JP2015154350A - Stereophonic acoustic apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve reproducibility of localization information of a sound image at a listening position of a listener when using a correction circuit having a correction property of approximating an inverse property of a transmission property in a space from a speaker to both ears of the listener.SOLUTION: A correction property C of a correction circuit 40 is set to minimize a differential between an acoustic property obtained by a space transmission property G and the correction property C of inversely approximating the space transmission property G, and a target acoustic property D. In such a case, as the target acoustic property D, a target acoustic property F is used which is multiplied by a target property frequency weight component Wfor emphasizing a frequency component to be a clue of sound image localization. As the differential between the acoustic property obtained by the space transmission property G and the correction property C and the target acoustic property D, a differential is used which is multiplied by a first error frequency weight component Wobtained by relatively enlarging a weight of a frequency band in which a crosstalk should be suppressed.

Description

  The present invention relates to a stereophonic sound device that can localize a sound image at a target position.

  For example, Patent Document 1 describes an apparatus that can correct a deviation of a center sound image, a spread of a left-right asymmetric sound field, and the like. The apparatus of Patent Document 1 obtains transfer functions corresponding to the transfer characteristics of the space from the left and right speakers to the left and right ears of the listener, and obtains a plurality of matrixes obtained by inverse matrices of the matrix having these transfer functions as elements. A correction circuit including a correction transfer function is provided. By providing such a correction circuit, the influence of the sound field characteristics of the above space is suppressed, and a natural sound field space can be provided to the listener.

JP 2001-57699 A

  However, in Patent Document 1, no matter how the correction circuit is designed, it is difficult to completely cancel the transfer characteristic of the space. Therefore, normally, as a correction circuit, a method such as a least square method is used, and an inverse filter that approximates the inverse characteristic of the transfer characteristic of the space is designed.

  Here, it is known that a human perceives the position (direction) of a sound image from the change in sound spectrum in addition to the sound pressure difference and phase difference of sound reaching both ears. In other words, from the sound source to the eardrum via the space, head, and ears, interference occurs due to reflection and wraparound by the head, reflection within the pinna, etc. The level changes. This frequency characteristic is called a head related transfer function (HRTF) and takes different values depending on the position (direction) of the sound source. For example, when this frequency characteristic is decomposed into a plurality of spectral peaks and spectral notches, the frequency of the first notch and the second notch on the low frequency side changes according to the vertical angle and is important for perceiving the angle. (See "Physical evaluation index of HRTF based on direction perception mechanism" Kazuhiro Iida et al., Proceedings of the Acoustical Society of Japan, September 2008). Therefore, conversely, if the frequency characteristics of the sound heard by the listener's left and right ears can be controlled to the frequency characteristics corresponding to the target localization position of the sound image, the sound image can be localized at a desired position. become able to.

  However, even if an acoustic signal having a frequency characteristic corresponding to the localization position of the sound image is created, the localization information of the sound image may not necessarily be reproduced at the listener's listening position when the above-described inverse filter is used.

  For example, consider a case where the inverse filter is designed by the method of least squares. In this case, the inverse filter is designed so that the error energy between the target acoustic characteristic to be heard by the listener and the actual acoustic characteristic actually heard by the listener is minimized as a whole. Therefore, for example, in a frequency band that is an important clue for the above-described sound image localization, there may be a case where the entire error energy is minimized by allowing a certain amount of error. Then, when the correction transfer function by the inverse filter is convoluted with an acoustic signal having frequency characteristics corresponding to the target sound image localization position, for example, it is sufficiently crossed in the frequency band that is an important clue for the above-described sound image localization. It is possible that the talk is not reduced and the reproducibility of the sound image localization information is lowered.

  The present invention has been made in view of the above points, and in the case of using a correction circuit having a correction characteristic that approximates a reverse characteristic of a transfer characteristic of a space from a speaker to a listener's both ears, An object of the present invention is to provide a stereophonic device capable of improving the reproducibility of localization information of a sound image at a listening position.

In order to achieve the above object, a stereophonic sound device according to the first invention comprises:
At least two speakers (70A, 70B) provided to correspond to the left and right ears of the listener;
Sound that creates a right acoustic signal (R _in ) to be output from the speaker (70A) corresponding to the right ear of the listener and a left acoustic signal (L _in ) to be output from the speaker (70B) corresponding to the left ear Signal generating means (30);
Correction means (40) for correcting the right acoustic signal and the left acoustic signal created by the acoustic signal creation means, and outputting the corrected right acoustic signal and the left acoustic signal from the corresponding speaker;
The sound signal creation means is configured so that the sound image is localized at a target position when the sound by the right sound signal is heard by the listener's right ear and the sound by the left sound signal is heard by the listener's left ear. , Create each acoustic signal,
The correction characteristic (C) by the correcting means is that when a predetermined right acoustic signal and a left acoustic signal are output at different timings, the sound of the output acoustic signal is to be heard by the listener. Transfer function (G _rr , G _ll ) indicating the transfer characteristics until reaching, and transfer function (G _rl , G _lr ) indicating the characteristics of crosstalk reaching the ear on the opposite side of the listener from the side to be listened to And the acoustic characteristic obtained by the correction characteristic that approximates the characteristic of the spatial transfer characteristic in the opposite direction so as to cancel the characteristic of the spatial transfer characteristic and the target acoustic characteristic (D ) To minimize the difference between
A first frequency weight component (W _1r ) for emphasizing a frequency component that is a clue to sound image localization is used as a target acoustic characteristic corresponding to the ear of the listener on the side where the output acoustic signal is to be heard. The target acoustic characteristic multiplied is used.

  Thus, in the first invention, when the correction characteristic (C) by the correcting means is set, the predetermined right acoustic signal and the left acoustic signal are output at different timings. Then, a first frequency weighting component for emphasizing a frequency component serving as a clue to sound image localization as a target acoustic characteristic corresponding to the ear of the listener on the side where one of the output acoustic signals is to be heard The correction characteristic by the correction means is set using the target acoustic characteristic multiplied by. For this reason, when an acoustic signal having a frequency characteristic corresponding to the target sound image localization position is corrected by the correction characteristic of the set correction means, the listener can emphasize the frequency component that is a clue to sound image localization. You will hear sound with characteristics. As a result, the listener can obtain localization information of the sound image from the emphasized frequency component even if some crosstalk remains in the frequency band which is an important clue for sound image localization. That is, according to the first invention, it is possible to improve the reproducibility of the localization information of the sound image at the listening position of the listener.

A stereophonic device according to the second invention is
At least two speakers (70A, 70B) provided to correspond to the left and right ears of the listener;
Sound that creates a right acoustic signal (R _in ) to be output from the speaker (70A) corresponding to the right ear of the listener and a left acoustic signal (L _in ) to be output from the speaker (70B) corresponding to the left ear Signal generating means (30);
Correction means (40) for correcting the right acoustic signal and the left acoustic signal created by the acoustic signal creation means, and outputting the corrected right acoustic signal and the left acoustic signal from the corresponding speaker;
The sound signal creation means is configured so that the sound image is localized at a target position when the sound by the right sound signal is heard by the listener's right ear and the sound by the left sound signal is heard by the listener's left ear. , Create each acoustic signal,
The correction characteristic (C) by the correcting means is that when a predetermined right acoustic signal and a left acoustic signal are output at different timings, the sound of the output acoustic signal is to be heard by the listener. Transfer function (G _rr , G _ll ) indicating the transfer characteristics until reaching, and transfer function (G _rl , G _lr ) indicating the characteristics of crosstalk reaching the ear on the opposite side of the listener from the side to be listened to And the acoustic characteristics obtained by the spatial transfer characteristics (G) having the element and the correction characteristics approximating the characteristics of the spatial transfer characteristics in the opposite direction so as to cancel the characteristics of the spatial transfer characteristics, and the target acoustic characteristics ( B, D) is set to minimize the difference,
Crosstalk is suppressed as the difference between the target acoustic characteristics and the acoustic characteristics obtained by the spatial transfer characteristics and correction characteristics corresponding to the ears of the listener on the opposite side of the output acoustic signal. A difference obtained by multiplying a second frequency weight component (W _2l ) in which the weight of the frequency band to be used is relatively large is used.

  Thus, in the second invention, when the correction characteristic (C) by the correcting means is set, the predetermined right acoustic signal and the left acoustic signal are output at different timings. As a difference between the target acoustic characteristic and the acoustic characteristic obtained by the spatial transfer characteristic and the correction characteristic corresponding to the ear of the listener on the opposite side to the side on which one of the outputted acoustic signals is to be heard The correction characteristic by the correction means is set using the difference obtained by multiplying the second frequency weight component in which the weight of the frequency band in which crosstalk should be suppressed is relatively increased. For this reason, the correction characteristics by the correction means minimize the difference between the acoustic characteristics obtained by the spatial transfer characteristics and the correction characteristics and the target acoustic characteristics, with particular emphasis on the characteristics of the frequency band in which crosstalk should be suppressed. Will be determined. As described above, in the second invention, since the frequency band in which crosstalk should be suppressed can be controlled, it is possible to reliably suppress crosstalk in the frequency band including the localization information of the sound image. As a result, according to the second invention, it is possible to improve the reproducibility of the localization information of the sound image at the listening position of the listener.

The stereophonic sound device according to the third invention is
At least two speakers (70A, 70B) provided to correspond to the left and right ears of the listener;
Sound that creates a right acoustic signal (R _in ) to be output from the speaker (70A) corresponding to the right ear of the listener and a left acoustic signal (L _in ) to be output from the speaker (70B) corresponding to the left ear Signal generating means (30);
Correction means (40) for correcting the right acoustic signal and the left acoustic signal created by the acoustic signal creation means, and outputting the corrected right acoustic signal and the left acoustic signal from the corresponding speaker;
The sound signal creation means is configured so that the sound image is localized at a target position when the sound by the right sound signal is heard by the listener's right ear and the sound by the left sound signal is heard by the listener's left ear. , Create each acoustic signal,
The correction characteristic (C) by the correcting means is that when a predetermined right acoustic signal and a left acoustic signal are output at different timings, the sound of the output acoustic signal is to be heard by the listener. Transfer function (G _rr , G _ll ) indicating the transfer characteristics until reaching, and transfer function (G _rl , G _lr ) indicating the characteristics of crosstalk reaching the ear on the opposite side of the listener from the side to be listened to The acoustic characteristics obtained by the spatial transfer characteristic (G) having the characteristic of the spatial transfer characteristic and the corrected transfer characteristic having a characteristic that approximates the characteristic of the spatial transfer characteristic in the opposite direction so as to cancel the characteristic of the spatial transfer characteristic, Is set to minimize the difference from the acoustic characteristics (D)
A first frequency weight component (W _1r ) for emphasizing a frequency component that is a clue to sound image localization is used as a target acoustic characteristic corresponding to the ear of the listener on the side where the output acoustic signal is to be heard. The difference between the target acoustic characteristic and the acoustic characteristic obtained by the spatial transmission characteristic and the corrected transmission characteristic, corresponding to the listener's ear opposite to the side to be listened to, using the target acoustic characteristic multiplied As described above, a difference obtained by multiplying the _second frequency weight component (W ₂ ) in which the weight of the frequency band in which crosstalk should be suppressed is relatively large is used.

  As described above, the stereophonic device according to the third aspect of the invention uses both the first frequency weighting component in the first aspect and the second frequency weighting component in the second aspect at the same time. As a result, it is possible to simultaneously obtain the enhancement effect of the frequency component that is a clue to the sound image localization in the first invention and the crosstalk suppression effect in the desired frequency band in the second invention. Therefore, it is possible to finely control the frequency characteristics of the sound that the listener listens to by the correction characteristics of the correction means, and it is possible to further improve the reproducibility of the localization information of the sound image at the listener's listening position. Become.

Furthermore, in any one of the second and third inventions described above, the acoustic characteristic and the target obtained by the spatial transfer characteristic and the correction characteristic corresponding to the ear of the listener on the side where the outputted acoustic signal is to be heard. A difference obtained by multiplying the third frequency weight component (W _2r ) obtained by relatively increasing the weight of the frequency band that should enhance the followability to the target acoustic characteristic may be used as the difference from the acoustic characteristic to be performed. A third frequency weighting component (W _2r) is applied to the difference between the acoustic characteristic obtained by the spatial transfer characteristic and the correction characteristic corresponding to the ear of the listener on the side where the acoustic signal is to be heard and the target acoustic characteristic. ), The correction characteristic of the correction means is set with particular emphasis on the difference in the frequency band in which the weight is relatively large. As a result, the correction characteristic of the correction means is set so that the difference between the frequency bands described above becomes small, so that the acoustic characteristic obtained by the spatial transfer characteristic and the correction characteristic can follow the target acoustic characteristic. Can be increased. For example, if the frequency band that you want to follow the target acoustic characteristics is set so as to include frequency components that are clues for sound image localization, the reproducibility of the localization information of the sound image at the listener's listening position can be further improved. Can do.

  The reference numerals in the parentheses merely show an example of a correspondence relationship with a specific configuration in an embodiment described later in order to facilitate understanding of the present invention, and are intended to limit the scope of the present invention. Not intended.

  Further, the technical features described in the claims of the claims other than the features described above will become apparent from the description of embodiments and the accompanying drawings described later.

It is the block diagram which showed the structure of the stereophonic sound apparatus which concerns on embodiment. It is a graph which shows an example of the frequency characteristic of the sound which a listener listens with a right ear and a left ear when the sound generation source is located in the direction of 45 degrees right from the front of the listener. It is explanatory drawing for demonstrating the frequency characteristic of an acoustic. It is explanatory drawing for demonstrating the basics of the correction circuit for correct | amending the spatial transfer characteristic regarding the sound between the speaker of right side and the left side, and both ears of a listener, and the spatial transfer characteristic. It is explanatory drawing for demonstrating the method of determining each transfer function used as the element of a spatial transfer characteristic. It is explanatory drawing for demonstrating the malfunction which may occur when the approximate inverse characteristic of a filter is defined with the conventional method. FIG. 5 is a diagram illustrating a relationship of first and second frequency weight components with respect to a target acoustic characteristic, a correction circuit, and a spatial transfer characteristic in calculating an evaluation function. It is a graph which shows an example of a target characteristic frequency weight component. It is a graph which shows an example of the 1st error frequency weight component. It is a graph which shows an example of the 2nd error frequency weighting component. It is a graph which shows an example of the impulse response characteristic by each correction transfer function of the correction circuit of this embodiment. It is the figure which showed only the right side acoustic signal being input into a correction circuit, and making the left side acoustic signal zero. It is explanatory drawing for demonstrating calculating | requiring the approximate inverse characteristic of a filter using only a target characteristic frequency weight component, without using the 1st and 2nd error frequency weight component. It is a graph which shows the frequency characteristic of the acoustic signal observed in the listening position corresponding to a listener's right ear and left ear when the right acoustic signal is corrected by the approximate inverse characteristic of the filter obtained in FIG. It is explanatory drawing for demonstrating calculating | requiring the approximate inverse characteristic of a filter using only a 1st error frequency weight component, without using a target characteristic frequency weight component and a 2nd error frequency weight component. It is a graph which shows the frequency characteristic of the acoustic signal observed in the listening position corresponding to a listener's right ear and left ear when the right acoustic signal is correct | amended by the approximate inverse characteristic of the filter calculated | required in FIG. It is explanatory drawing for demonstrating calculating | requiring the approximate inverse characteristic of a filter using a target characteristic frequency weight component and a 1st error frequency weight component, without using a 2nd error frequency weight component. It is a graph which shows the frequency characteristic of the acoustic signal observed in the listening position corresponding to a listener's right ear and a left ear when the right acoustic signal is correct | amended by the approximate inverse characteristic of the filter calculated | required in FIG. It is explanatory drawing for demonstrating calculating | requiring the approximate inverse characteristic of a filter using all the target characteristic frequency weighting components and the 1st and 2nd error frequency weighting components. It is a graph which shows the frequency characteristic of the acoustic signal observed in the listening position corresponding to a listener's right ear and a left ear when the right acoustic signal is correct | amended by the approximate inverse characteristic of the filter calculated | required in FIG.

  Hereinafter, a stereophonic device according to an embodiment of the present invention will be described with reference to the drawings. The stereophonic device according to the present embodiment generates a stereophonic sound field by outputting sound signals from two or more speakers to localize a sound image at a desired position in a three-dimensional space. In the present embodiment, an example in which the three-dimensional sound device is applied to a vehicle as a vehicle will be described.

  The vehicle has an obstacle detection device that detects obstacles around the vehicle and notifies the driver, a parking assistance device that provides guidance for guiding the vehicle to the parking position when parking, and further guides the vehicle to the destination. Various in-vehicle devices such as a navigation device may be mounted.

  In such a case, for example, when an obstacle is detected by the obstacle detection device, if a notification is given by voice from the direction in which the obstacle exists, the driver of the vehicle intuitively has the direction in which the obstacle exists. Can be grasped. Further, when guidance is given by the parking assistance device, for example, when instructing steering to the right, when the guidance is heard from the right direction of the driver, the driver intuitively understands the guidance content. Easy to understand. Further, when performing route guidance in the navigation device, if the route guidance sound is heard from the direction corresponding to the direction to turn left or right, the driver can intuitively recognize the direction to proceed.

  As described above, in a vehicle such as a vehicle, it may be effective to direct the occupant's attention in a certain direction when performing notification or warning. As will be described in detail later, the stereophonic device according to the present embodiment can accurately reproduce the localization information for perceiving the direction in which the sound image is localized at the listening position of the listener. Therefore, the stereophonic device according to the present embodiment is very effective when used in the vehicle as described above when it is desired to direct the attention of the passenger (listener) in a certain direction. However, the use of the three-dimensional sound device according to the present embodiment is not limited to the above-described example, and can be applied to sound devices such as various facilities and homes.

  FIG. 1 shows the configuration of the stereophonic device according to the present embodiment. In FIG. 1, the stereophonic device 20 is connected to the in-vehicle device 10 such as the obstacle detection device, the parking support device, and the navigation device described above. The in-vehicle device 10 has a sound source signal such as a warning sound and a notification sound for warning and notification, and a sound image indicating from which direction the sound source signal should be heard by the driver (listener) of the vehicle. The position information is output to the stereophonic device 20.

  The stereophonic device 20 includes an acoustic signal creation circuit 30, a correction circuit 40, D / A converters 50A and 50B, amplifiers 60A and 60B, and speakers 70A and 70B. Although FIG. 1 shows a configuration using two-channel speakers, it is possible to use a speaker configuration of three or more channels.

  Based on the sound source signal and sound image position information obtained from the in-vehicle device 10, the sound signal creation circuit 30 is as if the listener can hear the contents of the sound source signal from the sound image position corresponding to the sound image position information in the three-dimensional space. An acoustic signal having a perceived frequency characteristic is created. For example, the acoustic signal creation circuit 30 can create an acoustic signal having the above-described frequency characteristics by convolving a head-related transfer function set according to the sound image localization position into the sound source signal.

  The acoustic signal generation circuit 30 stores a plurality of acoustic signals corresponding to the contents of various sound source signals and having frequency characteristics corresponding to the direction of the sound image position to be perceived by the listener in advance. The sound signal to be output may be created by selecting the sound signal having the corresponding frequency characteristic based on the sound source signal and the sound image position information from the in-vehicle device 10.

  FIG. 2 shows an example of the frequency characteristics of the sound that the listener listens to with the right and left ears when the sound source is located in the direction of 45 degrees to the right from the front of the listener. As in the example shown in FIG. 2, when the position of the sound generation source is deviated from the front, the frequency characteristics of the sound heard by the listener are different between the left and right ears. When the angle in the left-right direction (azimuth angle) at which the sound source is located changes, the frequency characteristics of the sound that is heard by the listener's right and left ears also change.

  In addition, as shown in FIG. 3, when the acoustic frequency characteristic is decomposed into a plurality of spectral peaks P1 to P4 and spectral notches N1 to N3, the frequencies of the first notch N1 and the second notch N2 on the low frequency side are It is known to change according to the angle in the vertical direction with the listener's directly facing position as a reference.

  As described above, the frequency characteristic (head-related transfer function) of the sound takes different values depending on the direction of the sound source with respect to the listener. Therefore, localization information for localizing the sound image to a desired position is provided by controlling the frequency characteristic of the sound reaching the left and right ears (the eardrum) of the listener to a frequency characteristic corresponding to the localization position of the target sound image. Will be able to.

Therefore, the acoustic signal generation circuit 30 includes the right acoustic signal R _in output from the right speaker 70A disposed so as to correspond to the listener's right ear, and the left speaker disposed so as to correspond to the listener's left ear. to create a left sound signal _{L in} output from the 70B. More specifically, the acoustic signal generation circuit 30 is positioned at a target position when the sound of the right acoustic signal is heard by the listener's right ear and the sound of the left acoustic signal is heard by the listener's left ear. A right acoustic signal R _in and a left acoustic signal L _in having frequency characteristics at which the sound image is localized are created.

However, the right and left speakers 70A, the 70B, be output on the right sound signal R _in and the left audio signal L _in having a frequency characteristic as described above, the acoustic space between each ear of the listener Depending on the transfer characteristics, the sound of each acoustic signal may not be correctly heard by the ear on the corresponding side of the listener. Further, for example, the sound that should be listened to changes due to the characteristics of crosstalk in which the sound generated by the right sound signal R _in output from the right speaker 70A is heard even in the left ear of the listener. May end up.

Therefore, conventionally, a correction circuit 40 that corrects the right and left acoustic signals R _in and L _in output from the right and left speakers 70A and 70B in order to reduce the influence of the acoustic transmission characteristics and crosstalk characteristics of the space described above. Was established. The conventional correction circuit 40 will be briefly described with reference to FIG.

As shown in FIG. 4, the right and left speakers 70A, the spatial transfer characteristic 80 on Acoustic between ears of 70B and listener, acoustic by right audio signal R _in and the left audio signal L _in, respectively, Transfer functions G _rr and G _ll indicating transfer characteristics until the listener reaches the ear to be listened to, and transfer indicating characteristics of crosstalk reaching the ear on the opposite side of the listener to be listened to The functions G _rl and G _lr can be expressed as elements. Note that these transfer functions G _rr , G _rl , G _ll , and G _lr are usually determined based on impulse response characteristics, that is, the right and left speakers 70A and 70B have different timings as shown in FIG. The acoustic signal as shown in FIG. 5B is observed at a position corresponding to both ears of the listener while giving a pulsed acoustic signal as shown in FIG. The observed acoustic signal is sampled to obtain a discrete number sequence, and the discrete number sequence is z-transformed to obtain each transfer function G _rr , G _rl , G _ll , G _lr which is an element of the spatial transfer characteristic 80. be able to.

The conventional correction circuit 40 is configured to have a correction characteristic having a characteristic that approximates the characteristic of the spatial transfer characteristic 80 in the opposite direction in order to cancel the characteristic of the spatial transfer characteristic 80. For example, Patent Document 1 defines a matrix with 2 rows and 2 columns each having transfer functions G _rr , G _rl , G _ll , and G _lr representing the spatial transfer characteristics 80 as elements, and each matrix obtained by the inverse matrix It is described that the correction circuit 40 is configured to have a correction element corresponding to the transfer function.

  However, in reality, it is difficult to completely cancel the spatial transfer characteristic 80 by the correction characteristic of the correction circuit 40. For this reason, generally, using a digital filter such as an IIR filter or FIR filter, the least square method or the like is used so that the error energy between the target acoustic characteristic and the actually observed acoustic characteristic is minimized. Thus, the inverse characteristic of the spatial transfer characteristic 80 is approximated.

  Specifically, for example, when the target acoustic characteristic is B, the approximate inverse characteristic of the filter that is the correction characteristic of the correction circuit 40 is C, and the spatial transfer characteristic is G, the square value of the evaluation function E shown in the following Equation 1 The approximate inverse characteristic C of the filter is determined so that (corresponding to error energy) is minimized.

(Equation 1)
E = BC
Therefore, the approximate inverse characteristic C of the filter can be expressed by the following formula 2.

(Equation 2)
C = (G ^T · G) (G ^T · B)
However, in this case, since the error between the target acoustic characteristic B and the actual acoustic characteristic (corresponding to C · G) is considered uniformly over the entire frequency range, from the viewpoint of reproducing the localization information of the sound image May not necessarily determine the approximate inverse characteristic C of the optimum filter.

For example, FIG. 6 shows a case where the right acoustic signal R _in input to the correction circuit 40 is corrected by the approximate inverse characteristic C of the filter determined by the above-described method and then output from the right and left speakers 70A and 70B. 2 shows the frequency characteristics of the acoustic signal observed at the listening position corresponding to the right ear and the left ear of the listener. From the graph of FIG. 6, although the right acoustic signal R _in should originally be heard only by the right ear of the listener, the acoustic signal heard by the left ear has a frequency of 7 to 9 kHz. It can be seen that the band is not sufficiently attenuated and as a result, crosstalk occurs.

  Here, as described above, the amplitude spectrum of the head-related transfer function is an important clue for localization of the sound image (called a spectral cue), and there are many spectral cues in the frequency band of about 2 to 15 kHz. . For this reason, as shown in FIG. 6, if crosstalk occurs in a frequency band where there are many spectral cues, the reproducibility of the sound image localization information decreases, and the listener perceives the direction of the sound image relative to himself / herself. It becomes difficult to do.

In order to solve such a problem, in this embodiment, when the approximate inverse characteristic C of the filter as the correction circuit 40 is determined, the frequency with respect to the target acoustic characteristic B (Br, Bl) is set as shown in FIG. Target characteristic frequency weighting component W ₁ giving weight (W _1r , W ₁₁ ), and error frequency giving frequency weight to error between weighted target acoustic characteristic D and actual acoustic characteristic (corresponding to C · G) The evaluation function E is obtained using the weight component W ₂ (W _2r , W _2l ). The evaluation function E in the present embodiment can be expressed as Equation 3 below.

(Equation 3)
D = W ₁ B
E = W ₂ (DCG)
And the approximate inverse characteristic C of the filter in this embodiment can be represented by the following numerical formula 4.

(Equation 4)
^{_{C = (G T W 2 T}} W 2 G) -1 (G T W 2 T W 2 D)
Figure 8 (a), (b) is a graph showing an example of a target characteristic frequency weight component _{W 1.} 8 (a) is a target characteristic frequency weight component W ₁ shows the frequency axis, Fig. 8 (b), the target characteristic frequency weight component W ₁ shown in FIG. 8 (a) is converted to the time domain It is shown on the time axis.

Further, FIG. 9 (a), (b), among the error frequency weight component W _2, one output of a predetermined right audio signal and the left audio signal for determining the approximate inverse characteristic C of the filter as a compensation circuit 40 The first error frequency weighting component (of the right acoustic signal) multiplied by the error (D-C · G) corresponding to the ear of the listener on the opposite side of the output acoustic signal. In the case of W _2l and W _{2r in} the case of the left acoustic signal). FIG. 9A shows the first error frequency weight component on the frequency axis, and FIG. 9B shows the time error by converting the first error frequency weight component of FIG. 9A to the time domain. It is shown in the axis.

Further, FIG. 10 (a), (b), among the error frequency weight component W _2, when one of the predetermined right audio signal and the left audio signal for determining the approximate inverse characteristic C of the filter is output, the A second error frequency weighting component (W _{2r in} the case of a right acoustic signal, left acoustic signal) multiplied by an error (DCG) corresponding to the ear of the listener on the side where the output acoustic signal is to be heard In this case, W _2l is applicable). FIG. 10A shows the second error frequency weighting component on the frequency axis, and FIG. 10B converts the second error frequency weighting component of FIG. It is shown in the axis.

Target characteristic frequency weight component W ₁ corresponds to the first frequency weight components of the present invention, as shown in FIG. 8 (a), and relatively large weight of the spectral peak is an important clue sound localization Conversely, the spectral notch weight is set to be relatively small. That is, the target characteristic frequency weight component W ₁ is made shall be able to emphasize the frequency component as a clue to the sound image localization (Spectral peaks and spectral notches). Accordingly, using the target acoustic characteristics D multiplied by the target characteristic frequency weight component W _1, if you set the approximation filter inverse characteristic C (correction characteristic of the correction circuit 40), an acoustic signal corrected by the correction characteristic Therefore, it is possible to provide the listener with sound having frequency characteristics in which frequency components that are clues for sound image localization are emphasized.

The target characteristic frequency weighting component W ₁ preferably has different characteristics depending on the position (direction) where the sound image should be localized. This is because frequency components (spectral peaks and spectral notches) that are clues to sound image localization change according to the localization position of the sound image. In that case, as the correction characteristic of the correction circuit 40 (approximate inverse characteristic C of the filter), the number of correction characteristics corresponding to the position (direction) where the sound image should be localized is set in advance. The plurality of correction characteristics set in this way are stored in the correction circuit 40. Then, the correction circuit 40 selects a corresponding correction characteristic based on the sound image localization information from the in-vehicle device 10, and corrects the input acoustic signal using the selected correction characteristic.

However, by setting the target characteristic frequency weight component W ₁ so that it can substantially cover the change in the frequency component as a clue to the sound image localization (Spectral peaks and spectral notches), regardless of the position (direction) to localize a sound image It is also possible to use a common target characteristic frequency weighting component W1.

First error frequency weight component W ₂ corresponds to the second frequency weighting components of the present invention, as shown in FIG. 9 (a), the frequency to be suppressed crosstalk frequency band containing a large amount of information of the sound image localization The band is determined such that the weight of the frequency band becomes relatively large. The first error frequency weight component W ₂ is the acoustic characteristics to the side to the acoustic signal is heard corresponds to the opposite side of the listener's ear, the acoustic characteristics and the target obtained by the correction characteristic and spatial transfer characteristics Is multiplied by the error. In this manner, by calculating the evaluation function E using the first error frequency weight component W _2, with particular emphasis on the characteristics of the frequency band to be suppressed crosstalk, approximation of the spatial transfer characteristic G and filters The approximate inverse characteristic C of the filter is determined so as to minimize the difference between the acoustic characteristic obtained by the inverse characteristic C and the weighted target acoustic characteristic D. As a result, the frequency band in which the crosstalk should be suppressed can be controlled, so that the crosstalk in the frequency band including the localization information of the sound image can be reliably suppressed.

As shown in FIG. 9 (a), the first error frequency weight component W ₂ may define the weight of the entire frequency band containing a large amount of information of the sound image localization so that relatively large. Therefore, the first error frequency weight component W ₂ does not depend on the position to localize a sound image (direction), it is possible to use those same characteristics. However, with respect to the first error frequency weight component W _2, depending on the position to localize a sound image (direction), it may be used of different characteristics. The first error frequency weighting characteristic of the component W ₂ shown in FIG. 9 (a) is merely an example, unless defined as the weight of the frequency band to be suppressed crosstalk is relatively large, different properties May be used.

Second error frequency weight component W ₂ corresponds to the third frequency weight components of the present invention, as shown in FIG. 10 (a), the weight of the frequency band to enhance the conformability to the acoustic target properties It is determined to be relatively large. In the example shown in FIG. 10A, the frequency band that should improve the followability to the target acoustic characteristic is determined to be a frequency band that includes a frequency component that is a clue to sound image localization. The second error frequency weight component W ₂ corresponds to the ear side of a listener to the sound signal is heard, the error between the acoustic characteristics of the acoustic characteristics and the target obtained by the correction characteristic spatial transfer characteristic Multiplied. In this manner, by calculating the evaluation function E using the second error frequency weight component W _2, the difference in the frequency band the size of the weight is relatively large by the second error frequency weight component W ₂ With particular emphasis, the approximate inverse characteristic C of the filter is determined. As a result, the approximate inverse characteristic C of the filter is set so that the difference in the frequency band described above becomes small, so that the acoustic characteristic obtained by the spatial transfer characteristic and the correction characteristic follows the target acoustic characteristic. Can increase the sex. Therefore, if the frequency band you want to follow the target acoustic characteristics is set so as to include the frequency component that is a clue to sound image localization, the reproducibility of the localization information of the sound image at the listener's listening position can be further improved. Can do.

Incidentally, according to Equation 4, in order to determine an approximate inverse characteristic C of the filter, the target acoustic characteristic D, space transfer characteristic G, it is necessary to define the first and second error frequency weight component W ₂ a represents a matrix, each item Can be obtained from a change in characteristics on the time axis. For example, in determining the matrix of target acoustic characteristic D, first, in the frequency axis, multiplying the target characteristic frequency weight component W ₁ to the target acoustic characteristic B. Next, the weighted target acoustic characteristic D is converted into the time domain. Then, by sampling the weighted target acoustic characteristics D every predetermined time on the time axis, a matrix of the target acoustic characteristics D whose elements are the sampling values can be determined. It is possible to define a matrix in the same manner for other items.

11A to 11D, the approximate inverse characteristic C of the filter is obtained as described above, and each correction transfer function C _rr , C _rl for realizing the approximate inverse characteristic C in the correction circuit 40 is obtained. , C _ll , C _lr are graphs showing an example of impulse response characteristics by the respective correction transfer functions C _rr , C _rl , C _ll , C _lr .

Further, the setting of the approximate inverse characteristic C of the filter as the correction circuit 40 will be described. First, in this embodiment, a predetermined right acoustic signal and left acoustic signal for determining the approximate inverse characteristic C of the filter are output at different timings. For example, as shown in FIG. 12, when the right acoustic signal R _in is output, the right target characteristic frequency weighting component W _1r gives a frequency weight to the right acoustic signal R _in indicating the target acoustic characteristic. Further, for an error (D-CG) corresponding to the left ear of the listener on the opposite side to the side on which the right acoustic signal is to be heard, a left error frequency weight component W ₂₁ that is a first error frequency weight component is Give frequency weights. Furthermore, the right error frequency weight component _W2r, which is the second error frequency weight component, gives a frequency weight to the error (D-CG) corresponding to the right ear of the listener on the side where the right acoustic signal is to be heard. . As described above, the right and left errors (D-CG) to which the frequency weights are given by the right target characteristic frequency weight component W _1r , the left error frequency weight component W ₂₁ and the right error frequency weight component W _2r are minimized. For example, the approximate inverse characteristic C of the filter is determined by the least square method. When setting the approximate inverse characteristic C of the filter, as described above, the frequency weight components W ₁ and W ₂ may be changed according to the localization position of the sound image.

However, the approximate inverse characteristic C of the filter set when the right acoustic signal is output is only a correction transfer function (C _rr , C _rl ) for correcting the right acoustic signal Rin, as shown in FIG. is there. Therefore, the correction transfer function ( _Cll , _Clr ) that outputs the left acoustic signal and corrects the right acoustic signal Rin at different timings is set. Thereby, the approximate inverse characteristic C of the filter can be set.

Next, the operational effects obtained by each of the target characteristic frequency weight component W ₁ , the first error frequency weight component W ₂ , and the second error frequency weight component W ₃ will be described in detail with reference to the drawings. In the following description, correction transfer functions (C _rr , C _rl ) for correcting the right acoustic signal are set. Accordingly, the target characteristic frequency weight component _{W 1} is the right target characteristic frequency weight component _{W 1r,} and the first error frequency weight component _{W 2} is left error frequency weight component _{W 2l,} and the second error frequency weight component _{W 2} is It becomes the right error frequency weight component _W2r .

First, it will be described effects obtained by the target characteristic frequency weight component W _1. Therefore, as shown in FIG. 13, the approximate inverse characteristic C of the filter is obtained using only the target characteristic frequency weight component W _1r without using the first and second error frequency weight components W ₂₁ and W _2r .

When the correction transfer functions C _rr and C _rl of the approximate inverse characteristic C of the filter set in this way are _convolved with the right acoustic signal R _in and output from the respective speakers 70A and 70B, FIG. 14 shows. An acoustic signal having such frequency characteristics was observed at listening positions corresponding to the right and left ears of the listener. From the graph of FIG. 14, it is understood that the frequency component that is a clue to sound image localization is emphasized with respect to the acoustic signal that is heard by the listener's right ear. Therefore, even if some crosstalk remains, the listener can sufficiently obtain localization information of the sound image from the emphasized frequency component.

Next, as shown in FIG. 15, the approximate inverse characteristic C of the filter is obtained by using only the first error frequency weight component W ₂₁ without using the target characteristic frequency weight component W _1r and the second error frequency weight component W _2r. Asked.

When the right acoustic signal R _in input to the correction circuit 40 is corrected by the approximate inverse characteristic C of the filter set in this way and output from the right and left speakers 70A and 70B, the frequency shown in FIG. A characteristic acoustic signal was observed at listening positions corresponding to the listener's right and left ears. From the graph of FIG. 16, it can be seen that the acoustic signal that is heard in the listener's left ear is sufficiently attenuated (approximately -40 dB or less). In particular, it can be seen that there are many spectrum cues that are clues for sound image localization, and the degree of attenuation in the frequency band where crosstalk should be suppressed is large. From this result, the magnitude (attenuation degree) of a signal in a desired frequency band can be controlled by the first error frequency weighting component _{W2l in} the frequency characteristics of the sound that is heard by the listener's left ear. Recognize. Therefore, the first error frequency weighting component _W2l can surely suppress the crosstalk in the frequency band including the localization information of the sound image.

Next, as shown in FIG. 17, without using the second error frequency weight component W _2r, with the target characteristic frequency weight component W _1r, and a first error frequency weight component W _2l, approximate inverse characteristic of the filter C was determined.

In this case, when the right acoustic signal R _in inputted to the correction circuit 40 is corrected by the approximate inverse characteristic C of the obtained filter and outputted from the right and left speakers 70A and 70B, the frequency characteristic shown in FIG. 18 is obtained. The acoustic signal possessed was observed at the listening position corresponding to the listener's right and left ears. From the graph of FIG. 18, the enhancement effect of the frequency component that is a clue to the sound image localization explained using FIGS. 13 and 14 and the crosstalk suppression effect in the desired frequency band explained using FIGS. You can see that you can get. Thus, by defining the approximate inverse characteristic C of the filter using both the target characteristic frequency weighting component W _1r and the first error frequency weighting component W ₂₁ , the frequency characteristic of the sound that the listener listens to is controlled more finely. Therefore, the reproducibility of the localization information of the sound image at the listening position of the listener can be further improved.

Finally, as shown in FIG. 19, the approximate inverse characteristic C of the filter was obtained using all of the target characteristic frequency weight component W _1r and the first and second error frequency weight components W ₂₁ and W _2r .

In this case, when the right acoustic signal R _in input to the correction circuit 40 is corrected by the approximate inverse characteristic C of the obtained filter and output from the right and left speakers 70A and 70B, the frequency characteristics as shown in FIG. 20 are obtained. The acoustic signal possessed was observed at the listening position corresponding to the listener's right and left ears. Comparing the graph of FIG. 18 with the graph of FIG. 20, it can be seen that the frequency characteristic of the sound heard by the right ear is closer to the target frequency characteristic of the sound. In this way, by using the second error frequency weighting component W _2r , it is possible to improve the followability of the acoustic characteristic obtained by the spatial transfer characteristic G and the approximate inverse characteristic C of the filter to the target acoustic characteristic. . Therefore, the reproducibility of the localization information of the sound image at the listening position of the listener can be further improved.

  The preferred embodiments of the present invention have been described above. However, the present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the spirit of the present invention. It is.

For example, as described in the embodiment described above, by using the target characteristic frequency weight component W ₁ and the first error frequency weight component W ₂ alone, even if the determined approximate inverse characteristic C of the filter, the sound image This is effective in improving the reproducibility of localization information. Therefore, when setting the approximate inverse characteristic C of the filter, it may be used the target characteristic frequency weight component W ₁ and the first error frequency weight component W ₂ alone. Further, without using the second error frequency weight component W _2, using the target characteristic frequency weight component W ₁ by a first error frequency weight component W _2, it may be set approximate inverse characteristic C of the filter.

DESCRIPTION OF SYMBOLS 10 In-vehicle apparatus 20 Stereophonic device 30 Acoustic signal creation circuit 40 Correction circuit 70A Right speaker 70B Left speaker

Claims (10)

At least two speakers (70A, 70B) provided to correspond to the left and right ears of the listener;
A right acoustic signal (R _in ) to be output from the speaker (70A) corresponding to the right ear of the listener and a left acoustic signal (L _in ) to be output from the speaker (70B) corresponding to the left ear are created. Acoustic signal creation means (30);
Correction means (40) for correcting the right acoustic signal and the left acoustic signal created by the acoustic signal creation means and outputting the corrected right acoustic signal and left acoustic signal from the corresponding speaker; A stereophonic device,
The sound signal generating means is configured to detect a sound image at a target position when sound from the right sound signal is heard by the right ear of the listener and sound from the left sound signal is heard by the left ear of the listener. Create each acoustic signal so that
The correction characteristic (C) by the correction means is such that when the predetermined right acoustic signal and the left acoustic signal are output at different timings, the sound by the output acoustic signal is to be listened to by the listener. Transfer function (G _rr , G _ll ) indicating a transfer characteristic until reaching the ear of the user, and a transfer function (G _rl , indicating a characteristic of crosstalk reaching the ear on the opposite side of the listener to be listened to) An acoustic characteristic obtained by the spatial transfer characteristic (G) having G _lr ) as an element and the correction characteristic approximating the characteristic of the spatial transfer characteristic in the opposite direction so as to cancel the characteristic of the spatial transfer characteristic; Is set so as to minimize the difference from the acoustic characteristic (D),
The output acoustic signal is multiplied by a first frequency weighting component for emphasizing a frequency component that is a clue to sound image localization, as the target acoustic characteristic corresponding to the ear of the listener on the side to be listened to. A three-dimensional acoustic apparatus using a target acoustic characteristic. At least two speakers (70A, 70B) provided to correspond to the left and right ears of the listener;
A right acoustic signal (R _in ) to be output from the speaker (70A) corresponding to the right ear of the listener and a left acoustic signal (L _in ) to be output from the speaker (70B) corresponding to the left ear are created. Acoustic signal creation means (30);
Correction means (40) for correcting the right acoustic signal and the left acoustic signal created by the acoustic signal creation means and outputting the corrected right acoustic signal and left acoustic signal from the corresponding speaker; A stereophonic device,
The sound signal generating means is configured to detect a sound image at a target position when sound from the right sound signal is heard by the right ear of the listener and sound from the left sound signal is heard by the left ear of the listener. Create each acoustic signal so that
The correction characteristic (C) by the correction means is such that when the predetermined right acoustic signal and the left acoustic signal are output at different timings, the sound by the output acoustic signal is to be listened to by the listener. Transfer function (G _rr , G _ll ) indicating a transfer characteristic until reaching the ear of the user, and a transfer function (G _rl , indicating a characteristic of crosstalk reaching the ear on the opposite side of the listener to be listened to) An acoustic characteristic obtained by the spatial transfer characteristic (G) having G _lr ) as an element and the correction characteristic approximating the characteristic of the spatial transfer characteristic in the opposite direction so as to cancel the characteristic of the spatial transfer characteristic; Is set so as to minimize the difference from the acoustic characteristics (B, D)
As a difference between the acoustic characteristic obtained by the spatial transfer characteristic and the correction characteristic corresponding to the ear of the listener on the opposite side to the side on which the output acoustic signal is to be heard and the target acoustic characteristic, A stereophonic sound device characterized by using a difference multiplied by a second frequency weight component in which a weight of a frequency band in which crosstalk should be suppressed is relatively increased. At least two speakers (70A, 70B) provided to correspond to the left and right ears of the listener;
A right acoustic signal (R _in ) to be output from the speaker (70A) corresponding to the right ear of the listener and a left acoustic signal (L _in ) to be output from the speaker (70B) corresponding to the left ear are created. Acoustic signal creation means (30);
Correction means (40) for correcting the right acoustic signal and the left acoustic signal created by the acoustic signal creation means and outputting the corrected right acoustic signal and left acoustic signal from the corresponding speaker; A stereophonic device,
The sound signal generating means is configured to detect a sound image at a target position when sound from the right sound signal is heard by the right ear of the listener and sound from the left sound signal is heard by the left ear of the listener. Create each acoustic signal so that
The correction characteristic (C) by the correction means is such that when the predetermined right acoustic signal and the left acoustic signal are output at different timings, the sound by the output acoustic signal is to be listened to by the listener. Transfer function (G _rr , G _ll ) indicating a transfer characteristic until reaching the ear of the user, and a transfer function (G _rl , indicating a characteristic of crosstalk reaching the ear on the opposite side of the listener to be listened to) An acoustic characteristic obtained by the spatial transfer characteristic (G) having G _lr ) as an element and the correction characteristic approximating the characteristic of the spatial transfer characteristic in the opposite direction so as to cancel the characteristic of the spatial transfer characteristic; Is set so as to minimize the difference from the acoustic characteristic (D),
The output acoustic signal is multiplied by a first frequency weighting component for emphasizing a frequency component that is a clue to sound image localization, as the target acoustic characteristic corresponding to the ear of the listener on the side to be listened to. As the difference between the acoustic characteristic obtained by using the target acoustic characteristic and the spatial transmission characteristic and the correction characteristic corresponding to the ear of the listener on the opposite side to the side to be heard and the target acoustic characteristic, 3. A stereophonic sound device using a difference obtained by multiplying a second frequency weighting component in which a weight of a frequency band in which crosstalk is to be suppressed is relatively increased.   As the difference between the acoustic characteristic obtained by the spatial transfer characteristic and the correction characteristic corresponding to the ear of the listener on the side where the output acoustic signal is to be heard and the target acoustic characteristic, the target acoustic The stereophonic sound device according to claim 2 or 3, wherein a difference obtained by multiplying a third frequency weight component having a relatively large weight in a frequency band that should improve the follow-up characteristic is used.   The stereophonic apparatus according to claim 4, wherein a frequency band that should improve followability to the target acoustic characteristic is set so as to include a frequency component that is a clue to the sound image localization.   The stereophonic sound device according to claim 1 or 3, wherein the first frequency weighting component changes according to a position where the sound image is localized.   The stereophonic apparatus according to claim 2 or 3, wherein the frequency band in which the crosstalk is to be suppressed is determined to be a frequency band including a frequency component that is a clue to sound image localization.   4. The stereophonic device according to claim 2, wherein the second frequency weighting component changes according to a position where the sound image is localized. 5. As the correction characteristics of the correction means, a plurality of correction characteristics that differ depending on the position where the sound image is localized are preset,
The stereophonic device according to claim 6 or 8, wherein the correction unit selects a correction characteristic corresponding to a position where the sound image is localized.   10. The stereophonic device is mounted on a vehicle, and is used to cause an occupant of the vehicle to listen to sound for notification and warning from a desired direction. The three-dimensional acoustic apparatus according to item 1.

Images