which changes moment by moment, and express the virtual sound source 41 as if it were moving.

If the position of the virtual sound source 41 and the spatial structure of the virtual space 40 are equal, the same combined impulse response is combined every time. Therefore, by providing a temporary storage unit (cache mechanism) for the composite impulse response, the composite impulse response of the same condition in the temporary storage unit is searched from the position information received from the operation unit 7 or the controller 7a and is temporarily stored. In that case, a series of processes from the calculation of the amplitude delay data in the amplitude / delay calculation unit 13 to the combination of the impulse response fragment data in the impulse response combination unit 15 can be omitted.

In this specific example, the sound waveform from the virtual sound source 41 is reproduced at each speaker position, but it may be considered as a sound waveform reproduced at the ear position of the listener 100. That is, even when the sound image localization processing is performed based on the head related transfer function from the sound source to both ears of the listener, the above-described method of combining the combined impulse responses can be applied to this head related transfer function.

In the above example, the number of reflections on the wall 42 can be changed according to the basic performance such as the CPU and memory capacity of the voice processing device. Along with this, the spatial position information prepared in advance in the spatial configuration data storage unit 12 and the impulse response fragment data prepared in advance in the impulse response fragment data storage unit 14 can also be freely set according to the performance of the voice processing device.

In the above example, the amplitude / delay calculation unit 13
Only calculates the amplitude attenuation and time delay of the direct wave and the reflected wave, and the impulse response fragment data considers only the direct wave, the first-order reflected wave, or the second-order reflected wave. When the database of the response fragment data storage unit 14 is designed, more detailed information can be input by using a search key so that the final synthesized impulse response can be reproduced more faithfully.

For example, as shown in FIG. 10 and FIG.
The amplitude delay data has a data structure including additional information indicating the reflection condition of the sound in the virtual space 40, such as the incident angle of the sound wave from the virtual sound source 41 to the constituent elements forming the virtual space, the material of the constituent elements, and the like. The response fragment data storage unit 14 may be a database having a plurality of kinds of fragment data depending on the type of wall and the order of reflection. In this case, the amplitude / delay calculation unit 13 first makes the secondary reflected wave 53 in the sound ray of the reflected sound shown in FIG. 3 incident on the wall 42a and then reflects it on the wall 42b. In consideration of reaching the listening position, the amplitude attenuation and time delay of the sound wave are calculated. That is, the calculation is performed in consideration of the reflectance and sound absorption coefficient of the wall.

The amplitude and the time delay can be calculated from the attenuation rate and the speed of sound when traveling through the air when obtaining the sound ray. The impulse response fragment database also includes data obtained by reflecting the impulse response fragment data of the direct wave and the impulse response fragment data of the reflected wave twice on the wall 42a or the wall 42b.

Therefore, for the sound ray of the reflected wave 53 shown in FIG. 3, "data reflected on the wall 42a and subsequently on the wall 42b" is used. Here, although the incident angle is not used, for example, "...
It is also possible to create a more detailed database such as "data reflected at an incident angle of ○ °". In this case, the data including the incident angle x is retrieved from the database. In this way, the measurement data is fragmented and held for each fine condition, whereby the formation of a synthetic impulse response having various timbres can be realized.

Further, regarding the impulse response fragment data corresponding to the direct wave, the database is classified in detail such as "impulse response fragment data when an obstacle exists between the virtual sound source and the sound receiving position". be able to. When there is an obstacle on the sound ray, a plurality of data representing the diffraction effect may be prepared depending on the size and transmittance of the obstacle. However, in this case, it is necessary to store an item of additional information indicating whether or not Otomi has arrived through an obstacle even in the amplitude delay data.

As described above, it is preferable that the impulse response fragment data storage unit 14 is designed to closely cooperate with the amplitude delay data calculated by the amplitude / delay calculation unit 13.

As the voice processing section for executing the above-mentioned cross fade processing, for example, a well-known voice processing section having the configuration shown in FIG. 12 can be used. Voice processing unit 2
The value of 0 makes it possible to reproduce a complicated sound field by gradually switching while superimposing a plurality of voice transfer characteristics. The voice processing unit 20 performs a crossfade process on the voice signal S _i input from a voice signal input unit (not shown). As a sound source corresponding to the voice signal input unit, for example, a linear recording medium, An audio signal output unit or the like in a reproducing device that reproduces an audio signal recorded on a disc-type recording medium or the like corresponds to this.

The voice processing unit is provided at the subsequent stage of the voice signal input unit.
A plurality of linear filters 31 ₍₁₎ , 31 ₍₂₎ , ...
, 31 _(N) are arranged in parallel, and variable weighting units 32 _{( 1)} , 32 ₍₂₎ , ... Corresponding to the number of linear filters are provided at the subsequent stages of the respective linear filters.
, 32 _(N) are arranged in parallel, and one adder circuit 33 is connected to the rear stage of these variable weighting units. The audio processing unit 20 shown in FIG. 12 shows only one channel, but in this specific example, since it is a multi-channel system of 8 channels, this audio processing unit 20 has eight channels.
It is prepared as a set. Signal S after crossfade processing
_o is taken out from the output terminal of the adder circuit 33, and D / A
Is supplied to the section 17.

Each linear filter is connected to a terminal of the audio signal input section at the preceding stage through a common contact, and is wired and connected so that output signals from the audio signal input section are input in parallel. Each linear filter superimposes a filter coefficient D _f transferred from a controller 7a, which will be described later, on the input audio signal S _i , and the superposed signal S _(
1) , S ₍₂₎ , ..., S _(N) are output to the variable weighting unit 32 in the subsequent stage. Here, the filter coefficient D _f is, for example, a coefficient for changing the signal level or phase of the input signal S _i , and represents the acoustic transfer characteristic. The filter coefficient is an impulse response obtained by simulation or actual measurement in order to represent a virtual sound source (hereinafter, referred to as a sound image, if necessary) as an arbitrary position to be localized in a virtual space. The waveform is shown.

Each variable weighting unit is connected to a linear filter corresponding to each of the preceding stages, and based on a command signal S _c from the controller 7a, a fade-in process or a superimposition signal S from the corresponding linear filter is connected. Fade out processing is being performed.

On the other hand, on the basis of the movement request signal S _d of the sound image (virtual sound source) supplied from the operation unit 7, the variable weighting unit which previously performed the fade-in process is instructed to perform the fade-out process. The command signal S _{c 2} for outputting the signal S _c1 and performing the fade-in process on the variable weighting unit that previously performed the fade-out process.
Is output. Further, the control unit 34 outputs a command code C _s including sound image position data in response to the output of the command signal S _c .

The filter coefficient switching section 35 is arranged to
Linear filter to which the previous filter coefficient Df was transferred
Filter coefficient D returned to a linear filter unit other than_f
To transfer. This causes the signal S_iIs each linear fill
, The filter coefficient D _fSignal S on which is superimposed
And the superimposed signal S from these linear filter units is
Fade-in process or fade door
It is processed and output.

[0076] Specifically, the signal Si is the linear filter unit the previous filter coefficient D _f is transferred, the last filter coefficient D _f is superimposed processed, this newly from the filter coefficient switching unit 35 In the linear filter unit to which the filter coefficient D _f is transferred, the current filter coefficient D _f
Will be superposed.

As a result, the fade-out processing is performed in the corresponding variable weighting section by the superimposed signal S from the linear filter section to which the previous filter coefficient D _f has been transferred, and the linear filter to which the current filter coefficient D _f has been transferred. Fade-in processing is performed in the corresponding variable weighting section by the superimposed signal S from the filter section. The output signal from each variable weighting unit is subjected to addition processing by the addition circuit 33 in the subsequent stage, and is supplied to the D / A unit 17 as the final output signal, that is, the crossfade processing signal S _o .

Therefore, by the above-described processing, the voice processing device 1 can realistically represent the positional relationship between the virtual sound source and the listener 100 in accordance with the situation displayed on the display 5. In other words, listener 100
For you, the sound source sounds as if it were actually there. The sound processing device 1 is configured so that a sound source moves in a virtual space with respect to the listener 100, moves with respect to the sound source of the listener 100, or changes in surrounding conditions, for example, relative to a wall, a floor, an obstacle, or the like where sound waves from the sound source are reflected. It can be expressed as if the positional relationship between the sound source and the listener 100 and the transmission characteristic of the sound wave are actually changed in response to the movement.

In the concrete example described above, the voice waveform output from the speaker 2 by the process of finally taking out the combined impulse response formed on the basis of the amplitude delay data and the impulse response fragment data and convolving this into the input voice signal. However, for example, when the amplitude / delay calculation unit 13 can handle only a very small amount of reflected wave data due to restrictions on the capability of the device, impulse response fragment data storage is performed for the same reason. If the amount of impulse response fragment data that can be stored in the unit 14 is limited,
It is also possible to implement the same processing as the specific example described above by another specific example shown below.

The sound processing device 70 shown in FIG. 13 is similar to the sound processing device 1 in that it has a positional relationship in the virtual space between a sound source in the virtual space and a sound receiving point for receiving the sound from the virtual sound source. 1 is a voice processing device that generates a voice according to FIG.
By arranging the respective components according to the positional relationship as shown in, it is possible to obtain a realistic sound effect.
With respect to the configuration of the voice processing device 70 having the same functions as those of the voice processing device 1, the same numbers as in FIG. 1 are assigned and detailed description thereof is omitted.

The basic configuration of the voice processing device 70 is the same as that of the voice processing device 1. However, among the measured values of the impulse response from the sound source measured in the real space corresponding to the virtual space to the respective speaker positions, Voice data in which superimposed voice data in which impulse response fragment data obtained by extracting a predetermined section corresponding to a portion corresponding to a sound and a portion corresponding to a reflected sound are preliminarily superimposed on each sound source voice data is stored A feature is that the storage unit 71 is provided, and the voice data synthesizing unit 72 synthesizes the superimposed voice data with the amplitude delay data.

FIG. 14A shows the voice data storage section 71.
For the voice waveform of the sound source voice data shown in FIG.
(B) to be superimposed previously Found impulse response fragment data D ₂ of the extracted direct wave impulse response fragment data D ₁ and the reflected wave from each shown, shown in FIG. 14 (c), FIG. 14 (d) It is stored as such superimposed voice data. Here, since the superimposed voice data is stored in a database so that the data regarding the time delay is minimized, it can be made much shorter than the amplitude delay data synthesized by the method described above. That is, the amount of amplitude delay data is small, and a resource for storing the data is not required so much.

The voice data synthesizing unit 72 synthesizes voice data in correspondence with the amplitude delay data from the amplitude / delay calculating unit 13. That is, the amplitude-delayed data as shown in FIG. 15A is combined with the superimposed voice data shown in FIG. 15B to obtain the voice waveform shown in FIG. 15C.

According to the voice processing device 70, the period length up to the delay time of the reflected wave that arrives most late is shortened, so that the data capacity is saved and the real time property of voice processing is improved.

As described above, according to the voice processing device 1 and the voice processing device 70 according to the present invention, when forming the synthetic impulse response or the synthetic voice data, not all are calculated but calculation is performed. By effectively separating the part and the part that uses impulse response fragment data obtained from actual measurements in the real space, it is possible to acoustically create a more realistic virtual space with a smaller amount of calculation than when all are calculated. Can be expressed as Moreover, even when creating a sound field when a virtual sound source moves in the virtual space, it is possible to combine the measurement and calculation of several points without measuring the impulse response at all points where the virtual sound source moves in the real space. It is possible to more simply express a more realistic sound field by using the sound field.

It should be noted that the present invention is not limited to the above-described embodiments, and it goes without saying that various modifications can be made without departing from the gist of the present invention.

[0087]

As described in detail above, in the voice processing device according to the present invention, when the voice waveform for reproducing the sound field is generated, the amplitude delay calculation means propagates from the virtual sound source to the sound receiving point. Amplitude delay data consisting of at least amplitude and delay of direct sound response of sound wave and main reflected sound response,
By effectively separating into a part that is calculated based on the data for spatial configuration and a part that uses the impulse response fragment data that has extracted a predetermined section of the impulse response measured in the real space corresponding to the virtual space, all A more realistic acoustic space can be reproduced with a smaller amount of calculation than the case where is calculated, and the final improvement of sound quality can be achieved at the same time.

Further, according to the speech processing apparatus of the present invention, when simulating a state where a sound source moves in a virtual space, the sound source can be generated by inputting only a few synthetic impulse responses characteristic in the real space. It is possible to realistically express the situation where a person moves freely in this virtual space.

Further, by reducing the amount of calculation, it is possible to reproduce a realistic sound field even in a situation where real time processing is required for voice processing such as a game.

Further, in the audio processing device according to the present invention, the amplitude delay calculation means stores the amplitude delay data consisting of the direct sound response of the sound wave propagating from the virtual sound source to the sound receiving point and the amplitude and delay of the main reflected sound response. Impulse response fragment data representing each of the direct sound response and the reflected sound response calculated from the configuration data and extracted from the impulse response of the sound wave propagating from the virtual sound source to the sound receiving point is superimposed on the sound data of the sound source in advance. By synthesizing based on the superimposed voice data and the amplitude delay data, it is possible to reproduce a more realistic acoustic space with a smaller amount of calculation than when all are calculated, and at the same time improve the final sound quality. it can.

Furthermore, by reducing the amount of calculation, it is possible to reproduce a realistic sound field even in a situation where real-time processing is required for voice processing such as a game.

Further, according to the speech processing apparatus of the present invention, when simulating the state where the sound source moves in the virtual space, the sound source can be measured by measuring a few points of the characteristic impulse response in the real space. It is possible to realistically represent the situation of freely moving in this virtual space.

Further, the voice processing method according to the present invention, when generating the voice waveform for reproducing the sound field, has a direct sound response of the sound wave propagating from the virtual sound source to the sound receiving point in the amplitude delay calculation step and the main sound response. A part that calculates amplitude delay data consisting of at least the amplitude and delay of the reflected sound response based on the data for spatial composition, and impulse response fragment data that extracts a predetermined section of the impulse response measured in the real space corresponding to the virtual space. By effectively separating it from the part to be used, a more realistic acoustic space can be reproduced with a smaller amount of calculation than the case where all are calculated, and the final improvement in sound quality can be achieved at the same time.

Further, according to the voice processing method of the present invention, when simulating a state in which a sound source moves in the virtual space, the sound source can be measured by measuring only a few characteristic impulse responses in the real space. It is possible to realistically represent the situation of freely moving in this virtual space.

Further, by reducing the amount of calculation, it is possible to reproduce a realistic sound field even in a situation where real time processing is required for voice processing such as a game.

Further, according to the voice processing method of the present invention, in the amplitude delay calculation step, an amplitude delay including at least the amplitude and delay of the direct sound response of the sound wave propagating from the virtual sound source to the sound receiving point and the main reflected sound response. Data is calculated based on the spatial composition data, and impulse response fragment data representing each of the direct sound response and the reflected sound response extracted from the impulse response of the sound wave propagating from the virtual sound source to the sound receiving point is compared with the sound data of the sound source. By combining the pre-superimposed superposed voice data with the amplitude delay data, it is possible to reproduce a more realistic acoustic space with a smaller amount of calculation than when calculating all and to improve the final sound quality. Can be achieved at the same time.

Furthermore, by reducing the amount of calculation, it is possible to reproduce a realistic sound field even in a situation where real-time processing is required for voice processing such as a game.

Further, according to the voice processing method of the present invention, when simulating a state in which a sound source moves in the virtual space, the sound source can be measured by measuring only a few characteristic impulse responses in the real space. It is possible to realistically represent the situation of freely moving in this virtual space.

Further, according to the control program of the present invention, when a voice waveform for reproducing a sound field is generated,
A portion for calculating the amplitude delay data consisting of at least the amplitude and delay of the direct sound response of the sound wave propagating from the virtual sound source to the sound receiving point and the main reflected sound response by the amplitude delay calculation processing, and the virtual space. By effectively separating the predetermined section of the impulse response measured in the real space into the part that uses the extracted impulse response fragment data, it is more realistic with a smaller amount of calculation than when calculating all. While it can reproduce various acoustic spaces,
The final improvement in sound quality can be achieved at the same time.

Further, according to the control program of the present invention, when simulating the state in which the sound source moves in the virtual space, the sound source can be measured by measuring a few points of the characteristic impulse response in the real space. It is possible to realistically represent the situation of freely moving in the virtual space.

Further, by reducing the amount of calculation, it is possible to reproduce a realistic sound field even in a situation where real time processing is required for voice processing such as a game.

According to the control program of the present invention, the amplitude delay data including at least the amplitude and delay of the direct sound response of the sound wave propagating from the virtual sound source to the sound receiving point and the main reflected sound response in the amplitude delay calculation process. Is calculated based on the spatial composition data, and impulse response fragment data representing each of the direct sound response and the reflected sound response extracted from the impulse response of the sound wave propagating from the virtual sound source to the sound receiving point is compared with the sound data of the sound source. By synthesizing the pre-superimposed superimposed voice data, it is possible to reproduce a more realistic acoustic space with a smaller amount of calculation as compared with the case where all of them are calculated, and at the same time, improve the final sound quality.

Furthermore, by reducing the amount of calculation, it is possible to reproduce a realistic sound field even in a situation where real time processing is required for voice processing such as a game.

Further, according to the control program of the present invention, when simulating the state in which the sound source moves in the virtual space, the sound source can be measured by measuring a few characteristic impulse responses in the real space. It is possible to realistically represent the situation of freely moving in the virtual space.

[Brief description of drawings]

FIG. 1 is a configuration diagram showing an outline of a configuration of a voice processing device shown as a specific example of the present invention.

FIG. 2 is a configuration diagram illustrating a configuration of a voice processing device shown as a specific example of the present invention.

FIG. 3 is a diagram illustrating a positional relationship between a virtual sound source, a speaker, and a listener in a virtual space represented by a voice processing device shown as a specific example of the present invention.

FIG. 4A is a schematic diagram for explaining how to calculate a path of a primary reflected sound between a virtual sound source and a sound receiving point by the virtual image method, and FIG. 4B is a virtual image method. It is a schematic diagram explaining a mode that the path of the secondary reflected sound between the virtual sound source and the sound receiving point is calculated by.

FIG. 5 is a diagram showing amplitude / delay data calculated based on spatial configuration data in an amplitude / delay calculation unit of a voice processing device as a specific example of the present invention.

FIG. 6 is a diagram illustrating acquisition of impulse response fragment data stored in an impulse response fragment data storage unit of a voice processing device as a specific example of the present invention.

FIG. 7 is a diagram showing impulse response fragment data stored in an impulse response fragment data storage unit of a voice processing device shown as a specific example of the present invention.

FIG. 8 is a schematic diagram showing how impulse response fragment data is synthesized based on amplitude / delay data in a voice processing device as a specific example of the present invention.

FIG. 9 is a flowchart illustrating a process in which a voice processing device shown as a specific example of the present invention generates a voice waveform of a virtual sound source moving in a virtual space.

FIG. 10 is a schematic diagram showing an example of a table of amplitude delay data calculated by an amplitude delay calculation unit of the audio processing device shown as a specific example of the present invention.

FIG. 11 is a diagram showing another example of impulse response fragment data stored in an impulse response fragment data storage unit of the audio processing device shown as a specific example of the present invention.

FIG. 12 is a configuration diagram illustrating a configuration of an audio processing unit in an audio processing device shown as a specific example of the present invention.

FIG. 13 is a configuration diagram illustrating a configuration of a voice processing device shown as another specific example of the present invention.

14A is a waveform diagram showing a voice waveform of sound source voice data, and FIG. 14B is a waveform diagram showing a waveform of impulse response fragment data extracted from an actual measurement value. FIG. 14C is a waveform diagram showing a voice waveform in which impulse response fragment data indicating a direct wave is combined with the voice waveform of the voice source voice data, and FIG. 14D is a voice waveform of the voice source voice data. FIG. 9 is a waveform diagram showing superimposed voice data obtained by synthesizing impulse response fragment data indicating a reflected wave.

15 (a) is a diagram showing amplitude delay data calculated based on spatial configuration data, and FIG.
FIG. 15B is a waveform diagram showing superimposed voice data constructed based on amplitude delay data, and FIG. 15C is a waveform diagram of FIG.
It is a waveform diagram which shows the audio | voice waveform obtained by synthesize | combining each audio | voice data of 5 (b).

[Explanation of symbols]

1, 70 voice processing device, 2 speaker, 3 closed curved surface,
4 audio signal control generation unit, 5 display, 6 video signal control generation unit, 7 operation unit, 7a controller, 1
1 voice data storage unit, 12 spatial configuration data storage unit, 13 amplitude / delay calculation unit, 14 impulse response fragment data storage unit, 15 impulse response synthesis unit, 16
Folding unit, 17 D / A unit, 18 amplifier, 19 internal bus, 20 audio processing unit, 21 object display control unit,
22 image signal processing unit, 31 linear filter, 32 variable weighting unit, 33 adder circuit, 34 control unit, 35
Filter coefficient switching unit, 36 filter coefficient generating unit, 40
Virtual space, 41 virtual sound source, 41 wall, 51 direct wave, 52a primary reflected wave, 52b primary reflected wave, 53 secondary reflected wave, 60 real space, 61 sound source, 62 microphone, 71 voice data storage section, 72 voice data synthesis Part, 100 listeners

Claims (16)

[Claims] 1. A sound processing device for generating a sound according to a positional relationship in a virtual space between a sound source in the virtual space and a sound receiving point for receiving a sound from the virtual sound source, Spatial configuration data storage means in which spatial configuration data relating to constituent elements are stored, and an amplitude composed of at least an amplitude and a delay of a direct sound response of a sound wave propagating from the virtual sound source to the sound receiving point and a main reflected sound response. From the impulse response of the sound wave propagating from the virtual sound source measured in the real space corresponding to the virtual space to the sound receiving point, a predetermined value Based on the impulse response fragment data storage means for storing the impulse response fragment data in which the section is extracted, and the amplitude delay data and the impulse response fragment data. And the impulse response combining means for generating a composite impulse response, the speech processing apparatus; and a convolution means for convolving the been the synthetic impulse response synthesized in the impulse response combining means to the voice data of the sound source. 2. The impulse response fragment data is extracted from the impulse response by a time window in which the amplitude of the impulse response fragment data is gradually attenuated in the latter half of the predetermined section. The voice processing device described. 3. The voice processing apparatus according to claim 1, wherein the space configuration data includes additional information indicating a reflection condition of voice in the virtual space. 4. The reflected sound response includes at least a first-order reflection or diffraction response to a reflector or an obstacle based on the spatial configuration of the virtual space in the virtual space. Voice processor. 5. The amplitude delay calculation means calculates a direct sound response of a sound wave propagating from the virtual sound source to the sound receiving point and a main reflected sound response by using a virtual image method. 1. The voice processing device according to 1. 6. The voice processing apparatus according to claim 1, further comprising operation means for changing a positional relationship between the sound source and the sound receiving point in a virtual space. 7. A voice processing device for generating a sound according to a positional relationship in a virtual space between a sound source in the virtual space and a sound receiving point for receiving a sound from the virtual sound source, From the impulse response of the sound wave propagating from the virtual sound source measured in the corresponding real space to the sound receiving point, impulse response fragment data obtained by extracting a predetermined section is superimposed on the sound data of the sound source in advance. A voice data storage means in which is stored, a space configuration data storage means in which space configuration data relating to the elements forming the virtual space is stored, and a direct sound response of a sound wave propagating from the virtual sound source to the sound receiving point. And amplitude delay calculation means for calculating amplitude delay data consisting of at least the amplitude and delay of the main reflected sound response based on the spatial configuration data, and the superimposed voice. Speech processing apparatus characterized by a chromatography data and a synthesizing means for synthesizing on the basis of said amplitude delay data. 8. A sound processing method for generating a sound according to a positional relationship in a virtual space between a sound source in a virtual space and a sound receiving point for receiving a sound from the virtual sound source, the sound processing method comprising: Amplitude delay calculation for calculating the amplitude delay data including at least the amplitude and delay of the direct sound response of the sound wave propagating to the sound receiving point and the main reflected sound response on the basis of the data for space configuration related to the elements configuring the virtual space. Step, from the impulse response of the sound wave propagating from the virtual sound source measured in the real space corresponding to the virtual space to the sound receiving point, based on the impulse response fragment data and the amplitude delay data extracted a predetermined section The impulse response synthesizing step of generating a synthetic impulse response, and the synthetic impulse response synthesized in the impulse response synthesizing step are combined with the sound of the sound source. And a convolution step of convolving the voice data. 9. The impulse response fragment data is extracted from the impulse response by a time window such that the amplitude of the impulse response fragment data is gradually attenuated in the latter half of the predetermined section. Described voice processing method. 10. The voice processing method according to claim 8, wherein the space configuration data includes additional information indicating a reflection condition of voice in the virtual space. 11. The voice processing method according to claim 8, wherein the reflected sound response includes at least a primary reflection or diffraction response to a reflector or an obstacle based on a spatial configuration of the virtual space. . 12. The amplitude delay calculation step calculates a direct sound response of a sound wave propagating from the virtual sound source to the sound receiving point and a main reflected sound response by using a virtual image method. 8. The voice processing method described in 8. 13. The voice processing method according to claim 8, further comprising an operation step of changing a positional relationship between the sound source and the sound receiving point in a virtual space. 14. A sound processing method for generating a sound according to a positional relationship in a virtual space between a sound source in a virtual space and a sound receiving point for receiving a sound from the virtual sound source, comprising: Amplitude delay calculation for calculating the amplitude delay data including at least the amplitude and delay of the direct sound response of the sound wave propagating to the sound receiving point and the main reflected sound response on the basis of the data for space configuration related to the elements configuring the virtual space. Steps, from the impulse response of the sound wave propagating from the virtual sound source measured in the real space corresponding to the virtual space to the sound receiving point, impulse response fragment data obtained by extracting a predetermined section from the sound data of the sound source. And a synthesizing step of synthesizing preliminarily superimposed voice data based on the amplitude delay data. 15. A control of a computer controllable voice processing device for generating a sound according to a positional relationship in a virtual space between a sound source in the virtual space and a sound receiving point for receiving a sound from the virtual sound source. In the program, the amplitude delay data consisting of at least the amplitude and delay of the direct sound response of the sound wave propagating from the virtual sound source to the sound receiving point and the main reflected sound response is used as the data for space configuration related to the elements configuring the virtual space. Amplitude delay calculation processing based on the impulse response fragment data obtained by extracting a predetermined section from the impulse response of the sound wave propagating from the virtual sound source measured in the real space corresponding to the virtual space to the sound receiving point, and Impulse fragment synthesizing process for generating a synthetic impulse response based on the amplitude delay data, and And a convolution process for convolving the synthesized impulse response with the voice data of the sound source, the control program being executed. 16. A control of a computer controllable voice processing device for generating a sound according to a positional relationship in a virtual space between a sound source in the virtual space and a sound receiving point for receiving a sound from the virtual sound source. In the program, the amplitude delay data consisting of at least the amplitude and delay of the direct sound response of the sound wave propagating from the virtual sound source to the sound receiving point and the main reflected sound response is used as the data for space configuration related to the elements configuring the virtual space. Amplitude delay calculation processing based on the above, impulse response fragment data obtained by extracting a predetermined section from the impulse response of the sound wave propagating from the virtual sound source measured in the real space corresponding to the virtual space to the sound receiving point Performs, on the audio processing device, a synthesis process of synthesizing the overlaid voice data that has been overlaid on the voice data of the sound source based on the amplitude delay data. A control program for causing.