WO2022145154A1

WO2022145154A1 - Stereophonic processing device, stereophonic processing method, and stereophonic processing program

Info

Publication number: WO2022145154A1
Application number: PCT/JP2021/043354
Authority: WO
Inventors: 宏岩村; 興一郎落合
Original assignee: オーディージョンサウンドラボ合同会社
Priority date: 2020-12-29
Filing date: 2021-11-26
Publication date: 2022-07-07
Also published as: JP7319687B2; JP2022104731A

Abstract

[Problem] To easily realize acoustic localization with a low amount of computation and storage capacity. [Solution] The present invention comprises: a stereo processing unit (1) that performs an out-of-head localization process on at least a first frequency band (91) of an input signal; and a reflection processing unit (2) that is connected to the stereo processing unit (1) and adds a reflected sound component to at least a portion of a second frequency band (92) of the input signal. The stereo processing unit (1) includes stereo processing filters (11a-11d) in which the coefficient of a transmission function for only a left-right pair is applied in advance or selectively, the transmission function of only the left-right pair being a high-frequency suppression transmission function for the left-right pair only, in which the band corresponding to the second frequency band (92) of a sound source signal is suppressed or attenuated for a pair of head transmission functions from the sound source position at only one angle within a 45-90° horizontal plane azimuth, to the left and right ears.

Description

3D sound processing device, 3D sound processing method and 3D sound processing program

The present invention relates to a stereophonic processing device that localizes the reproduced sound for earphones out of the head, a processing method thereof, and a program thereof.

When playing sound through speakers, the sound emitted from one of multiple speakers is divided into direct sound that reaches the ear directly and reflected sound that is reflected by the wall or floor and reaches the ear. The direct sound reaches the eardrum linearly, and also reaches the eardrum under the influence of wraparound / reflex in the head, reflex / resonance in the pinna and the ear canal. A head-related transfer function (HRTF) that expresses the change in the physical characteristics of the incident sound due to the periphery of the head in the frequency domain is added to the direct sound.

Normally, the head related transfer function represented in the graph by the relative amplitude [dB] (vertical axis) with respect to the frequency [kHz] (horizontal axis) is the relative amplitude [dB] (vertical axis) with respect to the time region [ms] (horizontal axis). ) Is the relationship between the Head-Related Impulse Response shown in the graph and the Fourier transform pair. The relative amplitude is expressed by multiplying the logarithm of the amplitude ratio of the received signal to the sound source signal by 20. Head-related transfer functions differ between the left and right ears. In order to obtain the left and right head related transfer functions HRTF _L , HRTF _R , first, in the free sound field of the unsounded room, from the sound source to the dummy head (pseudo head) or the microphone attached to both ears of the human head. Left and right head-related transfer responses g _L , g _R are divided by the blank head-related transfer response g _C from the same sound source to the microphone at the position corresponding to the head center without the head, and the left and right head-related transfer responses (g _L / g _C) . = HRIR _L , g _R / g _C = HRIR _R ) is acquired, and then the left and right head-related transfer functions are obtained by Fourier transform F for the left and right head-related transfer responses HRIR _L and HRIR _R , respectively (FHRIR). _L = HRTF _L , FHRIR _R = HRTF _R ).

On the other hand, when playing sound with earphones or earphones with a headband (headphones), the sound from the sound output part such as headphones reaches the eardrum almost without being affected by the shape around the head, unlike the sound from the speaker. However, the head related transfer function is not taken into account. For this reason, the brain of the headphone wearer cannot recognize the stereophonic effect of sound (3D sound, 3D sound image, 3D sound, sound image localization, out-of-head localization), and obtains only a narrow sound image limited to the inside of the head between the left and right ears. Not possible (intra-head localization). A narrow sound image not only does not produce a three-dimensional sound, but also causes discomfort to the headphone wearer and causes brain fatigue due to long-term listening.

As a method of obtaining a three-dimensional sound by playing with earphones or headphones, a sound source including a head-related transfer function is recorded with a microphone attached to both human ears or a dummy head, and the sound source signal is played back with headphones. A binaural reproduction method that takes into account the characteristics is known. However, the recognition of stereophonic sound (head-related transfer function) varies greatly among individuals, and in the binaural reproduction method, a person with a microphone or a person other than a dummy head whose head-related transfer function is close to correctly recognizes the stereophonic sound image. Can not. In fact, only a person with a microphone can obtain the effect of a stereophonic image by this reproduction method, and even a dummy head that attempts standardization and averaging cannot eliminate the harmful effects of individual differences in the head-related transfer function.

As a method of obtaining stereophonic sound by reproducing headphones or the like, a method of convolving a head-related transfer function measured in advance or measured at the time of reproduction into a sound source signal and correcting the sound source signal to form a stereophonic sound image is known. This is a method of personalizing, that is, measuring a head-related transfer function peculiar to the listener and correcting the sound source signal at the time of reproduction, contrary to the above-mentioned standardization and averaging attempt by a dummy head or the like. However, the measurement of the head-related transfer function needs to be performed with high accuracy by a special device in an anechoic chamber, which is not realistic for general listeners. There is also a simple measurement method for the head-related transfer function, but due to lack of accuracy, the head-related transfer function unique to each person cannot be obtained with high accuracy, and the three-dimensional effect cannot be sufficiently obtained.

The head-related transfer function differs not only depending on the individual, but also on the position of the sound source, that is, the angle and distance, and its frequency characteristics are extremely complicated. Therefore, when the head-related transfer function is convoluted into the sound source signal and corrected, the amount of calculation is extremely large, which leads to a decrease in calculation processing speed and an increase in size of the device. In order to realize stereophonic sound using the head-related transfer function, it is necessary to reproduce the reproduced sound that matches the characteristics peculiar to the headphone wearer with high accuracy by arithmetic processing. Since the sound image cannot be formed and the sound quality changes and the localization in the head occurs, the capacities of the arithmetic processing device and the storage device must be inevitably increased.

In addition, when correcting the signals of multiple localized sound sources (multi-track, multi-object), processing by the head-related transfer function for each localized sound source is required, and more calculations are required.

Only head-related transfer functions at typical and characteristic peak (mountain) frequencies and dip (valley) or notch (especially steep negative peak) frequencies for out-of-head localization to achieve sound image localization with less computation. Attempts have been made to make the headphone playback sound three-dimensional by a simplified method used for calculation. However, a listener having a peak or the like that is not the same as the typical peak or the like cannot obtain a sufficient stereophonic image, and such a simplified method cannot sufficiently eliminate the harmful effects of individual differences.

Patent Document 1 estimates the center frequency and bandwidth of the first peak from the measurement dimensions of a specific part of the listener's ear, further estimates the center frequency and bandwidth of the first notch, and obtains each transfer function. Disclosed is a three-dimensional sound reproduction device that convolves into a sound source signal and personally adapts to generate a binoral signal. However, it is difficult to accurately measure each dimension of the auricle portion, and the device of Patent Document 1 cannot generate a personalized stereophonic signal even if the amount of calculation can be reduced by the low-precision measurement.

In addition, when the audio signal is compressed, the head-related transfer function information in the high frequency band is lost or not stored by the compression process in the audio signal that convolves the general head-related transfer function, and the stereophonic effect is reduced. do. That is, the delay of the reflected sound in the high frequency band due to the head-related transfer function is as short as several hundred μs (0.0001 seconds) from the direct sound, and it is reduced by compression processing as a sound that is difficult to recognize from the auditory psychological point of view. Or it will be deleted. In order to prevent this, the high band component of the audio signal is amplified and corrected before compression, but the sound quality changes.

Patent Document 2 includes a reflected sound synthesis processing unit that synthesizes a reflected sound component with an input audio signal, and an out-of-head localization processing unit that filters the output of the reflected sound component and outputs the left and right audio signals. A coefficient based on the measurement of the head-related transfer function including the floor-reflected sound in the sound source direction is stored in each circuit of the reflected sound synthesis processing unit and the out-of-head localization processing unit, and discloses an apparatus for performing virtual sound image localization processing. However, in Patent Document 2, since the head-related transfer function that has not been suppressed or attenuated is convoluted with respect to the entire band of the audio signal, information on the head-related transfer function is used in the high frequency band when the audio signal is compressed. Disappears from the audio signal, and the stereophonic effect is significantly reduced.

Furthermore, in the case of a sound source from a position of 45 ° to 90 ° with respect to the listener, the direct sound reaches the ear predominantly, and the ear on the opposite side is shielded by the head and the direct sound hardly reaches. Generally, the head related transfer function is measured in a free sound field (anechoic chamber) for accurate measurement. However, since there is no reflection in the free sound field, the head-related transfer function on the opposite side of the sound source cannot be measured with sufficient accuracy and S / N ratio, and if the head-related transfer function is used as it is, it will be reproduced in a real environment with reflection. Accuracy deteriorates and human sound source space recognition does not work properly.

Japanese Unexamined Patent Publication No. 2017-85362 Japanese Unexamined Patent Publication No. 2009-105565

Therefore, an object of the present invention is to provide a stereophonic processing device for earphones or headphones that generates high-quality stereophonic sound with a small amount of calculation and storage capacity, and a processing method thereof. Another object of the present invention is to provide a stereophonic sound processing device and a processing method thereof for generating a stereophonic sound with little recognition difference between listeners and almost no change in sound quality with respect to a sound source. Another object of the present invention is to provide a stereophonic processing device for forming a stereophonic sound image by high-speed processing using a transfer function and a processing method thereof. Further, it is an object of the present invention to provide a stereophonic processing apparatus and a processing method thereof in which the stereophonic effect does not deteriorate with respect to the compression processing of the audio signal.

The stereophonic processing apparatus (10,10', 10 ") of the present invention has a first frequency band (91) including the lowest region of the audible region and a second frequency band (91) higher than the first frequency band (91). A stereophonic sound reproduction signal is generated from the sound source signal including 92) and supplied to the earphone (50) or the headphone. The stereophonic processing unit (1) that performs out-of-head localization processing at least the first frequency band (91) of the input signal. ) And a reflection processing unit (2) connected to the stereophonic processing unit (1) and adding a reflected sound component to at least a part of the second frequency band (92) of the input signal. Includes a stereophonic processing filter (11a-11d) to which the coefficients of only one pair of left and right transmission functions are given in advance or by selection, and the only pair of left and right transmission functions are at only one angle from 45 ° to 90 ° in the horizontal plane. A pair of left and right high-frequency suppression transmission functions in which the band (92') corresponding to the second frequency band of the sound source signal is suppressed or attenuated for the pair of head transmission functions from the sound source position to each of the left and right ears. be.

In the present invention, the transfer function of only one angle is used, and the high frequency suppression transfer function (BRTF) in which the band (92') corresponding to the second frequency band is suppressed or attenuated is used, so that the amount of calculation is small. And, depending on the memory capacity, the sound source signal can be processed in three dimensions. In addition, since the high-frequency suppression transfer function (BRTF) is obtained from the head-related transfer function (HRTR) with a horizontal plane azimuth (θ) of 45 ° to 90 °, individual differences in stereophonic sound recognition are small, and earphones (50) or For all listeners wearing headphones, there is no in-head localization and high-quality three-dimensional sound can be reproduced. Further, in the present invention, the coefficient (h _B0 to h _Bn ) of the high frequency suppression transmission function (BRTF) is given to the three-dimensional processing unit (1) in advance and is not updated during operation, so that a storage device and an additional arithmetic unit are used. The entire device can be made smaller and lighter without the need, and can be built into earphones (50) or headphones. Alternatively, in the present invention, since the coefficient (h _B0 to h _Bn ) of the high frequency suppression transmission function (BRTF) at one horizontal plane azimuth (θ) can be selected, the type of sound depends on the type of earphone (50) or headphone. Therefore, it is possible to realize optimum reproduction of three-dimensional sound by switching according to the taste of the listener.

Another embodiment of the stereophonic processing apparatus (10''') of the present invention includes a first and second input terminals (6a, 6b) for inputting a sound source signal to the stereophonic processing apparatus (10'''). It is provided with a stereophonic processing unit (1) that performs out-of-head localization processing at least the first frequency band (91) of the input signals from the first and second input terminals (6a, 6b). The three-dimensional processing unit (1) includes a first and second three-dimensional processing filter (11a, 11b) connected to the first input end (6a) and a third and fourth three-dimensional processing filters (11a, 11b) connected to the second input end (6a). Both output signals of the three-dimensional processing filter (11c, 11d) and the second and third three-dimensional processing filters (11b, 11c) are input as left and right signals, separated into a center signal and a side signal, and a reflected sound component is added to the side signal. And the other adder (16b) that adds the output signal of one of the M / S processor (51) and the output signal of the fourth stereoscopic processing filter (11d). , The one adder (16a) for adding the other output signal of the M / S processor (51) and the output signal of the first three-dimensional processing filter (11a) is provided. In the first to fourth three-dimensional processing filters (11a-11d), the transfer function of only one pair of left and right is given in advance or by selection, and the transfer function of only one pair of left and right is one angle of horizontal plane orientation angle of 45 ° to 90 °. Only a pair of left and right high-frequency transfer functions in which the band (92') corresponding to the second frequency band of the sound source signal is suppressed or attenuated for the pair of head-related transfer functions from the sound source position to the left and right ears. It is a function. By separating the center and side signals with the M / S processor (51) and adding reflected sound to the side signal, for example, the side signal on the opposite side of the sound source, it is possible to generate a sound that is easy for humans to recognize three-dimensional spatial sound. ..

The stereophonic processing method of the present invention is from a sound source signal including a first frequency band (91) including the lowest region of the audible region and a second frequency band (92) higher than the first frequency band (91). , Generates a stereophonic sound reproduction signal and supplies it to the earphone (50) or headphones. The process of preliminarily assigning the transfer function coefficient to the three-dimensional processing filter (11a-11d) of the three-dimensional processing unit (1) and the sound source signal to the three-dimensional processing filter (11a-11d) to which the transfer function coefficient is preliminarily assigned. Alternatively, the process of inputting the processing signal of the reflection processing unit (2) and performing out-of-head localization processing of at least the first frequency band (91) of the input signal to generate the processing signal of the three-dimensional processing unit (1) is included. The process of assigning the transfer function coefficient in advance is the process of preparing a pair of left and right head transfer functions from the sound source position to each of the left and right ears at only one angle of the horizontal plane orientation angle of 45 ° to 90 °, and the process of preparing a pair of left and right heads. The process of suppressing or attenuating the band (92') corresponding to the second frequency band with respect to the transfer function to generate a pair of left and right high-frequency suppression transfer functions, and the process of inversely fooling the pair of left and right high-frequency suppression transfer functions. A three-dimensional processing filter (11a- 11 d) includes the process of pre-giving.

Further, in another embodiment of the three-dimensional acoustic processing method of the present invention, a sound source signal is input to the process of generating and accumulating the coefficient of the transfer function and the three-dimensional processing filter (11a-11d) of the three-dimensional processing unit (1). It includes a process of inputting a processing signal of the reflection processing unit (2), performing out-of-head localization processing of at least the first frequency band (91) of the input signal, and generating a processing signal of the three-dimensional processing unit (1). The process of generating and accumulating transfer function coefficients is the process of preparing multiple pairs of left and right head transfer functions from each sound source position to each of the left and right ears at multiple angles of horizontal plane orientation angle of 45 ° to 90 °, and multiple pairs of left and right. With respect to the head transfer function of, the band (92') corresponding to the second frequency band is suppressed or attenuated by the coefficient generator (3) to generate a plurality of pairs of left and right high-frequency transfer functions. , The process of obtaining the high-frequency suppression impulse response of multiple pairs of left and right by arithmetically processing the high-frequency suppression transfer function of multiple pairs of left and right by inverse Fourier transform, and the high of multiple pairs of left and right corresponding to the value of the high-frequency suppression impulse response. It includes the process of accumulating the coefficient of the region suppression transfer function in the coefficient storage unit (4). The process of generating the processing signal of the three-dimensional processing unit (1) is the coefficient of the high-frequency suppression transfer function of only one pair of left and right among the coefficients of the high-frequency suppression transfer function of multiple pairs of left and right stored in the coefficient storage unit (4). For the process of selecting and the three-dimensional processing filter (11a-11d) to which the transfer function coefficient is pre-assigned or not assigned, only a pair of left and right high-frequency suppression transmissions at only one selected horizontal plane azimuth angle. The sound source signal or the processing signal of the reflection processing unit (2) is passed through the process of assigning the function coefficient and the stereoscopic processing filter (11a-11d) to which only one pair of left and right high-frequency suppression transfer function coefficients are assigned. This includes a process of generating a processing signal of the three-dimensional processing unit (1) that has been subjected to out-of-head localization processing for at least the first frequency band (91).

The stereophonic processing program of the present invention causes a computer to function as each part of the stereophonic processing apparatus.

The stereophonic processing device, processing method, and program of the present invention can realize stereophonic sound with high sound quality and natural sound quality with little influence of individual differences. Further, with a small amount of calculation and storage capacity, the stereophonic processing device of the present invention, which can be mounted as software on a personal computer, a smartphone, a game machine or the like, and which is downsized and lightweight, can be built in an earphone or a headphone. Further, in the present invention, even a wireless earphone or headphone that requires transmission of a compressed signal can realize a high-quality stereophonic effect without correction such as amplification.

A block diagram showing a stereophonic processing apparatus according to the first embodiment of the present invention. A block diagram illustrating the configuration of the three-dimensional processing unit of FIG. A block diagram illustrating the configuration of the three-dimensional processing unit of FIG. A block diagram illustrating the configuration of the three-dimensional processing filter of FIGS. 2A and 2B. A block diagram illustrating the configuration of the reflection processing unit of FIG. A block diagram illustrating the configuration of the reflection processing unit of FIG. A block diagram illustrating the configuration of the reflection processing unit of FIG. A block diagram illustrating the configuration of the reflection processing unit of FIG. A block diagram illustrating the configuration of the reflection processing filter of FIGS. 4A to 4D. A block diagram illustrating the configuration of the frequency filter of FIG. Schematic diagram showing how to obtain a head impulse response from a sound source Flowchart to generate high frequency suppression transfer function from head impulse response Graph showing each transfer function of left ear (a) and right ear (b) Graph showing each impulse response of the left ear (a) and the right ear (b) A graph showing each audio signal processed by the prior art (a) and the method (b) of the present invention. A graph showing each audio signal obtained by compressing each audio signal in FIG. 11. A block diagram showing a stereophonic processing apparatus according to a second embodiment of the present invention. A block diagram showing a stereophonic processing apparatus according to a third embodiment of the present invention. A block diagram illustrating the configuration of the M / S delay unit of FIG. A block diagram showing a stereophonic processing apparatus according to a fourth embodiment of the present invention.

The stereophonic processing apparatus according to the present invention, the processing method thereof, and the embodiment of the program will be described with reference to FIGS. 1 to 16. The following embodiments are illustrative and do not limit the interpretation of the present invention.

FIG. 1 shows a block diagram of a two-channel stereophonic processing device 10 that generates a stereophonic sound reproduction signal from a sound source signal stored in the sound source apparatus 30 and supplies the stereophonic sound reproduction signal to the earphone 50 or headphones. The sound source signal includes a first frequency band 91 including the lowest region of the audible region and a second frequency band 92 higher than the first frequency band 91. The human audible range is generally said to have a frequency range of 20 Hz to 20,000 Hz, the lowest of which is about 20 Hz. The stereophonic processing device 10 of the present invention is connected to a stereophonic processing unit 1 that performs out-of-head localization processing of at least the first frequency band 91 of the input signal and a stereophonic processing unit 1, and has a height exceeding the first frequency band 91 of the input signal. A reflection processing unit 2 for adding a reflected sound component is provided in at least a part of the second frequency band 92 including the region.

The three-dimensional processing unit 1 shown in FIG. 1 is connected to a sound source device 30 in which a sound source of digital or analog data is stored via the first and

second input terminals

6a and 6b. When the stored sound source is analog data, an analog-to-digital conversion device (not shown) is required between the sound source device 30 and the stereophonic processing device 10. The three-dimensional processing unit 1 includes a three-dimensional processing filter 11a-11d to which only a pair of left and right transfer function coefficients are given in advance, and the three-dimensional processing filter (11a-11d) passes an input signal of a sound source signal to a digital signal. Out-of-head localization processing is performed on the sound source signal by processing.

The block diagram of FIG. 2A illustrating the embodiment of the three-dimensional processing unit 1 includes the first three-dimensional processing filter 11a and the second three-dimensional processing filter 11b, each of which is connected to the first input end 6a of the stereophonic processing apparatus 10. , Connected to the third stereophonic processing filter 11c and the fourth stereophonic processing filter 11b, one of which is connected to the second input end 6b of the stereophonic processing apparatus 10, and the other of the first and third

stereophonic processing filters

11a, 11c. The adder 16a is provided, and the adder 16b connected to the other of the second and fourth steric processing filters 11b and 11d is provided. It is a so-called tasuki hanging structure. The output sides of the

adders

16a and 16b are connected to the reflection processing unit 2 as shown in FIG. As shown in the block diagram of FIG. 2B, the M / S processor 51 may be arranged between the second and third three-dimensional processing filters 11b and 11c and the two

adders

16a and 16b. In the M / S processor 51, left and right signals are input to the L terminal and the R terminal from the

branch points

55a and 55b between the second and third three-dimensional processing filters 11b and 11c and the

adder

16a and 16b. The M / S converter 52, which is separated into a center signal and a side signal and outputs the center and side signals from the M terminal and the S terminal, respectively, and the center signal and the side signal from the M / S converter 52 are input and reflected, respectively. From the

reflection processing filters

54a and 54b that add sound and the

reflection processing filters

54a and 54b, the signal after reflection processing is input to the M terminal and S terminal, respectively, and returned to the left and right signals, and the left and right feedback signals are sent from the L terminal and R terminal. The L / R converter 53 that outputs It is equipped with two

adders

57a and 57b that output to the

adders

16b and 16a by tapping.

The first to fourth three-dimensional processing filters 11a-11d are FIR filters (finite impulse response filters) or IIR filters (infinite impulse response filters). The block diagram of FIG. 3 illustrates the configuration of the three-dimensional processing filter 11a by a general FIR filter. That is, the output signals of both the delay device 12 ₁ and the multiplier 13 _{0 connected to the input terminal 6a, the multiplier 13 1 connected to the delay device 12 1 and the multiplier 13 0} _and _the _multiplier 13 ₁ are added. It is equipped with an adder 141, a delay device 12 ₁ , a multiplier 13 ₁ and an adder 14 ₁ to form _a tap, and a plurality of taps to form a higher-order (n-th order) FIR filter.

The coefficient of the transfer function of only the pair of left and right given to the first to fourth three-dimensional processing filters 11a-11d in advance is shown in FIG. 7 at any one angle of the horizontal plane azimuth (θ) 45 ° to 90 °. It is a coefficient obtained by arithmetically processing a pair of head impulse responses HRIR _L and HRIR _R from the position of the sound source 71 to the left and

right ears

70 L and 70 R. Since only one angle is used, the amount of calculation and time can be reduced in the present invention. Here, the horizontal azimuth (θ) 45 ° to 90 ° is an angle of both left and right directions + 45 ° to + 90 ° and −45 ° to −90 when the front direction of the head 70 of the human or dummy head is 0 °. Includes any °. In the angle range, the influence of individual differences is small and a stereophonic sound image can be formed. Any angle of horizontal azimuth (θ) includes, for example, plus or minus 45 °, 50 °, 55 °, 60 °, 65 °, 70 °, 75 °, 80 °, 85 ° or 90 °. It is not limited to these as long as it is within the range of 45 ° to 90 °. The vertical angle (elevation angle) in which the individual difference in the head impulse response is large is set to 0 ° and is not considered in the present invention.

The pre-assigned transfer function coefficient is a pair of left and right head-related transfer obtained by arithmetically processing a pair of left and right head-related transfer responses HRIR _L and HRIR _R at one of the horizontal plane orientation angles of 45 ° to 90 °. It is the coefficient of the high frequency suppression transfer functions BRTF _L , BRTF _R of only one pair on the left and right, in which the band 92'corresponding to the second frequency band 92 of the sound source signal is suppressed or attenuated with respect to the functions HRTF _L and HRTF _R. .. The band 91'of the first frequency band 91 and the corresponding transfer functions HRTF and BRTF is a band included in the range of 0 to 8 kHz, and is preferably 0.1 kHz to 8 kHz. The second frequency band 92 and the corresponding transfer functions HRTF and BRTF band 92'are included in the range of 8 kHz to 96 kHz, and are usually 8 kHz to 20 kHz in the audible range, preferably 8 kHz to 15 kHz. .. The upper limit is half the sampling frequency Fs (Fs / 2) for the convenience of processing, for example, the sound source is 44.1kHz / 2 for CD, 48kHz / 2, 96kHz / 2, 192kHz / 2 for DVD / Blu-ray, etc. You may.

In the stereophonic processing device 10 shown in FIG. 1, one is connected to the stereophonic processing unit 1 and the other is connected to the digital-to-analog (D / A) conversion device 40 via the

output terminals

7a and 7b. To prepare for. The D / A conversion device 40 is wired or wirelessly connected to the headphones 50 by a transmission line 41, and transmits a stereophonic sound reproduction signal to the earphones 50. Further, the digital signal is wirelessly transmitted as it is from the transmitter (indicated by reference numeral 40 in this case) connected to the

output terminals

7a and 7b without providing the D / A conversion device on the stereophonic processing device 10 side, and the bluetooth is transmitted. (Registered trademark) It may be converted into an analog signal by a D / A conversion device (not shown) added or built into an earphone or the like.

FIG. 4A is a block diagram illustrating the configuration of the two-channel reflection processing unit 2. The reflection processing unit 2 includes a first reflection processing filter 21a connected to a first branch point 25a between the first input end 6a (FIG. 1) of the three-dimensional processing device 10 and the earphone 50 to input a signal. .. The processing signal of the three-dimensional processing unit 1 and the feedback signal from the first reflection processing filter 21a are added to the first addition point 26a between the first input terminal 6a and the first branch point 25a of the three-dimensional processing apparatus 10. A first adder 27a is provided. Similarly, the reflection processing unit 2 is connected to the second branch point 25b between the second input end 6b (FIG. 1) of the three-dimensional processing device 10 and the earphone 50, and the signal is input to the second reflection processing filter. Equipped with 21b. The processing signal of the three-dimensional processing unit 1 and the feedback signal from the second reflection processing filter 21b are added to the second addition point 26b between the second input end 6b and the second branch point 25b of the three-dimensional processing apparatus 10. A second adder 27b is provided. Further, as shown in FIGS. 4B and 4D, the feedback signal from the first reflection processing filter 21a is input to either or both of the second adder 27b and the first adder 27a by the

switches

17a and 17b. Alternatively, the feedback signal from the second reflection processing filter 21b may be input to either or both of the first adder 27a and the second adder 27b. In FIG. 4B, an M / S converter 28 is provided between the first and

second branch points

25a and 25b and the first and second

reflection processing filters

21a and 21b, and the stereo left and right signals L and R are centered and side signals. It can be separated into M and S and processed individually. Perform a Fourier transform and perform a Fourier transform, | L amplitude |-| R amplitude |, (L amplitude) ^2- (R amplitude) ² , or arctan (| L amplitude | / | R amplitude |) within a certain range (preferably arctan (0.7). ) ≤ arctan (| L amplitude | / | R amplitude |) ≤ arctan (1.3)) may be separated as a center signal component, and the others may be separated as side signal components and returned by inverse Fourier transform (||). Is an absolute value). In general, since the low frequency component is often arranged in the center of the music signal, the coefficient of the low frequency portion (preferably about 200 Hz) may be used as the center signal component regardless of the arctan value. Only the center signal component may be extracted and subtracted from the original signal to be used as the side signal component. The conversion formulas are M = (L + R) / 2 and S = (LR) / 2. In FIG. 4B, an L / R converter 29 is provided between the first and second

reflection processing filters

21a and 21b and the first and

second adders

27a and 27b, and the center and side signals M and S are stereo left and right signals. Return to L and R. The conversion formulas are L = M + S and R = MS. 4C shows the first and third

reflection processing filters

21a and 21c, one of which is connected to the M terminal of the M / S converter 28, and the first and one of which are connected to the S terminal of the M / S converter 28. Adder 19a connected to the M terminal of the L / R converter 29 by adding the outputs of the second and fourth

reflection processing filters

21b and 21d and the first and third

reflection processing filters

21a and 21c, and the second and second. It is provided with an adder 19b connected to the S terminal of the L / R converter 29 by adding both outputs of the fourth

reflection processing filter

21b and 21d. Further, in FIG. 4D, switching

switches

18a and 18b are provided between the third and fourth

reflection processing filters

21c and 21d and the

adders

19a and 19b, and two output signals from the third reflection processing filter 21c are added. It can be switched and transmitted to either or both of the

devices

19a and 19b, and the output signal from the fourth reflection processing filter 21d can be switched and transmitted to either or both of the two

adders

19b and 19a.

FIG. 5 illustrates the configuration of the first reflection processing filter 21a of the reflection processing unit 2, from the delayer 22 connected to the first branch point 25a (FIGS. 4A to 4D) and inputting a signal, and the delayer 22. Includes a multiplier 23 that amplifies the output signal of .. FIG. 6 is a block diagram showing an IIR filter, which is an embodiment of the frequency filter 24 of FIG. The frequency filter 24 is a multiplier that multiplies the input signal X ₂₄ by a coefficient b ₀ , a delayer 62 ₁ that delays the input signal X ₂₄ , and a multiplier that multiplies the signal by the delayer 62 ₁ by _{a coefficient b 1} _. 63 ₁ and the delayer 62 ₂ that delays the signal by the delayer 62 ₁ , the multiplier 63 ₂ that multiplies the signal by the delayer 62 ₂ by the coefficient b ₂ , and the signal by the multiplier 63 ₀ and the multiplier 63 ₁ It is provided with a feed forward unit including an adder 64 ₁ for adding a signal and an adder 64 ₂ for adding a signal by the adder 64 ₁ and a signal by a multiplier 63 ₂ .

Further, the frequency filter 24 uses a delay device _{6 2 3} for inputting a part of the output signal Y ₂₄ generated from the signal by the adder 64 ₂ through the

adders

64 ₃ and 64 ₄ and a coefficient a ₁ for the signal by the delay device ₆ 23. A multiplier 63 ₃ that multiplies by, a delayer 62 ₄ that delays the signal by the delayer 62 ₃ , a multiplier 63 ₄ that multiplies the signal by the delayer 62 ₄ by the coefficient a ₂ , and a signal by the adder 64 ₂ . It is provided with a feedback unit including an adder 64 ₃ that adds the signal by the multiplier 63 ₃ and an adder 64 ₄ that adds the signal by the adder 64 ₃ and the signal by the multiplier 63 ₄ . Since the second reflection processing filter 21b of FIGS. 4A to 4D has the same configuration as the first reflection processing filter 21a, the description thereof will be omitted. The reflected sound includes an initial reflected sound and an additional reflected sound. Early reflection is a reflected sound that arrives in a short time after a direct sound that arrives directly from a sound source, and is referred to as a reflected sound that is reflected only once by an object. The additional reflected sound is a reflected sound having two or more reflections, and is also referred to as a rear reverberation sound, simply a reverberation sound, or a reverb.

A method for three-dimensionally processing a sound source signal using the stereophonic device 10 of the present invention shown in FIG. 1 will be described in detail below.
In the present invention, first, the coefficient of the transfer function is given in advance to the three-dimensional processing filter 11a-11d of the three-dimensional processing unit 1. In the process of assigning the transfer function coefficient in advance, a pair of left and right head-related transfer functions HRTF _L , HRTF from the sound source position to each of the left and right ears at only one angle of the horizontal plane azimuth angle of 45 ° to 90 ° in an anechoic chamber. Includes the process of preparing _R. For example, in FIG. 7, a dummy head or a microphone (not shown) provided on the left and

right ears

70L and 70R of the human head 70, which is about 1 m away from the sound source 71 at an impulse (θ) real + 45 ° as one angle, is used. The left and right impulse responses g _L and g _R are recorded, and the blank impulse response g _C is recorded by a microphone (not shown) placed at the position corresponding to the center of the head 70 C without the head 70, and the left and right impulse responses g are recorded. By dividing _L and g _R by the blank impulse response g _C , the left and right head impulse responses (g _L / g _C = HRIR _L , g _R / g _C = HRIR _R ) are obtained.

FIG. 8 shows a flowchart (90) for generating an impulse response (high frequency suppression impulse response BRIR _L , BRIR _R ) of a high frequency suppression transfer function from the left and right head impulse responses HRIR _L , HRIR _R (80). First, the left and right head impulse responses HRIR _L and HRIR _R are arithmetically processed by Fourier transform F (s81), and the common logarithm is multiplied by 20 and converted into a dB value (s82) to transmit a pair of left and right heads. The function is prepared (FHRIR _L = HRTF _L , FHRIR _R = HRTF _R ). Alternatively, a pair of left and right head-related transfer functions HRTF _L , HRTF _R obtained from or corrected from a database of a research institution, a university, a general company, etc. may be used. In FIGS. 9 (a) and 9 (b), the horizontal axis is the frequency [kHz] and the vertical axis is the relative amplitude [dB]. The head-related transfer function HRTF _L of FIG. 9B is shown, and the solid line in FIG. 9B shows the head-related transfer function HRTF _R of the right ear 70R at a horizontal azimuth (θ) of 45 °. In FIG. 9, the sampling frequency Fs is 44.1 kHz.

For the obtained pair of left and right head-related transfer functions HRTF _L and HRTF _R , a band 92'corresponding to the second frequency band 92 of the sound source signal, for example, a component having a frequency exceeding 8 kHz is passed through a low-pass filter. Suppression or attenuation processing (s83) produces a pair of left and right high-frequency suppression transfer functions BRTF _L , BRTF _R , which are shown by the broken lines in FIGS. 9 (a) and 9 (b). By the suppression or attenuation processing (s83) by the low-pass filter, fine irregularities disappear on the time axis and the waveform is smoothed. For example, it may be smoothed by a moving average method. In FIGS. 9A and 9B, the head-related transfer functions HRTF _L and HRTF _R (solid line) before the high-frequency suppression treatment are compared with the high-frequency suppression transfer functions BRTF _L and BRTF _R (broken line) after the treatment. Then, in the band 91'corresponding to the first frequency band from 0 to 8 kHz, the solid line and the broken line show almost the same value. On the other hand, in the band 92'corresponding to the second frequency band exceeding 8 kHz, the solid line and the broken line show similar changes and behaviors, but the high frequency suppression transfer functions BRTF _L and BRTF _R are head-related transfer functions HRTF _L. , Shows a value attenuated by a few dB from the HRTF _R.

Next, for the pair of left and right high-frequency suppression transfer functions BRTF _L , BRTF _R (broken line in Fig. 9) at one angle of the horizontal plane azimuth (θ), the logarithmic value is returned to the amplitude value (s84), and the inverse Fourier transform F ^-1 . Then (F ^-1 BRTF _L , F ^-1 BRTF _R ) (s85), and further, window function processing is performed by, for example, a humming window (s86), and a pair of left and right high-frequency suppression impulse responses BRIR _L , BRIR _R are obtained (F -1 BRTF L, F -1 BRTF R). 90). A square wave window, a Hanning window, or a Bartlett window may be used as the window function. FIGS. 10 (a) and 10 (b), in which the horizontal axis represents the number of samples (maximum value 512) and the vertical axis represents the amplitude, show a pair of left and right high-frequency suppression impulse responses BRIR _L , BRIR _R (solid lines) obtained by the present invention. ) Is a graph showing in comparison with a pair of left and right head impulse responses HRIR _L , HRIR _R (broken line) of the prior art. The impulse responses HRIR _L and BRIR _L on the left ear 70L side proximal to the sound source 71 are shown in FIG. 10 (a), and the impulse responses BRIR _R and BRIR _R on the right ear 70R side distal to the sound source 71 are shown in FIG. Shown in (b). Due to the difference in distance from the sound source 71, FIG. 10 (a) has a larger amplitude as a whole as compared with FIG. 10 (b).

As shown in FIGS. 10 (a) and 10 (b), in the high frequency suppression impulse response BRIR _L , BRIR _R (broken line) obtained by the present invention, the amplitude converges to 0 when the number of samples is about 150, whereas the amplitude converges to 0 in the conventional case. In the head impulse response HRIR _L , HRIR _R (solid line) of the technology, the amplitude does not converge up to the maximum number of samples 512 (right end of FIG. 10). In the present invention, in order to suppress or attenuate the band 92'corresponding to the second frequency band, the time-delayed impulse response (with 150 or more samples) mainly composed of high frequency components is suppressed. caused by. The value of the amplitude of the impulse response corresponds to the filter coefficient value in a frequency filter such as an FIR filter, and the number of samples corresponds to the number of taps (n + 1) of the FIR filter. If the number of taps (n + 1) is small, the amount of calculation by the FIR filter can be reduced. Therefore, in the present invention in which the high-frequency suppression impulse responses BRIR _L and BRIR _R are adopted, the effect of significantly reducing the amount of calculation can be obtained as compared with the prior art.

Since the obtained values of the pair of left and right high-frequency suppression impulse responses BRIR _L and BRIR _R (broken line in FIG. 10) directly correspond to the coefficients of the FIR filter, only the pair of left and right high-frequency suppression at one angle of the horizontal plane azimuth θ The coefficients h _B0 to h _Bn of the transfer function are given in advance to the three-dimensional processing filters 11a-11d. Specifically, in the FIR filter 11a having a filter length (number of taps) n + 1 shown in FIG. 3, the high frequency suppression impulse response BRIR _L is set as the filter coefficients h _B0 to h _Bn for a plurality of multipliers 13 ₀ -13 _n . , BRIR _R values are given respectively. The transfer function H (z) of the FIR filter is expressed as Equation 1.
H (z) = h _B0 + h _B1 z ^-1 + ... + h _Bn z ^-n (Equation 1)

The frequency response of Equation 1 is expressed by Equation 2 by converting e- ^jωT to z ^-1 . T is the time interval [seconds] between data, ω is the angular frequency [rad / sec] represented by 2πFc / Fs, Fs is the sampling frequency [Hz], and Fc is the cutoff frequency [Hz]. For example, if Fc is set to 8 kHz, the band 91'corresponding to the first frequency band becomes a band of 8 kHz or less.
H (jω) ＝ h _B0 ＋ h _B1 e ^-jωT ＋・・＋ h _Bn e ^-jnωT (Equation 2)

By inputting the values of the high frequency suppression impulse responses BRIR _L and BRIR _R obtained according to the flowchart of FIG. 8 into h _B0 to h _Bn of Equation 2 in consideration of the frequency characteristics, the three-dimensional processing filter 11a of the three-dimensional processing unit 1 is input. Is given in advance the coefficients of only a pair of left and right transfer functions. When the sound source is passed through the three-dimensional processing filter 11a to which the coefficient is given in advance, the input signal Xa (FIG. 3) is converted into the output signal Ya by adding the coefficients of the high frequency suppression transfer functions BRTF _L and BRTF _R. (HXa = Ya). Similarly, the high-frequency suppression impulse responses BRIR _L and BRIR are given in advance as coefficients to the other three-dimensional processing filters 11b-11d. When the sound source signal from the sound source device 30 is input to the three-dimensional processing unit 1 to which the coefficient of the transfer function is assigned through the first and

second input terminals

6a and 6b, the input signal is subjected to out-of-head localization processing to perform three-dimensional processing. The processing signal of part 1 is obtained. In the present invention, the entire first and

second frequency bands

91 and 92 of the sound source signal are subjected to out-of-head localization processing, but in the high frequency suppression transfer function BRTF, the band 92'corresponding to the second frequency band is suppressed or attenuated. Therefore, the influence of the high frequency suppression transfer function BRTF on the second frequency band 92 is smaller than the influence on the first frequency band 91. In other words, signal processing using the coefficient of the high-frequency suppression transfer function BRTF for the first frequency band 91 is effective for forming a stereophonic image with little individual difference.

Subsequently, the processed signal subjected to the out-of-head localization processing by the three-dimensional processing unit 1 is input to the reflection processing unit 2, and is applied to at least a part of the second frequency band 92 including the high frequency band exceeding the first frequency band 91 of the input signal. , A reflected sound component is added to generate a processing signal of the reflection processing unit 2. Further, in the reflection processing unit 2, when the processing signal of the three-dimensional processing unit 1 input to the reflection processing unit 2 passes through the reflection processing unit 2, a second frequency band higher than the first frequency band 91 of the input signal is generated. For 92, an additional reflected sound component (reverberation sound, reverb) delayed from the direct sound by 0.001 to 10 seconds is added together with the initial reflected sound.

Specifically, the input signal introduced from the

branch points

25a and 25b of FIGS. 4A to 4D to the

reflection processing filters

21a and 21b of the reflection processing unit 2 has the delay device 22 shown in FIG. 5 and an arbitrary coefficient of the reflection processing gain. It passes through the applied multiplier 23 and is input to the frequency filter 24. In the biquadratic (BiQuad) filter of the IIR filter shown in FIG. 6 as the frequency filter 24, in the case of a low shelf filter (LSF: a filter that amplifies components below the cutoff frequency by the amount of amplification), the transfer function H (z). And its frequency response are represented by

Equations

3 and 4.
H (z) = A (z ^-2 + (A ^1/2 / Q) z ^-1 + A) / (Az ^-2 + (A ^1/2 / Q) z ^-1 +1) (Equation 3)
H (jω) = A (e ^-j2ω + (A ^1/2 / Q) e ^-jω + A) / (Ae ^-j2ω + (A ^1/2 / Q) e ^-jω +1) (Equation 4)

Q is the filter Q value (a value that affects the amplification curve around the cutoff frequency), G is the amplification amount [dB] that determines the filter characteristics, and ω is the angular frequency [rad / sec] expressed by 2πFc / Fs. Fs indicates the sampling frequency [Hz], and Fc indicates the cutoff frequency [Hz]. For example, if Fc is set to 8 kHz, the band 92'corresponding to the second frequency band is in the range of 8 kHz or more. A is represented by 10 ^{G / 40} and α is represented by sinω / 2Q. In the FIR filter, the impulse response is equal to the coefficient, but the coefficients a ₁ , a ₂ , b ₀ , b ₁ , b ₂ of the LSF in the IIR filter 24 in FIG. 6 are the following equations 5 to 5 in consideration of the frequency characteristics. Determined by 9.
a ₁ = -2 ((A-1) + (A + 1) cosω) (Equation 5)
a ₂ = (A + 1) + (A-1) cosω-2αA ^1/2 (Equation 6)
b ₀ = A ((A + 1)-(A-1) cosω + 2αA ^1/2 ) (Equation 7)
b ₁ = 2A ((A-1)-(A + 1) cosω) (Equation 8)
b ₂ = A ((A + 1)-(A-1) cosω-2αA ^1/2 ) (Equation 9)
In the frequency filter 24, the coefficients a ₁ , a ₂ , b ₀ , b ₁ , b ₂ can be determined by arbitrarily setting the filter Q value, the amplification amount G, and the cutoff frequency Fc according to the earphone characteristics and the listener's preference. .. Further, instead of the LSF, other filter types such as low-pass, high-pass, band-pass, notch, high-shelf, peaking, and all-pass filters may be used. It may be combined with other filters.

FIG. 11 (a) schematically shows a three-dimensional sound reproduction signal 34 in which a coefficient of a conventional head-related transfer function HRTF having a horizontal plane orientation angle of substantially 45 ° is added to the sound source signal, and FIG. 11 (a) shows the same According to the present invention, the stereoscopic sound reproduction signal 35 to which the coefficient of the high frequency suppression transmission function BRTF is added to the sound source signal by the stereoscopic processing unit 1 and the components of the initial reflected sound and the additional reflected sound are added by the reflection processing unit 2 is outlined. Show. For comparison, the sound source signal 33 (broken line) is shown in both figures. In FIGS. 11A and 11B, the horizontal axis is frequency [kHz] and the vertical axis is relative amplitude [dB], and the data on the left ear 70L side when the sound source 71 is substantially + 45 ° (FIG. 7) is illustrated. do. The compatible sound reproduction signals 34 and 35 of FIGS. 11A and 11B show almost the same characteristics in the first frequency band 91 of substantially 8 kHz or less, but in the second frequency band 92 of substantially more than 8 kHz. The characteristics are different.

The three-dimensional sound reproduction signal 35 generated by both the three-dimensional processing unit 1 and the reflection processing unit 2 is transmitted to the digital-to-analog conversion device 40 via the

output terminals

7a and 7b, and further sent to the earphone 50 or the headphone as the three-dimensional sound reproduction signal. Supply 35. In this case, for example, when transmitting the three-dimensional sound reproduction signal 35 to the wireless earphone, the signal compression processing is required. FIG. 12 (a) schematically shows an audio signal 34'that has been compressed during transmission to, for example, a wireless earphone, with respect to the audio signal 34 (FIG. 11 (a)) to which a coefficient of the conventional head-related transfer function HRTF is added. .. In FIG. 12 (a), the audio signal 35 (FIG. 11 (a)) processed by the three-dimensional processing unit 1 and the reflection processing unit 2 is compressed in the same manner as described above by the processing method of the present invention, and after the compression processing. The audio signal 35'is outlined. In the voice signal 34'by the conventional technique, the sound image information of the second frequency band 92 in the high band is reduced or disappears by adding the coefficient of the head related transfer function HRTF, that is, adding a short delay component of about 0.0001 seconds from the direct sound. do. On the other hand, in the voice signal 35'by the processing method of the present invention, an additional reflected sound component delayed from the direct sound by 0.001 to 10 seconds is added separately from the transmission function together with the initial reflected sound component. The reflected sound component added to the frequency band 92 does not disappear, and the stereophonic effect is maintained.

FIG. 13 shows the stereophonic device 10'of the second embodiment according to the present invention.
A stereophonic sound reproduction signal is generated from a sound source signal including the first frequency band 91 including the lowest region of the audible region and the second frequency band 92 higher than the first frequency band 91 and supplied to the earphone 50 or the headphone. The stereophonic processing device 10'is connected to a stereophonic processing unit 1 that performs out-of-head localization processing of at least the first frequency band 91 of the input signal, and is connected to the stereophonic processing unit 1 to cover at least a part of the second frequency band 92 of the input signal. The point that the reflection processing unit 2 for adding the reflected sound component is provided is the same as that of the stereophonic device 10 of FIG. FIG. 13 differs from FIG. 1 in that the coefficient generation unit 3 and the coefficient storage unit 4 are provided. The coefficient generator 3 receives sound source signals for a plurality of pairs of head-related transfer functions HRTF _L , HRTF _R from each sound source position 71 to the left and

right ears

70 L, 70 R at a plurality of angles from the horizontal plane azimuth angle of 45 ° to 90 °. Generates the coefficients of a plurality of pairs of left and right high-frequency suppression transfer functions BRTF _L and BRTF _R in which the band corresponding to the second frequency band 92 is suppressed or attenuated. The coefficient storage unit 4 accumulates the coefficients of a plurality of pairs of left and right high-frequency suppression transfer functions BRTF _L and BRTF _R. The three-dimensional processing unit 1 is given a selected pair of the coefficients of the high-frequency suppression transfer functions BRTF _L and BRTF _R stored in the coefficient storage unit 4, and passes the input signal to perform out-of-head localization processing. It is equipped with a three-dimensional processing filter 11a-11d.

Next, a three-dimensional processing method according to the present invention using the three-dimensional sound device 10'in FIG. 13 will be shown.
In the processing method using the stereophonic device 10', in order to generate and accumulate the transfer function coefficient, first, the left and right ears 70L from each sound source 71 position at multiple angles of horizontal plane azimuth angle 45 ° to 90 °, Prepare multiple pairs of left and right head-related transfer functions HRTF _L and HRTF _R up to 70R. Specifically, coefficient generation of multiple pairs of head-related transfer responses HRIR _L , HRIR _R (80) from each sound source 71 position to each left and

right ear

70 L, 70 R at multiple angles θ with a horizontal azimuth angle of 45 ° to 90 °. In Part 3, the head-related transfer functions HRTF _L and HRTF _R are prepared by performing arithmetic processing by Fourier transform (s81) and multiplying the common logarithm by 20 to convert it to a dB value (s82). Processing is performed in the same manner as in the flowchart (FIG. 8) of the first embodiment, except that a plurality of pairs are prepared.

Second, for the left and right head-related transfer functions HRTF _L and HRTF _R , the band 92'corresponding to the second frequency band is suppressed or attenuated by the coefficient generator 3 (s83), and the left and right multiple pairs. Generates the high frequency suppression transfer functions BRTF _L and BRTF _R. Third, for the high-frequency suppression transfer functions BRTF _L and BRTF _R , the logarithmic value is returned to the amplitude value (s84), inverse Fourier transform is performed (s85), and window function processing is performed (s86). Suppressive impulse responses BRIR _L , BRIR _R are obtained (90). Fourth, the coefficients h _B0 to h _Bn of a plurality of pairs of left and right high-frequency suppression transfer functions corresponding to the values of the high-frequency suppression impulse responses BRIR _L and BRIR _R are stored in the coefficient storage unit 4.

Next, in order to generate the processing signal of the three-dimensional processing unit 1, first, only the left and right pairs of the coefficients h _B0 to h _Bn of the left and right plurality of pairs of high-frequency suppression transfer functions stored in the coefficient storage unit 4 are generated. The coefficients of the high frequency suppression transfer function of are selected manually or automatically or according to the program. Secondly, for the three-dimensional processing filter 11a-11d to which the transfer function coefficient is given or not given in advance, the coefficient h of the high-frequency suppression transfer function of only one pair on the left and right at only one selected horizontal plane azimuth angle. _B0 to h _Bn is given. Third, the sound source signal is passed through the three-dimensional processing filter 11a-11d to which the coefficients h _B0 to h _Bn of the high-frequency suppression transfer function of only one pair on the left and right are given, and at least the first frequency band 91 is localized out of the head. Generates the processed signal of the processed three-dimensional processing unit 1.

Further, the processing signal of the three-dimensional processing unit 1 is input to the reflection processing unit 2, and a reflected sound component is added to at least a part of the second frequency band 92 including the high band exceeding the first frequency band 91 of the input signal. Generates the processing signal of the reflection processing unit 2. Finally, the three-dimensional sound reproduction signal generated by both the three-dimensional processing unit 1 and the reflection processing unit 2 is supplied to the earphone 50 or the headphones. Similar to the stereophonic device 10 (FIG. 1), the stereophonic device 10'(FIG. 13) and the processing method using the stereophonic device 10'(FIG. 1) have an excellent effect of forming high-quality stereophonic sound with a small amount of calculation and small individual differences. The effect is obtained.

In the first and second embodiments shown in FIGS. 1 and 13, the sound source signals input to the stereophonic devices 10 and 10'are processed in the order of the stereophonic processing unit 1 and the reflection processing unit 2. , The same effect can be obtained by reversing the order and processing in the order of the reflection processing unit 2 and the three-dimensional processing unit 1. Further, in the first and second embodiments, the two-channel stereophonic device 10,10'is shown, but two or more channels may be used. Further, in the first and second embodiments, the entire first and

second frequency bands

91 and 92 of the sound source signal are subjected to out-of-head localization processing by the stereoscopic processing unit 1, but only in the first frequency band 91. Overhead localization processing may be performed by the coefficient of the region suppression transfer function BRTF.

FIG. 14 shows a stereophonic device 10 "according to the third embodiment, which includes an M / S delay unit 93 connected to the reflection processing unit 2 at positions after both processing of the stereophonic processing unit 1 and the reflection processing unit 2. FIG. 15 shows an exemplary configuration of the M / S delay unit 93. The M / S delay unit 93 inputs the output signals from the reflection processing unit 2 to the L and R terminals, respectively, and converts them into center and side signals. The M / S converter 98, the reflection processing filter 21a that delays the center and side signals from the M / S converter 98, and the delay signal from the reflection processing filter 21a are input to the M and S terminals, respectively, and left and right signals. It is equipped with an L / R converter 99 that transmits left and right signals from the L and R terminals to the

output terminals

7a and 7b, respectively. The reflection processing filter 21a has the same reference numerals as those shown in FIG. For the center and side signals separated by the M / S converter 98, the delay device 22 of the reflection processing filter 21a delays each signal separately or only to one of them, and then L / R conversion is performed again. By returning to the left and right signals with the device 99, the stereoscopic effect and the sense of localization of the sound due to the hearth effect (a phenomenon in which the localization is deviated in the direction of the earliest arrival when the same sound is heard from multiple directions at the same volume) are emphasized. Even if both the center signal and the side signal contain the same component and the stereoscopic effect sounds weak, a relatively different delay is given to the center signal and the side signal, the localization is biased to the preceding sound, and the stereophonic sound is produced. Alternatively, a Fourier transform is performed and there are | L amplitude |-| R amplitude |, (L amplitude) ^2- (R amplitude) ² , or arctan (| L amplitude | / | R amplitude |). Separate the range (preferably arctan (0.7) ≤ arctan (| L amplitude | / | R amplitude |) ≤ arctan (1.3)) as the center signal component and the other as the side signal component, and return them by inverse Fourier transform. (|| is an absolute value). Generally, in music signals, the low frequency component is often placed in the center, so the coefficient of the low frequency region (preferably about 200Hz) is set to the above arctan value. Regardless, the center signal component may be used. Only the center signal component may be extracted and subtracted from the original signal to be used as the side signal component.

FIG. 16 shows a stereophonic device 10 ″ according to a fourth embodiment of the present invention, which includes an M / S processor 51 that separates a center signal and a side signal, but does not include a reflection processing unit 2. Out-of-head localization of at least the first frequency band 91 of the first and

second input terminals

6a and 6b for inputting the sound source signal to the three-dimensional processing device 10'''and the input signals from the first and second input ends 6a and 6b. It is provided with a three-dimensional processing unit 1 for processing. The three-dimensional processing unit 1 includes first and second three-

dimensional processing filters

11a and 11b connected to the first input end 6a, and third and fourth three-

dimensional processing filters

11c and 11d connected to the second input end 6a. M / S that inputs both output signals of the second and third three-dimensional processing filters 11b and 11c as left and right signals, separates them into a center signal and a side signal, and adds reflected sound to at least the side signal or only the side signal. One adder 16a that adds the output signal of the processor 51 and the other of the M / S processor 51 (from FIG. 2B adder 57b) and the output signal of the first stereoscopic processing filter 11a, and the M / S processor. The other adder 16b for adding the output signal of one of 51 (from FIG. 2B adder 57a) and the output signal of the fourth stereoscopic processing filter 11d is provided. Since the specific configuration of the M / S processor 51 is substantially the same as the configuration of reference numeral 51 illustrated in FIG. 2B, detailed description thereof will be omitted. The first to fourth three-dimensional processing filters 11a-11d are given the coefficients of only a pair of left and right transfer functions in advance or by selection. The transfer function of only one pair on the left and right is the band corresponding to the second frequency band of the sound source signal with respect to the pair of head-related transfer functions from the sound source position to each ear on the left and right at only one angle of the horizontal plane azimuth angle of 45 ° to 90 °. 'Is a high-frequency suppression transfer function with only a pair of left and right, suppressed or attenuated. With this configuration, the signal on the sound source side (output signal from the first or fourth

stereoscopic processing filter

11a or 11d) does not pass through the M / S processor 51, while the signal on the opposite side of the sound source (second or second). 3 The output signal from the three-

dimensional processing filter

11b or 11c) can be passed through the M / S processor 51 to add a reflected sound component to enhance human spatial acoustic recognition. In the case of a stereo sound source, since the sense of spaciousness of the sound is greatly affected by the side signal component, additional processing such as reflected sound may be performed only on the side signal component after conversion to the center and side signals.

The present invention may be a stereophonic processing program for operating a computer as each part of the

stereophonic processing apparatus

10, 10', 10 ", 10'''. In this case, the processing content of the function of each part may be. By being described in the program and executing the program on the computer, the processing of each part can be realized on the computer. For example, although not shown, a storage unit for storing the stereophonic processing program and a storage unit connected to the storage unit of the program The storage unit includes a control unit that controls the processing operation. The storage unit includes a computer-readable storage medium such as a magnetic storage device, an optical disk, a photomagnetic storage medium, and a semiconductor memory. Further, the first and second storage units are included. The stereophonic processing method described in detail according to the embodiment may be executed by the stereophonic processing program stored in the storage unit.

The stereophonic processing device, processing method, and processing program of the present invention can be used not only for music sound sources but also for all stereo sound sources such as environmental sounds, conversation sounds, and game sounds.

1 ... 3D processing unit, 2 ... Reflection processing unit, 3 ... Coefficient generation unit, 4 ... Coefficient storage unit, 6a, 6b ... Input end, 10,10'... 3D sound processing device, 11a-11d・・ Solid processing filter, 16a, 16b ・・ Adder, 21a, 21b ・・ Reflection processing filter, 22 ・・ Delayer, 23 ・・ Multiplier, 24 ・・ Frequency filter, 25a, 25b ・・ Branch point, 26a , 26b ... Adder points, 27a, 27b ... Adder, 51 ... M / S processor,

Claims

Stereophonic sound processing that generates a stereophonic sound reproduction signal from a sound source signal including a first frequency band including the lowest region of the audible region and a second frequency band higher than the first frequency band and supplies it to an earphone or a headphone. In the device
A three-dimensional processing unit that performs out-of-head localization processing at least the first frequency band of the input signal,
It is connected to the three-dimensional processing unit and includes a reflection processing unit that adds a reflected sound component to at least a part of the second frequency band of the input signal.
The three-dimensional processing unit includes a three-dimensional processing filter in which only a pair of left and right transfer function coefficients are given in advance or by selection.
The transfer function of only one pair on the left and right has a band corresponding to the second frequency band of the sound source signal with respect to the pair of head-related transfer functions from the sound source position to each ear on the left and right at only one angle of the horizontal plane azimuth angle of 45 ° to 90 °. A stereophonic processing device characterized by having only a pair of left and right high-frequency suppression transfer functions that have been suppressed or attenuated.
Stereophonic sound processing that generates a stereophonic sound reproduction signal from a sound source signal including a first frequency band including the lowest region of the audible region and a second frequency band higher than the first frequency band and supplies it to an earphone or a headphone. In the device
The first and second input ends that input the sound source signal to the three-dimensional processing device, and
It is provided with a three-dimensional processing unit that performs out-of-head localization processing at least in the first frequency band of the input signals from the first and second input ends.
The three-dimensional processing unit is
The first and second steric processing filters connected to the first input end,
The third and fourth steric processing filters connected to the second input end,
An M / S processor that inputs both output signals of the second and third stereoscopic processing filters as left and right signals, separates them into a center signal and a side signal, and adds a reflected sound component to the side signal.
The other adder that adds the output signal of one of the M / S processors and the output signal of the 4th solid processing filter, and
It is provided with one adder that adds the other output signal of the M / S processor and the output signal of the first solid processing filter.
In the first to fourth three-dimensional processing filters, the coefficients of only a pair of left and right transfer functions are given in advance or by selection.
The transfer function of only one pair on the left and right has a band corresponding to the second frequency band of the sound source signal with respect to the pair of head-related transfer functions from the sound source position to each ear on the left and right at only one angle of the horizontal plane azimuth angle of 45 ° to 90 °. A stereophonic processing device characterized by having only a pair of left and right high-frequency suppression transfer functions that have been suppressed or attenuated.
The reflection processing unit is
A reflection processing filter in which a signal is input from a branch point between an input end for inputting a sound source signal to a stereoscopic processing device and an earphone or a headphone, and a reflection processing filter.
An adder that adds the processing signal of the sound source signal or the three-dimensional processing unit and the feedback signal from the reflection processing filter at the addition point between the input end and the branch point.
The reflection processing filter is
A delayer that inputs a signal from a branch point, and
A multiplier that amplifies the signal from the delayer, and
The stereophonic processing apparatus according to claim 1, further comprising a frequency filter to which a coefficient is added, a reflected sound component is generated from a signal amplified by a multiplier, and the reflected sound component is transmitted to an adder as a feedback signal.
The stereophonic sound according to any one of claims 1 to 3, wherein the first frequency band is a band included in the range of 0 to 8 kHz, and the second frequency band is a band included in the range of 8 kHz to 96 kHz. Processing equipment.
Stereophonic sound processing that generates a stereophonic sound reproduction signal from a sound source signal including a first frequency band including the lowest region of the audible region and a second frequency band higher than the first frequency band and supplies it to an earphone or a headphone. In the method
The process of assigning the transfer function coefficient to the 3D processing filter of the 3D processing unit in advance,
A sound source signal or a processing signal of the reflection processing unit is input to a three-dimensional processing filter to which a coefficient of the transmission function is given in advance, and at least the first frequency band of the input signal is subjected to out-of-head localization processing to process the processing signal of the three-dimensional processing unit. Including the process of generating
The process of assigning the transfer function coefficient in advance is
The process of preparing a pair of left and right head-related transfer functions from the sound source position to the left and right ears at only one angle of horizontal plane azimuth of 45 ° to 90 °, and
The process of generating a pair of left and right high-frequency suppression transfer functions by suppressing or attenuating the band corresponding to the second frequency band for the pair of left and right head-related transfer functions.
The process of obtaining a pair of left and right high-frequency suppression impulse responses by arithmetically processing a pair of left and right high-frequency suppression transfer functions by inverse fury transformation.
A stereophonic processing method comprising a process of preliminarily assigning a coefficient of only a pair of left and right high-frequency suppression transfer functions to a stereophonic processing filter, which corresponds to a value of a high-frequency suppression impulse response.
Stereophonic sound processing that generates a stereophonic sound reproduction signal from a sound source signal including a first frequency band including the lowest region of the audible region and a second frequency band higher than the first frequency band and supplies it to an earphone or a headphone. In the method
The process of generating and accumulating transfer function coefficients,
A process in which a sound source signal or a processing signal of the reflection processing unit is input to the three-dimensional processing filter of the three-dimensional processing unit, and at least the first frequency band of the input signal is subjected to out-of-head localization processing to generate a processing signal of the three-dimensional processing unit. Including
The process of generating and accumulating transfer function coefficients is
The process of preparing multiple pairs of head-related transfer functions from each sound source position to each of the left and right ears at multiple angles of horizontal plane azimuth angle of 45 ° to 90 °, and
The process of generating a high-frequency suppression transfer function of multiple pairs of left and right by suppressing or attenuating the band corresponding to the second frequency band with respect to the head-related transfer function of multiple pairs of left and right.
The process of obtaining the high-frequency suppression impulse response of multiple pairs of left and right by arithmetically processing the high-frequency suppression transfer function of multiple pairs of left and right by inverse Fourier transform.
Including the process of accumulating the coefficients of multiple pairs of left and right high-frequency suppression transfer functions in the coefficient storage unit, which correspond to the values of the high-frequency suppression impulse response.
The process of generating the processing signal of the three-dimensional processing unit is
Of the coefficients of the high-frequency suppression transfer function of multiple pairs of left and right stored in the coefficient storage, the process of selecting the coefficients of the high-frequency suppression transfer function of only one pair of left and right, and
The process of assigning the coefficient of the high-frequency suppression transfer function of only one pair on the left and right at only one selected horizontal plane azimuth angle to the three-dimensional processing filter to which the transfer function coefficient is given or not given in advance, and
A stereophonic processing unit that has been subjected to out-of-head localization processing for at least the first frequency band by passing a sound source signal or a processing signal of the reflection processing unit through a stereophonic processing filter to which only a pair of left and right high-frequency suppression transfer function coefficients are assigned. A stereophonic processing method comprising a process of generating a processing signal of.
The process of inputting the processed signal of the three-dimensional processing unit or the sound source signal to the reflection processing unit and adding the reflected sound component to at least a part of the second frequency band of the input signal to generate the processed signal of the reflection processing unit.
The stereophonic processing method according to claim 5 or 6, which includes a process of supplying a stereophonic sound reproduction signal generated by both the stereophonic processing unit and the reflection processing unit to earphones or headphones.
The process of generating the processing signal of the reflection processing unit is
The process in which the processing signal of the three-dimensional processing unit or the sound source signal is input to the reflection processing unit and passes through the reflection processing unit, and
The stereophonic processing method according to claim 7, further comprising a process of adding an initial reflected sound component and a reflected sound component delayed by 0.001 to 10 seconds from the direct sound to the second frequency band of the input signal.
The stereophonic sound according to any one of claims 5 to 8, wherein the first frequency band is a band included in the range of 0 to 8 kHz, and the second frequency band is a band included in the range of 8 kHz to 96 kHz. Processing method.
A stereophonic processing program characterized in that a computer functions as each part of the stereophonic processing apparatus according to any one of claims 1 to 4.