WO2025110110A1 - Audio reproduction system - Google Patents

Abstract

An audio reproduction system according to the present invention comprises a signal processing unit that performs processing on a plurality of ch signals that include a left ch signal L and a right ch signal R, a left speaker and a right speaker, and a front speaker that is provided in a forward direction further from a listener than the left speaker and the right speaker at a prescribed opening angle or lower and a prescribed elevation angle or higher. The signal processing unit generates a front speaker signal SF, a left speaker signal SLt, and a right speaker signal SRt from the signal L and the signal R. The SF includes at least 200–9 kHz among components that have a high correlation between L and R, SLt is principally components of the left ch signal L that have a low correlation with the right ch signal R, and SRt is principally components of the signal R that have a low correlation with the signal L.

Description

Sound Reproduction System

This disclosure relates to an audio reproduction system.

　There is currently technology for acoustic correction of speakers.

For example, there is a technology for audio equipment that corrects the bias of the center sound image in the playback sound field and the left-right asymmetric spread of the sound field (see Patent Document 1). This technology calculates a transfer function based on the impulse response series of the left and right speakers, and configures the audio equipment to have a correction circuit made up of a transfer function obtained by an inverse matrix with the transfer function as an element.

Furthermore, there is technology relating to an audio processing device that allows the listener to hear sound that is always in the correct state even if the position of the listener's head relative to the speaker is not constant (see Patent Document 2), and technology relating to a headphone device that minimizes the deviation between the image position and the sound image position (see Patent Document 3).

There is also technology related to central signal extraction processing that generates highly correlated and low correlated signal components from two inputs, one each on the left and right (see Patent Documents 4 to 6).

There are also other technologies related to generating virtual sound sources (see Non-Patent Documents 1 and 2).

JP 2001-057699 A JP 2003-092799 A Japanese Patent Application Publication No. 08-009489 JP 2009-025500 A JP 2009-027388 A JP 2011-023961 A

Virtual Sound Source Generation Technology for Stereo Reproduction Yoshitaka Murayama and Haruo Hamada, JAS Journal 2011 Vol.51 No.1 (January issue) Sound field reproduction using a transaural system with two speakers, ASJ 2011

If it is difficult to achieve an ideal speaker placement inside the vehicle, for example, there may be problems with sound quality degraded by primary reflections due to obstructions near the speakers, or it may be difficult to position the speakers stably in front of the vehicle.

Furthermore, when using virtual sound source processing to improve localization, the transfer function of the playback system changes when the listener's head moves, resulting in problems with the sound image localization, sound field, and sound quality not being as intended.

Patent Document 1 and Patent Document 2 propose a method of detecting head movement with a sensor and reflecting it in virtual sound source processing, but there is no mention of how to deal with head rotation, which frequently occurs when listening to an in-car system, for example.

In addition, the headphone playback in Patent Document 3 uses a gyroscope or other method to accommodate head rotation, which may be applicable, but there are problems with accuracy and the need to handle huge amounts of data, which increases the cost of the system.

The objective of this disclosure is to provide an audio reproduction system that can provide a sound source environment that can obtain forward positioning and a sense of realism even when basic speaker placement is not possible, and can obtain stable forward positioning even when head rotation occurs when using virtual sound source processing.

The sound reproduction system of the present disclosure includes a signal processing unit that processes signals on multiple channels including at least a left channel signal L and a right channel signal R, a left speaker and a right speaker that are positioned to the left and right of the listener at a predetermined angle of opening or less and a predetermined angle of elevation or more, and a front speaker that is positioned further forward from the listener than the left and right speakers, and the signal processing unit generates a front speaker signal SF, a left speaker signal SLt, and a right speaker signal SRt from the left channel signal L and the right channel signal R, respectively, and the front speaker signal SF includes at least 200 to 9 kHz of the components that are highly correlated between L and R, and the left speaker signal SLt is mainly composed of components of the left channel signal L that are less correlated with the right channel signal R, and the right speaker signal SRt is mainly composed of components of the right channel signal R that are less correlated with the left channel signal L.

In addition, in the sound reproduction system, the left speaker and the right speaker may be arranged such that the opening angle is within a predetermined range of 110° or less and does not exceed the line of sight in front of the listener, and the elevation angle is within a predetermined range of 30° or more and does not exceed the apex on the listener's ear side.

The sound reproduction system may also be configured so that the front speakers are composed of a front left speaker and a front right speaker positioned in a plane parallel to the coronal plane to the left and right of the target listener and facing the direction of the listener.

In addition, in the sound reproduction system, the signal processing unit may reproduce signal components in at least a band with a lower frequency of 0 to 200 Hz and an upper frequency of 400 to 600 Hz through the front speakers, regardless of the left-right correlation between the left speaker signal SLt and the right speaker signal SRt.

In addition, in the sound reproduction system, the listener may be multiple listeners, and the left speaker and the right speaker may be placed on the left and right of each listener's head, respectively, or the left speaker may be placed at the left end of each listener lined up horizontally in the left-right direction, and the right speaker may be placed at the right end.

In addition, in the sound reproduction system, the opening angle of the left speaker and the right speaker may be set to a range of 70 to 110°, and the elevation angle may be set to a predetermined range of angles from 40° or more to not exceeding the apex on the ear side of the listener, and the signal processing unit may perform virtual sound source processing on the left speaker signal SLt and the right speaker signal SRt to reproduce the signals.

In addition, in the case of an audio reproduction system that is installed as an in-vehicle speaker system, the left speaker and the right speaker may be located at the upper part of the ears of the listener in each seat, the front speakers may be located in a position related to the dashboard or the lower part of the left and right doors, and the signal processing unit may be incorporated into the in-vehicle audio system.

The sound reproduction system disclosed herein has the advantage of providing a sound source environment that can obtain forward localization and a sense of realism even when basic speaker placement is not possible, and can provide a sound source environment that can obtain stable forward localization even when head rotation occurs when virtual sound source processing is used.

FIG. 1 is a diagram showing a schematic configuration of a sound reproduction system. FIG. 2A is a diagram showing an example of a speaker arrangement. FIG. 2B is a diagram showing an example of a speaker arrangement. FIG. 2C is a diagram showing an example of a speaker arrangement. FIG. 3 shows an example of a signal processing mode of the signal processing unit. FIG. 4 is an example of a graph illustrating the basis for frequencies for ensuring a sense of realism. FIG. 5A shows an example of the time response and frequency response characteristics of the HRTF. FIG. 5B is an example of the time response and frequency response characteristics of the HRTF. FIG. 5C shows an example of the time response and frequency response characteristics of the HRTF. FIG. 6 shows the frequency response characteristics of the HRTF in front of the sound source. FIG. 7 shows an example of a configuration having a front left speaker and a front right speaker. FIG. 8 is a diagram showing the overall flow of signal processing in the second embodiment. FIG. 9 shows an example in which two listeners are arranged side by side in the left-right direction. FIG. 10 shows an example in which an LSP and an RSP are installed for each listener. FIG. 11 illustrates an example of a signal generating unit according to the third embodiment. FIG. 12 illustrates an example of a signal generating unit according to the third embodiment. FIG. 13 is a diagram showing the overall flow of signal processing in the third embodiment. FIG. 14 shows an example of an arrangement in an experimental example with an opening angle of 90° and an elevation angle of 30°. FIG. 15 shows an example of an arrangement with an opening angle of 150° and an elevation angle of 0°. FIG. 16 is a diagram showing changes that occur when the head is rotated. FIG. 17 is a graph showing the change in HRTF_L at each elevation angle during head rotation. FIG. 18 is a graph showing the change in HRTF_L at each elevation angle during head rotation. FIG. 19 is a graph showing the sound pressure difference and phase difference for each opening angle and elevation angle. FIG. 20 is a graph showing the sound pressure difference and phase difference for each opening angle and elevation angle. FIG. 21 is a graph showing the sound pressure difference and phase difference for each opening angle and elevation angle. FIG. 22 shows an example of speaker arrangement when applied to a listening chair or the like. FIG. 23 shows an example of a center signal extraction process in the case of application to a listening chair or the like. FIG. 24 shows an example of a center signal extraction process in the case of application to a listening chair or the like. FIG. 25A shows an example of speaker arrangement when applied to an in-vehicle system. FIG. 25B shows an example of speaker arrangement when applied to an in-vehicle system. FIG. 26 shows an example of a center signal extraction process in the case of application to an in-vehicle system. FIG. 27 shows an example of a center signal extraction process in the case of application to an in-vehicle system. FIG. 28 shows an example of a signal generating section when the front speakers are capable of high quality sound reproduction over the entire frequency range. FIG. 29 shows an example of center signal extraction processing when the front speakers have high bass reproduction capability and the LSP and RSP speakers are installed suitable for mid- and high-range reproduction.

Below, an example of an embodiment of the disclosed technology will be described with reference to the drawings. Note that the same reference symbols are used for identical or equivalent components and parts in each drawing. Also, the dimensional ratios in the drawings have been exaggerated for the convenience of explanation and may differ from the actual ratios.

Explain the premise and overview of the embodiment of this disclosure.

Most common 2ch stereo sources are designed to be played back on a system with basic speaker placement (usually 30 degrees to the left and right of the front) to create a sound image in front of the listener.

Many in-car audio systems have at least one speaker on each side of the listener, located near the door feet (lower), on the roof liner (ceiling), or in the rear tray (rear). When speakers are located below the door, the sound image is localized forward, but because the diaphragm is not facing the listener, high frequencies are easily attenuated, and seats and other sound-insulating and reflective objects cause problems with attenuation and deterioration of sound quality, particularly in the high frequencies. When speakers are located on the roof liner, the distance between the listener and the speaker is shorter, so the signal reproduction quality is high and the speaker is less susceptible to the effects of reflected sound compared to other speaker placements, but there is an unnatural feeling due to the sound image being localized upwards. When speakers are located on the rear tray, there is both an unnatural feeling due to the sound image being at the rear and a large deterioration in sound quality due to the effects of reflected sound.

For the problem of sound image localization, the speaker-to-ear transfer function (HRTF: Head related transfer function) is used. Speaker-to-ear generally refers to the distance from the sound source to the entrance of the ear canal or above the eardrum. In the method using HRTF, a method is often applied in which the HRTF for a virtual intended speaker arrangement is input and the inverse transfer function of the transfer function of the playback system including crosstalk is processed (virtual sound source processing) and then played back to obtain a playback sound field with a virtual speaker arrangement.

As mentioned above, depending on the speaker arrangement, there are problems with sound quality and stability. Therefore, the method according to this embodiment provides an audio reproduction system that can obtain forward localization and a sense of realism even when basic speaker arrangement is not possible, and can obtain stable forward localization even if head rotation occurs when virtual sound source processing is used.

[First embodiment]
A first embodiment will be described. Fig. 1 is a diagram showing a schematic configuration of an audio reproduction system. As shown in Fig. 1, the audio reproduction system 1 includes a signal processing unit 10, a left speaker 12L and a right speaker 12R arranged on the left and right sides of the head of a listener (UL), and a front speaker 14 located in the forward direction farther away from the listener than the left speaker and the right speaker. The signal processing unit 10 processes signals of multiple channels including at least a left channel signal L and a right channel signal R. The signal processing unit 10 may be realized by a hardware configuration of a computer having a CPU, a RAM, a storage, etc. (not shown).

The left speaker 12L and the right speaker 12R are placed to the left and right of the listener at a predetermined angle of opening or less and a predetermined angle of elevation or more. In the following description, for ease of explanation, when the speakers are referred to by their abbreviations, the left speaker 12L will be referred to as (LSP), the right speaker 12R as (RSP), and the front speaker 14 as (FSP). In some embodiments described below, there may be multiple left speaker 12L, right speaker 12R, and front speaker 14.

<Combination of speaker arrangement and signal processing>
Here, the combination of speaker arrangement and signal processing of the method according to the present embodiment will be described. The signal processing unit 10 extracts signal band components (SF) of at least a frequency band of 200 Hz to 9 kHz (described later in A1) of a centrally localized sound signal (signal component with high internal correlation between L and R signals) from the L and R signals. The signal processing unit 10 also reproduces L signal components SLt and R signal components SRt of at least a band of 600 Hz or more (described later in A2) of a signal with low correlation. The L signal components SLt and R signal components SRt are reproduced from the LSP and RSP, respectively, arranged at an elevation angle of 30° or more and an opening angle of 110° or less (described later in A3) at a position closer to the listener side than the FSP. The signal band components SF are reproduced from the FSP. Note that the position closer to the listener side than the FSP may be a relatively close position. Each signal is corrected for time differences and level differences between at least the FSP and LSP (or RSP) caused by differences in speaker distance and efficiency.

Fig. 2A to Fig. 2C are diagrams showing an example of speaker placement. Fig. 2A is a diagram (2A) showing the listener (UL) from above, Fig. 2B is a diagram (2B) showing the listener (UL) from behind, and Fig. 2C is a diagram (2C) showing the listener (UL) from the left side. The speaker placement example is as follows: "Opening angle of LSP and RSP: 110° or less (example is 90°)" and "Elevation angle of LSP and RSP: 40° or more (example is 60°)."

FIG. 3 shows an example of the signal processing mode of the signal processing unit. (3A) is a diagram showing the overall signal processing flow of the signal processing unit 10 of the first embodiment. (3B) is an example of the signal generating unit 10A, and shows an example of the case where the 200 to 9 kHz band of the centrally localized sound is output from the FSP. (3C) is an example of a circuit mode including a central signal extracting unit 10C that performs central signal extraction processing. Note that the central signal extraction processing method can be applied with reference to the techniques of Patent Documents 4 to 6. For example, a signal processing method using an adaptive filter or the like may be used, and other configurations may also be used as long as they are capable of separating the signal of the centrally localized sound from other signals. An example of the method is described below.

As shown in (3A), the signal processing unit 10 receives a left channel signal L (Lch signal) from the LSP and a right channel signal R (Rch signal) from the RSP, and generates an L signal component SLt, an R signal component SRt, and a signal band component SF as signals in the signal generating unit 10A. Dly is a delay corresponding to the difference in distance between the LSP and RSP and the FSP. Part 10B performs gain control to correct the level difference due to the efficiency difference and distance difference between the LSP and RSP and the FSP, and performs power amplification to drive the speakers.

In the signal generating unit 10A (3B), signals in the frequency band between 200 Hz and 9 kHz are extracted from the L signal component SLt and the R signal component SRt via a bandstop filter (BSF: L1, R2) and a bandpass filter (BPF: L2, R1). Two bandpass filters are provided on each side. Dly is a delay corresponding to the central signal extracting unit 10C. In the central signal extracting unit 10C, low correlation signals SLt', SRt', and high correlation signal SF are extracted from the inputs of the L side input Sil and the R side input Sir. SLt' and SRt' are added to the delay signal in a calculator to output the signal component SLt and the R signal component SRt.

In the center signal extraction unit 10C (3C), the filter coefficients for extracting highly correlated signals are updated so as to minimize the error between the left and right signals, and the filter coefficients are convolved with the L-side input Sil and the R-side input Sir via an adaptive filter (ADF) to generate SLt' and SRt'. Other configurations may also be used as long as they can separate the sound signal localized in the center from other signals.

In this way, the signal processing unit 10 generates the front speaker signal SF, the left speaker signal SLt, and the right speaker signal SRt from the signal L and the signal R. The front speaker signal SF contains at least 200 to 9 kHz of the components with high correlation between L and R. The left speaker signal SLt is mainly composed of the components of the left channel signal L that are low in correlation with the right channel signal R, and the right speaker signal SRt is mainly composed of the components of the right channel signal R that are low in correlation with the left channel signal L.

Next, we will explain the principles (A1) to (A3) that characterize the method of this embodiment.

The frequency bands listed in (A1) are particularly effective for forward localization of centrally localized sounds. If sounds with frequencies below 9 kHz are reproduced from the FSP, it can be perceived as being sufficiently forward as the head rotates. Conversely, sounds above 9 kHz have short wavelengths and it is difficult to perceive direction based on the phase difference between the two ears. Also, the sensitivity of the ears may be low, and the contribution to horizontal localization is low, so it may be reproduced from the upper SP. In this case, high-frequency sounds that are strongly affected by obstructions are reproduced from the nearby LSP and RSP, making it possible to achieve high sound quality (in this case, there is no need to place a dedicated high-frequency speaker in front. Also, since the 7 to 8 kHz sound source that contributes to the sense of upward localization of centrally localized sounds is not reproduced from the LSP and RSP located above, a strong sense of upward localization can be avoided (the explanation of the 7 to 8 kHz that contributes to the sense of upward localization will be explained later in "Perception of the direction of the sound source").

The frequencies in the 600 Hz band and above mentioned in (A2) are the frequencies for ensuring a sense of realism. In order to ensure a sense of realism, it is necessary that the low correlation signals of the Lch and Rch can be heard in a state where there is a large interaural difference (difference in sound pressure between the left and right ears). In the example of a speaker placement with an elevation angle of 30° or more and an opening angle of 90°, the interaural difference (HRTF_R-L) for the output from the left is particularly clear above 600 Hz. Therefore, a sense of realism can be ensured by playing low correlation signals in at least the 600 Hz frequency band and above from speakers placed on the left and right. Note that low correlation signals contribute to the sense of realism as signals output from one speaker.

Figure 4 is an example of a graph that explains the basis for frequencies to ensure a sense of realism. The graph shows the interaural difference (the value obtained by subtracting the sound pressure HRTF-R at the right ear from the sound pressure HRTF-L at the left ear) for each elevation angle with an LSP with an opening angle of 90°. The greater the interaural difference that is reproduced when a signal with low correlation is played, the greater the sense of realism. The graph shows that the interaural difference is clear (approximately 3 dB or more) at 600 Hz and above for all elevation angles, which means that a sense of realism can be ensured by reproducing at least 600 Hz and above from the LSP and RSP.

Here, we will explain HRTF. HRTF, which serves as a clue for identifying the direction of a sound source through hearing, mainly includes the following elements: (1) Interaural difference: The difference in sound pressure level, time difference, and phase difference that occurs between the left and right ears contributes greatly to the perception of left and right direction. If there is no interaural difference, the sound source can be recognized as being on the median plane, but identifying the direction is difficult. (2) Sound pressure frequency characteristics: Characterized by the fact that they differ depending on the direction of the sound source due to the effects of reflection and resonance from the pinna. They change significantly not only left and right (horizontally) but also up and down, so they also affect the perception of up and down.

Figures 5A to 5C show examples of the time response and frequency response characteristics of HRTFs. Figure 5A shows an example (5A) where the sound source direction is 30° to the front and left (elevation angle 0°). As the HRTF from a sound source in this direction to the left and right ears (eardrums), the time response (5B) of Figure 5B and the frequency response characteristics (5C) of Figure 5C are obtained. Meanwhile, Figure 6 shows the frequency response characteristics of the HRTF for a sound source directly in front when the elevation angles are 0°, 20°, and 40°, respectively.

Here, we will explain the "perception of sound source direction". The sound pressure characteristics of HRTFs are different in all directions of the sound source, so they can be a clue to identifying the sound source direction in all directions. However, in past experiments on the perception of the direction of a stationary sound source, the correct answer rate was low for the sound source direction in the median plane and in the front-to-back direction, which suggests that the interaural difference is more dominant than the sound pressure characteristics of HRTFs. In addition, for a stationary sound source, it is usually difficult to distinguish whether the characteristics of the sound pressure characteristics that are generated are due to the HRTF or the characteristics of the sound source, but the change in the sound pressure characteristics caused by the movement of the sound source can be captured as a change in the transfer function, making it easier to identify. When a person tries to clearly identify the position of a sound source, rotating or tilting the head has the same effect as moving the sound source, and it is thought that the direction is clearly identified by sensing the "change" in the sound pressure characteristics and the "change" in the interaural difference. It has also been confirmed that when the level of a sound source played on a speaker is raised to 7-8 kHz, as in the case of HRTF changes due to elevation angle, the sound image appears to rise.

The arrangement given in (A3) is an arrangement that does not give the impression that the sound source is located at the LSP and RSP when the head rotates, i.e., an arrangement that causes little change in HRTF when the head rotates. The LSP and RSP, which mainly output signals with low LR correlation, are arranged to the left and right of the listener, respectively. Here, the arrangement is such that a strong sense of localization does not occur in the direction of the LSP and RSP, so that the sense of forward localization is not hindered when the head rotates. Based on the results of the "Listening Experiment" and "Confirmation of HRTF in Listening Experiment" described later, the elevation angle is set to 40° or more and the opening angle is set to 110° or less. The opening angle range is from 110° or less to an angle that does not exceed the line of sight when the listener is facing forward, and the elevation angle range is from 30° or more to an angle that does not exceed the vertex on the listener's ear side (or the vertex of the top of the head).

As described above, the sound reproduction system of the first embodiment can provide forward localization and a sense of realism even when basic speaker placement is not possible, and can provide a sound source environment in which stable forward localization can be obtained even when head rotation occurs when virtual sound source processing is used.

Moreover, in many music sources and TV and radio content, the main instruments and voices, such as the basic instruments, vocals, and narration, are created to be localized in the center. By playing these components with the front SPs, the overall sound is localized forward, even if the LSPs and RSPs are placed above. Furthermore, by not playing the 6 to 9 kHz centrally localized sound from the LSPs and RSPs placed above, the sound image of the centrally localized sound can be prevented from moving in the height direction of the LSPs and RSPs placed above. Also, when the head is turned, the forward localization sense provided by the FSPs takes precedence over the upward localization sense reproduced from the SPs above, so the forward localization sense can be maintained. Since the LSPs and RSPs are placed in close proximity to the listener, the effects of primary reflections can be reduced, resulting in higher sound quality. In addition, the driving power can be reduced.

Furthermore, by setting the elevation angle of the LSP and RSP to 30° or more, (1) it is possible to reduce the change in HRTF when the head rotates, (2) it is possible to reduce the change in the distance between the SP and the ear when the head rotates, and (3) it is possible to reduce the peaks and dips in the sound pressure characteristics. These effects make it possible to suppress the strong feeling of the real sound source position (SP position) due to the sound emitted from the LSP and RSP when the head rotates. Furthermore, when virtual sound source processing is used, it is possible to increase robustness against head rotation (making it less likely that localization in an unintended direction or a sense of echo will occur), and it is possible to maintain forward localization even if the head moves more than expected.

Furthermore, the opening angle of the left speaker 12L (LSP) and the right speaker 12R (RSP) may be in the range of 70 to 110°, and the elevation angle may be in a predetermined range of 40° or more but not exceeding the apex on the listener's ear side. Furthermore, the signal processing unit 10 may apply virtual sound source processing to the left speaker signal SLt and the right speaker signal SRt before reproducing them. This minimizes the change in the difference between the two ears when the head is rotated, enabling even more stable virtual sound source processing for reproduction. This makes it possible to suppress the phenomenon of unclear positioning and increased echo when the head is rotated.

[Second embodiment]
In the second embodiment, two front speakers are arranged. Note that the same components as those in the first embodiment are designated by the same reference numerals and the description thereof will be omitted.

In the second embodiment, the front speakers 14 have a front left speaker 14L (FLSP) and a front right speaker 14R (FRSP). FIG. 7 shows an example of a configuration having a front left speaker and a front right speaker. The front left speaker 14L and the front right speaker 14R are arranged in a plane parallel to the coronal plane to the left and right of the target listener (UL) and facing the direction of the listener (UL). By arranging the two front speakers in this way, the centrally located sound (phantom center) created in front of the listener is the FSP sound image SFr. The front left speaker 14L and the front right speaker 14R are an example of a left speaker and a left speaker of the present disclosure.

Here, we will explain the characteristics of the phantom center. When speakers are placed to the left and right of the listener (UL) and the same signal (monaural signal) is played back, the sound image is localized in front of the listener. Also, even if the listener moves to the left or right between the left and right speakers, the sound image is generally localized in front of the listener after the movement.

FIG. 8 is a diagram showing the overall flow of signal processing in the second embodiment. As shown in FIG. 8, the signal band component SF output from the signal generating unit 10A is branched and reproduced by FLSP and FRSP.

[Modification of the second embodiment]
FIG. 9 shows an example in which two listeners are arranged side by side. In the case of multiple listeners at different positions in the left and right direction, the FLSP is arranged further leftward and forward of the leftmost listener (UL1), and the FRSP is arranged further rightward and forward of the rightmost listener (UL2) in a plane parallel to the coronal plane of each listener. Each of the dotted front speakers is a fixed center position for the listeners (UL1) and (UL2). In this way, a left speaker (front left speaker 14L (FLSP)) is arranged at the left end of each listener arranged horizontally in the left and right direction, and a right speaker (front right speaker 14R (FRSP)) is arranged at the right end.

Figure 10 shows an example in which an LSP and an RSP are installed for each listener. In this way, the left speaker 12L and the right speaker 12R can be placed on the left and right of each listener's head, respectively.

According to the second embodiment, even if it is not possible to install an FSP in front of the listener, it is possible to localize a sound image of a centrally localized sound in front of the listener. For multiple listeners at different horizontal positions, a sound image of a centrally localized sound can be localized in front of each of them. In addition, it is possible to provide a similar playback space to each listener at a different horizontal position.

[Third embodiment]
In the third embodiment, the signal processor 10 reproduces signal components (midlow components) between the lower limit frequency of 0 to 200 Hz and the upper limit frequency of 400 to 600 Hz for the signals in the first and second embodiments from the front speakers 14 even if there is no correlation between the left and right signals. As a result, all midlow components of the left and right signals are reproduced in FSP (or FLSP and FRSP).

Most musical instruments include a frequency band between 200 and 400 Hz, so even instruments recorded on only one channel can be perceived as coming from the front. Also, if low-correlation signals above 600 Hz are played from the front speakers, the playback sound from the nearby LSP and RSP is reduced, which increases the influence of primary reflections and can easily lead to a deterioration in sound quality and a loss of realism. A good method is to separate the mid-low components including 200 to 400 Hz from the L and R signals in advance, mix this with a highly correlated signal extracted from the other components, and send it to the FSP (or FLSP and FRSP).

FIGS. 11 and 12 are examples of the signal generation unit 10A of the third embodiment. The example of FIG. 11 is an example of a case where all midlow (200 to 400 Hz example) signal components are reproduced by the front speakers. Bandpass filters (BPF: L1, L3, R1, R3) and bandstop filters (BSF: L2, R2) are provided on the left and right. This can be applied to both the case where there is only one FSP as in the first embodiment (FIG. 3), and the case where there are left and right (FLSP and FRSP) as in the second embodiment and the left and right signals are the same (FIG. 7).

The example in Figure 12 shows a case where all midlow (200-400 Hz example) signal components are played on the left and right front speakers. In this case, the front speakers are on the left and right (FLSP and FRSP), and the left and right midlow components are played on the respective front speakers. The highly correlated signal SF' is calculated at 0.5 and output as SFl and SFr to FLSP and FRSP.

FIG. 13 is a diagram showing the overall signal processing flow of the third embodiment. When playing from the front left and right speakers, as shown in FIG. 13, the SLt and SRt signals are Dlyed and played from the LSP and RSP, and the SFl and SFr signals are played from the FLSP and FRSP.

As described above, the signal processing unit 10 of the third embodiment reproduces signal components in at least the lower limit frequency band of 0 to 200 Hz and the upper limit frequency band of 400 to 600 Hz from the front speakers, regardless of the left/right correlation of the left speaker signal SLt and the right speaker signal SRt. According to the third embodiment, by reproducing the 200 to 400 Hz band contained in most musical instruments and human voices from the front speakers, it is possible to obtain not only the reproduced sound from the LSP and RSP but also a sense of front localization, even in the case of a source that contains almost no centrally localized sound. In addition, by setting the upper limit frequency to 600 Hz or less, the sense of realism can be maintained.

[Fourth embodiment]
In the fourth embodiment, unlike the first to third embodiments, in the sound reproduction system 1, the opening angle of the LSP and RSP is set to 70 to 110°, the elevation angle is 40° or more, and virtual sound source processing is performed on the L signal component SLt and the R signal component SRt.

When playing back signals that have undergone virtual sound source processing, it is desirable that there be little change in the HRTF and in the interaural difference when the head rotates. To achieve this, it is better to have little change in the distance between the left and right speakers (LSP/RSP) and the eardrum, and the greater the elevation angle of the LSP and RSP, the less change there is. Also, by setting the opening angle to around 90° (coronal plane), the change in the distance between the left and right speakers and the eardrum when the head rotates will be the same, so the change in the interaural difference can also be reduced.

A supplementary note on virtual playback (surround, etc.) of virtual sound source processing: Virtual sound source processing aims to reproduce the intended sound field and positioning by reproducing the sound pressure above the left and right eardrums. Processing of the sound source signal (Sound Source) includes setting a virtual sound source at a specific position and convolving the HRTF in that direction, and processing to cancel crosstalk components (HRTF is also used here) to control the sound pressure at the left and right ears when playing on a 2-channel speaker, for example. Therefore, it is assumed that the HRTF assumed when processing the signal is equal to the HRTF during playback. Possible factors that cause HRTF differences include individual differences in HRTF, the position of the head during playback, and changes in the position of the ears due to rotation, and dealing with this is important in building a practical system.

　Generally, with playback methods that use HRTF, (1) virtual playback is effective for sources with movement, such as movies. (2) For sources with no movement, localization, especially in the forward and backward directions, is often unclear. (3) If the listener's head moves during playback, localization can become unclear or can be localized in an unintended direction. There can also be a strong echo effect that sounds unnatural.

[Experimental Example]
Here, an example of a listening experiment and an HRTF analysis will be described with respect to the first embodiment. The listening experiment was a listening experiment of the forward localization feeling (with FSP) when the head was rotated by the arrangement of the LSP and RSP. The elevation angle is the angle between the line connecting the center surface part of the diaphragm of the LSP (or RSP) and the center of the head and the horizontal plane. The opening angle is the angle between the center surface part of the diaphragm of the LSP (or RSP) and the front direction of the listener's reference position. The speaker was oriented in the radial direction, and the LSP was arranged so that the diaphragm central axis faces the entrance direction of the ear canal of the left ear, and the RSP was arranged so that the diaphragm central axis faces the entrance direction of the ear canal of the right ear. The distance between the FSP and the center of the head was 1.4 m. The distance between the LSP and the RSP and the center of the head was 0.4 m.

Fig. 14 shows an example of an arrangement in an experimental example with an opening angle of 90° and an elevation angle of 30°. Fig. 15 shows an example of an arrangement with an opening angle of 150° and an elevation angle of 0°. Table 1 is a comparison table of opening angles and elevation angles. The symbols in the comparison table are ◯, △, and × to indicate the state of the sound source as follows: ◯: Stable forward localization, △: For some sources, the presence of the sound source is felt in the LSP and RSP directions when the head is rotated (somewhat unstable forward localization), ×: The presence of the sound source is clearly felt in the LSP and RSP directions when the head is rotated (forward localization is hindered).

(HRTF data confirmation under listening conditions)
Fig. 16 is a diagram showing the change when the head rotates. The change in HRTF_L when the head rotates was compared for each elevation angle for the change in HRTF-L when the head is rotated ±20° with an opening angle of 90° to the left (HRTF_L at directional angles of 70, 90, and 110° to the left). Figs. 17 and 18 are graphs showing the change in HRTF_L at each elevation angle when the head is rotated. It can be seen that with a speaker arrangement at an opening angle of 90°, the larger the elevation angle, the smaller the change in HRTF when the head is rotated. It can be seen that the change is particularly large at 0° and 20°.

The situation where there is no (or little) change in sound pressure in both ears when the head is rotated is similar to when the sound source is located above or when headphones are being used. The sound reproduced from the LSP and RSP in an arrangement that results in such a change in characteristics will be localized above or inside the head. However, this effect is weaker than the effect of perceiving a front sound source (sound reproduced from the FSP) when the head is rotated, and as long as the content (sound source signal) contains front-localized sound, the sense of front localization will be maintained.

Next, we compare the changes in interaural difference (HRTF_R - HRTF_L) when the head is rotated, depending on the speaker placement.

Figures 19 to 21 are graphs showing the sound pressure difference and phase difference for each opening angle and elevation angle. (19A) in Figure 19 shows the sound pressure difference and phase difference for a typical speaker arrangement with an opening angle of 30° and an elevation angle of 0°. In this case, there is a large change in both sound pressure and phase due to the large effect of head rotation and the large change in distance between the speaker and the ear. (19B) shows the case with an opening angle of 90° (in the coronal plane) and an elevation angle of 0°. In this case, there is little change in distance, so there is little change in phase, but there is a large change in sound pressure characteristics due to the effect of the pinna. (20A) in Figure 20 shows the case with an opening angle of 90° and an elevation angle of 40°, and (20B) shows the case with an opening angle of 77° and an elevation angle of 40°. Figure 21 shows the case with an opening angle of 103° and an elevation angle of 40°.

When the elevation angle is 0, the change in interaural difference when the head rotates is large, which can lead to the generation of unwanted echoes and the inability to reproduce sound image localization, especially when virtual sound source processing is used. When the elevation angle is 40° or more, and the opening angle is in the range of 77° to 103°, the change in interaural difference when the head rotates is relatively small in both sound pressure level and phase.

(Application examples of the embodiment)
Next, application examples of the above-mentioned embodiments will be illustrated. First, an application example to a listening chair or the like will be given. FIG. 22 shows an example of speaker arrangement when applied to a listening chair or the like. The FSP is installed in the front direction with a fixture connected to the armrest, and the signal processing unit 10 may be mounted on a PC software, a game machine, an audio amplifier, a playback player, or the like. The fixture may be movable for seating, or may be replaced with an FRSP and an FLSP. The LSP and RSP are located above the left and right ears, respectively (elevation angle 70 to 80°, opening angle 90°).

Figures 23 and 24 are an example of central signal extraction processing when applied to a listening chair, etc. Note that the reference numerals are the same as in the above example and are therefore omitted. When using the same components for all speakers, it is advantageous in terms of bass feeling and input resistance to have the LSP and RSP, which are placed close together, take charge of the low frequency range, so the low frequencies are reproduced by the LSP and RSP side. Also, since the FSP is small in diameter and has high high frequency reproduction capabilities, all highly correlated signals above 200 Hz are reproduced by the FSP. Also, when playing back virtual sound sources as in Figure 24, stable forward positioning can be maintained by performing this on SLt and SRt.

Also, in the case of multi-channel sources such as 5.1ch, the effect can be obtained by applying this processing to at least the front Lch and Rch. Note that it is also possible to deal with the other rear Lch and Rch by performing virtual playback with LSP and RSP. In this case, no effect can be obtained for this channel, but you can enjoy stable positioning and high sound quality for the entire multi-channel playback.

Secondly, an example of application to an in-vehicle system will be given. Figures 25A and 25B are examples of speaker placement when applied to an in-vehicle system. Figures 26 and 27 are an example of center signal extraction processing when applied to an in-vehicle system. In the in-vehicle system of Figures 25A and 25B, the LSP and RSP are placed overhead (roof lining) near the ears of the passenger and driver's seat listeners, respectively. In addition, the FSP is placed on the left and right sides of the dashboard. If placement on the dashboard is not possible, they may be placed under the left and right doors or in a location related to the door bottoms. The signal processing unit 10 is incorporated into the audio function built into the navigation system. Figure 26 shows the center signal extraction processing in the case of (25A). In addition, since the LSP and RSP are ones with sufficient bass and treble reproduction capabilities (for example, Patent Document 6), only a minimum signal (200 to 9 kHz, a signal with high correlation) is reproduced from the front SP.

(25B) in Figure 25 is an example where speakers are placed at the rear. Figure 27 shows the central signal extraction process for (25B). In this way, a circuit is provided that inputs the rear L channel and rear R channel separately from the front, and performs rear positioning. In the case of a multi-channel source (e.g. 5ch), rear speakers are installed to obtain stable rear positioning for the rear Lch and rear Rch as well. By combining the rear LSP (14Lb) and rear RSP (14Rb) and performing the same processing as for the main channels (front Lch, front Rch), the same effect can be obtained for the rear channel positioning as for the front channels.

As described above, when the sound reproduction system 1 is installed as an in-vehicle speaker system, the left speaker 12L and the right speaker 12R are placed at the top near the ears of the listeners in each seat. The front speakers 14 are placed in positions related to the dashboard or the bottom of the left and right doors. The signal processing unit 10 is incorporated into the in-vehicle audio system.

Figure 28 shows an example of a signal generation unit when the front speakers are capable of high quality sound reproduction across the entire frequency range. The signals sent to the front speakers (FSP or FLSP and FRSP) may be signals with high L/R correlation, including at least 200 to 9 kHz. Other highly correlated frequency components (components below 200 Hz and above 9 kHz) may be reproduced by the front speakers or by the LSP and RSP. Components below 200 Hz may be distributed between the front SP, LSP, and RSP. Select the speaker most suitable for the reproduction capabilities and installation conditions of each speaker. To avoid primary reflections, it may be possible to reproduce from the LSP and RSP, which are closer to the front speakers.

Figure 29 shows an example of center signal extraction processing when the front speakers have high bass reproduction capabilities and the LSP and RSP are speakers suitable for mid- and high-range reproduction. In this case, for example, all frequencies below 500 Hz are reproduced by the front speakers.

In addition, in the case where an LSP and an RSP are placed for each of multiple listeners, if the listeners are close to each other and the sound reproduced from an SP close to one listener affects the other listener, processing can be applied to cancel this.

The present invention is not limited to the above-described embodiment, and various modifications and applications are possible without departing from the spirit of the invention.

The disclosure of Japanese Patent Application No. 2023-197575, filed on November 21, 2023, is incorporated herein by reference in its entirety.

1 Sound reproduction system 10 Signal processing unit 12L Left speaker 12R Right speaker 14 Front speaker

Claims (7)

a signal processing unit that processes signals of multiple channels including at least a left channel signal L and a right channel signal R;
a left speaker and a right speaker disposed on the left and right of a listener at a predetermined angle or less and a predetermined angle or more;
a front speaker located in a forward direction farther from a listener than the left speaker and the right speaker,
the signal processing unit generates a front speaker signal SF, a left speaker signal SLt, and a right speaker signal SRt from the left channel signal L and the right channel signal R, respectively;
The front speaker signal SF includes at least 200 to 9 kHz among components with high correlation between L and R,
The left speaker signal SLt mainly includes a component of the left channel signal L that has a low correlation with the right channel signal R,
The right speaker signal SRt mainly includes a component of the right channel signal R that has a low correlation with the left channel signal L.
Sound reproduction system. The sound reproduction system of claim 1, in which the left speaker and the right speaker are arranged such that the opening angle is within a predetermined range of 110° or less and does not exceed the line of sight in front of the listener, and the elevation angle is within a predetermined range of 30° or more and does not exceed the apex on the ear side of the listener. The sound reproduction system of claim 1, wherein the front speakers are composed of a front left speaker and a front right speaker positioned in a plane parallel to the coronal plane to the left and right of the target listener and facing the direction of the listener. The sound reproduction system of claim 1, wherein the signal processing unit reproduces signal components in at least a band with a lower frequency of 0 to 200 Hz and an upper frequency of 400 to 600 Hz through the front speakers, regardless of the left-right correlation between the left speaker signal SLt and the right speaker signal SRt. The listener is intended for a plurality of listeners,
2. The sound reproduction system according to claim 1, wherein the left speaker and the right speaker are arranged on the left and right sides of each listener's head, respectively, or the left speaker is arranged at the left end of each listener arranged horizontally in the left-right direction, and the right speaker is arranged at the right end. For the left speaker and the right speaker, the opening angle is set to a range of 70 to 110 degrees, and the elevation angle is set to a predetermined range of angles from 40 degrees or more to not exceeding the apex on the ear side of the listener,
2. The sound reproducing system according to claim 1, wherein the signal processing unit performs virtual sound source processing on the left speaker signal SLt and the right speaker signal SRt and reproduces the signals. When installing it as an in-car speaker system,
The left speaker and the right speaker are disposed above the ears of the listener in each seat;
The front speakers are disposed in positions related to the dashboard or the lower portions of the left and right doors;
The sound reproducing system according to claim 1 , wherein the signal processing unit is incorporated in an in-car audio system.