JP2008311718A - Sound image localization controller, and sound image localization control program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To eliminate the unclearness of words and the addition of excessive reverberant sound and to express surround sound naturally and richly when arranging 2 channel speakers more closely to a listener than a normal position. <P>SOLUTION: When the speakers 42FL and 42FR are arranged more closely to the listener 41 than the normal position, an input signal analysis means 1 detects the input level of the audio signals of multiple channels at every prescribed interval of time, and a parameter setting means 2 decides whether or not the input level of this time is higher than the input level of the previous time by a prescribed value or more, sets the gain and/or delay time of rear reverberant sound to a value corresponding to the input level of this time when it is not higher, sets the gain and/or delay time of the rear reverberant sound to a value corresponding to the level in which the input level of the previous time and the prescribed value are added when it is higher, and controls a rear reverberant sound addition means 8. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a sound image localization control device and a sound image localization control program that generate a direct sound, an initial reflection sound, and a rear reverberation sound according to a sound field mode.

  Conventionally, in an AV (Audio / Video) amplifier or the like, a sound field as if listening to a concert hall, for example, is realized by digitally processing a reproduced audio signal. Various modes such as a large hall mode, a small hall mode, and an outdoor mode are known as sound field modes. In order to realize such a sound field, for example, a reverberation component is added by reverberation in the old days, a reflected sound situation of a hole is calculated in advance, and a reflected sound for each direction is a coefficient of an FIR (Finite Impulse Response) filter. As described above, a sound field feeling is produced by filtering the sound of each channel.

  Actually, as shown in FIG. 10, the initial reflected sound S1 is added at a position delayed by a predetermined period with respect to the original sound (direct sound S0), and the rear reverberant sound S2 is further added after a predetermined period. The delay time of the rear reverberation sound S2 with respect to the original sound is referred to as pre-delay, and it is possible to perform reverberation time, addition of sub-reverberation sound, fine level adjustment, etc., so that a wide range of sounds can be created. In addition, a so-called sound image localization technique has been established as a method for making it sound as if sound is being emitted from a place without a speaker. For example, in the following Patent Document 1, when there is no rear speaker, a sound image localization unit having a convolver in which a filter coefficient based on a head-related transfer function is set for each channel of a pair of left and right rear surround signals is provided. The pair of left and right rear surround signals via the means is added to the pair of left and right front stereo signals, and the sound images are localized at rear positions that are substantially symmetrical to the listener.

On the other hand, there is a listening method called near-field listening (for example, Non-Patent Document 1 below). This is a listening method in which the distance between the viewer and the two-channel speaker is extremely shortened, and the speaker is placed close to the viewer so that playback and viewing can be performed at a low volume. The arrangement of multi-channel speakers is defined in ITU-RBS 775-1, but there is no clear definition for near-field listening. For example, in Non-Patent Document 1 below, as an indication of the distance for near-field listening, The distance that the palm of the hand can reach the left and right speakers. Since two-channel speakers are installed close to each other, it is possible to reduce the influence of reflection in the viewing room, even when listening to the recorded sound itself or when viewing with sound image localization processing. It is an effective viewing method.
JP-A-8-265899 (abstract) stereo “Unbeatable low volume playback” (Music Notosha), February 2005 issue

  However, when performing sound image localization control by near-field listening, it is difficult to produce a sense of distance even if sound image localization processing is performed due to the proximity of a 2-channel speaker, and it is necessary to add a reverberation effect to produce a sense of distance. . When the reverberation effect is added, the early reflection sound S1 and the rear reverberation sound S2 described above are used. In sound image localization using multi-channel sound, in addition to the reverberation effect, two speakers are used. Since a lot of information for reproducing the surround sound is collected, there are problems such as deterioration of sound quality and unclearness of lines due to excessive reverberation.

  In view of the above-described problems of the prior art, the present invention has an unclearness or excessiveness of speech when a multi-channel signal is localized and reproduced by arranging a 2-channel speaker closer to the listener than the normal position. An object of the present invention is to provide a sound image localization control device and a sound image localization control program capable of eliminating surround sound and adding natural and rich surround sound.

In order to achieve the above object, the present invention provides a sound localization control apparatus for reproducing multi-channel audio signals by arranging left and right speakers for stereo reproduction at positions closer to the viewer than normal reproduction positions. Because
Input level detecting means for detecting the input level of the multi-channel audio signal every predetermined time;
It is determined whether or not the current input level detected by the input level detection means is greater than a predetermined value by a predetermined value or more. If not, the gain and / or delay time of the rear reverberant sound is determined as the current input level. A rear reverberation control means for setting the gain and / or delay time of the rear reverberation sound to a value according to a level obtained by adding the previous input level and a predetermined value when large,
Have.

In order to achieve the above object, the present invention provides a sound localization for reproducing multi-channel audio signals by arranging left and right speakers for stereo reproduction at positions closer to the viewer than normal reproduction positions. A control program,
An input level detecting step for detecting an input level of the multi-channel audio signal at predetermined time intervals;
It is determined whether or not the current input level detected by the input level detection means is greater than a predetermined value by a predetermined value or more. If not, the gain and / or delay time of the rear reverberant sound is determined as the current input level. A rear reverberation control step for setting the gain and / or delay time of the rear reverberation sound to a value corresponding to a level obtained by adding the previous input level and a predetermined value when the value is large,
Let the computer run.

  According to the present invention, when the front two-channel speaker is arranged at a position closer to the viewer than the normal stereo reproduction position and the multi-channel signal is localized and reproduced, the current input level is the previous input level. It is determined whether or not it is greater than a predetermined value above the level, and if not larger, the rear reverberation gain and / or delay time is set to a value corresponding to the current input level, and if larger, the rear reverberation gain is set. And / or delay time is set to a value corresponding to the level obtained by adding the predetermined value to the previous input level, so that the ambiguity of the lines and the addition of excessive reverberation are eliminated, and the surround sound is expressed naturally and richly. be able to.

  Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a first embodiment of a sound image localization control apparatus according to the present invention.

  As an example, FIG. 1 shows that 5.1 channel surround audio signals, ie, input signals FLch, Cch, FRch, SLch, SRch, and LFEch are downmixed to 2.1 channels of FLch, FRch, and LFEch (LFEch is through). The output device is shown. In the 5.1 channel surround speaker system, speakers 42FL, 42C, 42FR, 42SL, 42SR, and 42LFE of each channel are arranged with respect to the viewer 41 and the display 43 at the position shown in FIG. The Further, as shown in FIG. 2B, the speaker system of the normal position of the 2.1 channel uses only the front speakers 42FL, 42FR, 42LFE (hereinafter, “normal position”). On the other hand, in the near-field listening 2.1 channel speaker system according to the present invention, as shown in FIG. 3, the speakers 42FL, 42FR, and 42LFE are arranged away from the display 43 and close to the viewer 41 (hereinafter referred to as the viewer 41). , Near position). The arrangement position of the LFEch speaker 42LFE (subwoofer) is arbitrary, but is arranged close to the viewer 41 in FIG.

  In FIG. 1, among the 5.1 channel audio signals reproduced from a DVD player (not shown) and the like and input via the input terminal 12, the 5 channel audio signals excluding LFEch are input to the input analysis means 1. Via, it is applied to the transfer characteristic adding means 5. The LFEch audio signal is applied to the LFEch speaker 42LFE constituting the speaker system 13 through the amplitude amplifying unit 11 without passing through the transfer characteristic adding unit 5, the adding unit 9, and the crosstalk canceling unit 11.

  The transfer characteristic adding means 5 includes direct sound processing means 6, initial reflected sound processing means 7, and rear reverberation sound processing means 8 for processing the direct sound S0, initial reflected sound S1, and rear reverberation sound S2 shown in FIG. The parameter information holding means 3 generates a sound field corresponding to various sound field modes from the direct sound S0, the initial reflection sound S1, and the rear reverberation sound S2 shown in FIG. 10 for the five-channel audio signal excluding LFEch. Parameters (gain, delay time) are stored. When a certain sound field mode is set by the speaker position / sound field mode setting unit 4, a parameter corresponding to the set sound field mode is read from the parameter information holding unit 3 and directly sounded by the parameter setting unit 2. The processing unit 6, the initial reflection sound processing unit 7, and the reverberation sound processing unit 8 are set.

  The 5-channel audio signals processed by the direct sound processing means 6, the early reflection sound processing means 7, and the reverberation sound processing means 8 are downmixed by the adding means 9 into 2 channels of FLch and FRch. Next, after the crosstalk is canceled by the crosstalk cancellation processing means 10, the two-channel audio signals are applied to the FLch and FRch speakers 42 FL and 42 FR constituting the speaker system 13 via the amplitude amplification means 11. The

  In the above configuration, when the “normal position” shown in FIG. 2B is set by the speaker position / sound field mode setting unit 4, the sound field mode is set according to the sound field mode set by the speaker position / sound field mode setting unit 4. As parameters, the gain G0 and delay time D0 of the direct sound S0 such that the virtual sound source of each channel can be heard from that direction when reproduced at the “normal position”, the gain G1 and delay time D1 of the initial reflected sound S1, The gain G2 and delay time D3 of the rear reverberant sound S2 are read from the parameter information holding means 3, and set by the parameter setting means 2 to the direct sound processing means 6, the initial reflected sound processing means 7, and the reverberation sound processing means 8, respectively. The On the other hand, when the “near position” shown in FIG. 3 is set by the speaker position / sound field mode setting means 4, the direct sound processing means 6 and the initial reflected sound processing means 7 are reproduced at the “near position”. In addition, the gain G0 and delay time D0 of the direct sound S0, and the gain G1 and delay time D1 of the initial reflected sound, such that the virtual sound source of each channel can be heard from that direction, are read from the parameter information holding means 3 and parameter setting means 2 However, a gain G2 ′ and a delay time D2 ′ are set in the reverberation sound processing means 8 as follows.

  When the “near position” is set, the audio signal of 5 channels is input to the input analysis means 1 and the input level L is analyzed for each channel. The result output from the input analysis unit 1 is passed to the parameter setting unit 2. FIG. 4 is a flowchart showing the flow from input signal analysis to parameter setting. First, when an audio signal is input, the maximum level Lmax (f) of all channel input signals in the current frame f is detected with a certain number of samples as one frame (step 101).

  In the present invention, when the “near position” is set, the parameters (delay time D2 ′, gain G2 ′) of the rear reverberation sound S2 are set based on the maximum level Lmax of all channel input signals. Here, if the difference Δ between the maximum levels Lmax is large in consecutive frames, the rear reverberation sound S2 is suddenly changed, which causes unnaturalness. Therefore, the maximum level difference from the previous frame is compared. In step 102, the difference Δ (= Lmax (f) −Lmax (f−1)) between the maximum level Lmax (f) of the current frame f and the maximum level Lmax (f−1) of the previous frame f−1 is a reference value. Compare with TH. The reference value TH here is determined by the speaker position (normal position / near position) set by the speaker position setting / sound field mode setting means 4 and the sound field mode (large hall, small hall, etc.). "Is set to a value that does not cause a sudden change in the rear reverberation sound S2.

When the maximum level difference Δ from the previous frame is smaller than the reference value TH in step 102, the maximum level Lmax (f) is acquired, and the parameter D2 ′ of the rear reverberation sound S2 optimum for the maximum level Lmax (f) is obtained. , G2 ′ is set (step 103). On the other hand, if the maximum level difference Δ from the previous frame is larger than the reference value TH in step 102, that is,
Lmax (f) -Lmax (f-1)> TH
If it is, smoothing processing is performed between the maximum level Lmax (f) of the current frame and the maximum level Lmax (f-1) of the previous frame (step 104). In the smoothing process, for example, Lmax (f) ′ − Lmax (f−1) = TH
Then, parameters D2 ′ and G2 ′ of the rear reverberation sound S2 optimum for the maximum level Lmax (f) ′ after the smoothing process are set (step 103). Accordingly, when the maximum level difference Δ from the previous frame is larger than the reference value TH at the “near position”, the gain G2 ′ of the rear reverberation sound S2 can be reduced by Lmax (f) ′ (<Lmax (f)). The delay time D2 ′ is shortened. Then, the parameters D2 ′ and G2 ′ of the rear reverberation sound S2 are set for each frame (step 105).

  Further, when the “near position” is set, for example, when the scene is generally quiet, that is, when the maximum level Lmax is low, the rear reverberation sound S2 is prevented from being added in order to prevent the rear reverberation sound S2 from being added. By reducing the gain G2 ′ of the sound S2 and setting the delay time D2 ′ to be short, a sound field feeling corresponding to the scene can be obtained.

  Here, the transfer characteristics T for reproducing the sound field require a total of ten transfer characteristics from each of the five channels to the viewer's left and right ears, and each of them has a direct sound SO, It is divided into an initial reflected sound S1 and a rear reverberant sound S2. Here, as shown in FIG. 5, the rear reverberation sound S2 includes delay devices (delay times τ1, τ2, τ3, τ4, τ5, τ6) and amplifiers (gains G21, G22, G23, G24, G25, G26, G27). , G28), the reverberation processing means 8 is used in which the comb filters 21, 22, 23, 24 are configured in four stages and the all-pass filters 25, 26 are configured in two stages. In the parameter setting means 2, as parameters for generating the rear reverberation sound S2 based on the level analyzed by the input signal analysis means 1, for example, delay parameters (τ1 to τ6) and gain parameters (G21 to G28) in FIG. Are set in the reverberation processing means 8. In the sound field mode setting means 4, the gain G 0 and delay time D 0 of the direct sound S 0, the gain G 1 and delay time D 1 of the initial reflected sound, and the rear reverberation from the parameter information holding means 3 according to the selected sound field mode. The gain G2 and delay time D2 of the sound S2 are read and transferred to the transfer characteristic adding means 5. In addition, although FIG. 5 shows a configuration of four comb filters and two allpass filters, settings such as two comb filters may be performed in accordance with the sound field mode.

  Here, the transfer characteristic adding means 5 is divided into a direct sound processing means 6, an early reflection sound processing means 7 and a reverberation sound adding means 8. The direct sound processing means 6 reflects anywhere from the sound source in the transfer characteristics T. Instead, a process for a component called direct sound S0 that reaches the viewer's ear is performed. Here, the component of the direct sound S0 attaches importance to sound quality, so that the component measured in a room with little reflection such as an anechoic room is suitable, and a plurality of data are stored in the direct sound processing means 6 in advance. . Data used for processing is selected from the parameter information holding means 3 in accordance with the selected sound field mode, and the gain G0 and the delay value D0 are determined.

  The initial reflected sound processing means 7 is an initial reflected sound called initial reflected sound S1, which is the first reflected sound that reaches the viewer's ear after being reflected from the sound source to the wall or ceiling excluding the floor, and is 100ms from the direct sound S0. The process is performed for components that reach within. Since the initial reflected sound S1 is different from the direct sound S0 and uses reflection, a sound obtained by removing the direct sound component S0 from the transfer characteristic measured in a room with some reflection is used. Similar to the direct sound processing means 6, a plurality of data are held in advance in the initial reflected sound processing means 7, and the data used for processing is stored by the parameter information holding means 3 in accordance with the selected sound field mode. The gain G1 and the delay value D1 are determined.

  Since the 5-channel audio signal to which the spatial information for reproducing the set sound field is added by the transfer characteristic adding means 5 performs sound image localization by 2-channel reproduction, signals to the viewer's left and right ears Are added by the adder 9, and then the crosstalk cancellation processing is performed by the crosstalk cancellation processing means 10. The crosstalk canceling process is a technique for localizing a sound image as if there is a sound source at a place different from the actual sound source by a 2-channel speaker. When it is desired to localize the sound source X at a desired position, the head-related transfer function (HRTF) measured in advance with a real head or a dummy head is used as a coefficient of the FIR digital filter, and the signal of the sound source X is subjected to convolution calculation processing. When listening to the playback sound through the speaker, make sure that the playback sound component that reaches the listener's left ear from the right channel speaker and the playback sound component that reaches the listener's right ear cannot be ignored from the left channel speaker. Exists. For this reason, in order to properly obtain a three-dimensional sound field by two-speaker reproduction, the crosstalk component is removed from the acoustic signals of the left and right channels, so that there is no crosstalk for the listener. It is necessary to make the audio sound heard from the left and right speakers. This is the crosstalk cancellation process.

FIG. 6 is an example of this and is a principle diagram of the most basic crosstalk cancellation processing. In the figure, h _LL (t), h _LR (t), h _RL (t), and h _RR (t) are two speakers (l (t) and r (t) in the figure) and left and right ears (figure P _l (t), p _r (t))
d (t) = 1 / ((h _LL (t) × h _RR (t) −h _RL (t) × h _LR (t))
It is. As a result, the Lch signal is reproduced for the left ear and the Rch signal is reproduced for the right ear at the position of the ear (referred to as a control point) where the relationship in FIG. In the near field system, since two speakers are installed close to the viewer, the head-related transfer function measured at a position corresponding to the reproduction system is added to the head-related transfer function consisting of the relationship between the two speakers and the left and right ears. Is used. Here, p _l (t) and p _r (t) have already been added with transfer characteristics, and a signal added to the two channels is input. As a result, 5.1 channel surround can be reproduced by amplifying the audio signal subjected to the crosstalk cancellation processing by the amplitude amplifying means 11 and reproducing it by the 2.1 channel speaker system 13.

  The LFE signal is used for the subwoofer (speaker 42LFE) without passing through all the above-described processing. However, when only the 2-channel speakers 42FL and 42FR are used without using the subwoofer as shown in FIG. As shown, after passing through the attenuator 14, it is added to the L and R signals.

<Second Embodiment>
In the first embodiment described above, the technique for detecting the maximum level of all channels has been described. As a second embodiment, a method for setting only the center signal independently will be described. As shown in FIGS. 3 and 7, in the near field listening method of the present invention, the speakers 42FL and 42FR are installed close to the viewer, so that a distance difference between the image of the display 43 and the speakers 42FL and 42FR is generated. . Therefore, the initial reflected sound S1 and the rear reverberation sound S2 are important for providing a sense of distance. However, since the center signal is usually assigned to the center signal of the input audio signal, when the initial reflected sound S1 and the rear reverberation sound S2 are large, the intelligibility of the speech deteriorates or is not suitable for the scene. It is possible that there is a sound. Therefore, it is conceivable to set the gain G2 ′ and the delay value D2 ′ of the rear reverberation sound S2 added to the center channel independently of the other four channels in order to increase the clarity of the lines.

  FIG. 9 is a flowchart showing the flow from the input signal analysis to the parameter setting means 2 when the center channel is independent of the other four channels. When “near position” is set, first, when an audio signal of 5 channels is input, the maximum level Lmax of the input signal of each channel in one frame is detected with a certain number of samples as one frame (step) 111). Next, the maximum level Lmax (C) of the center channel is compared with the maximum level Lmax (4) of the other four channels (step 112).

In step 112, if the maximum level Lmax (C) of the center channel is larger than the maximum level Lmax (4) of the other four channels, the maximum level Lmax (C) of the center channel is used to The maximum level difference Δ is compared with the reference value TH so that the difference Δ between the level Lmax (C, f) and the maximum level Lmax (C, f-1) of the previous frame f-1 does not open too much (step 114). ). If the maximum level difference Δ is smaller than the reference value TH, the maximum level Lmax (C, f) of the current frame is acquired (step 115), and the optimum parameter setting for the acquired maximum level Lmax (C, f) is set as a parameter. This is performed for each frame by the setting means 2 (step 117). The reference value TH here is also determined by the sound field mode. In step 114, if the maximum level difference Δ is larger than the reference value TH, for example, Lmax (C, f) ′ − Lmax (C, f−1) = TH
Then, the smoothing process with the maximum level of the previous frame is performed (step 116). Then, the maximum level Lmax (C, f) ′ after the smoothing process is obtained (step 115), and the parameter setting means 2 sets the optimum parameter setting for the maximum level Lmax (C, f) ′ for each frame. Perform (step 117).

If the maximum level Lmax (C) of the center channel is smaller than the maximum level Lmax (4) of the other four channels in step 112, the level difference between the center channel and the other four channels Δ = Lmax (4) -Lmax (C)
Is obtained (step 113). Next, it is assigned to the parameter setting means 2 (step 117). Similarly to the case where the maximum level Lmax (C) of the center channel is larger, the maximum level difference between the current frame f and the previous frame f-1 of the four channels = Δ (f) −Δ (f−1). Are compared (step 114). When the maximum level difference = Δ (f) −Δ (f−1) is smaller than the reference value TH, the maximum level Lmax (4, f) of the frame f is acquired (step 115), and the maximum level Lmax (4 , f) optimal parameter setting is performed for each frame by the parameter setting means 2 (step 117).

The parameter setting means 2 is based on the maximum level Lmax (4) passed from the input signal analysis means 1 and the level difference Δ = Lmax (4) −Lmax (C) between the center channel and the other four channels. Set the parameters of the sound S2. When the level of the center signal is small, the phenomenon that the clarity of the center signal decreases and the speech becomes difficult to hear occurs, but considering the balance of the rear reverberation sound S2 by acquiring the level difference Δ with other channels, By setting the gain for the channel to be lower than usual, a natural surround feeling can be obtained without degrading the clarity of the speech.

1 is a block diagram showing a first embodiment of a sound image localization control device according to the present invention. It is a block diagram which shows a 5.1 channel speaker system and the 2.1 channel speaker system of a normal position. It is a block diagram in which the 2.1 channel speaker system was close to the viewer. It is a flowchart for demonstrating the process which concerns on this invention. It is a block diagram which shows the rear reverberation sound production | generation means of FIG. It is a block diagram which shows the crosstalk cancellation process of FIG. It is a block diagram which adjoined the viewer with the 2-channel speaker system. It is a block diagram which shows the sound image localization control apparatus of 2 channels. It is a flowchart for demonstrating the process of 2nd Embodiment. It is explanatory drawing which shows a direct sound, an early reflection sound, and a rear reverberation sound.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Input signal analysis means 2 Parameter setting means 3 Parameter information holding means 4 Speaker position / sound field mode setting means 5 Transfer characteristic addition means 6 Direct sound processing means 7 Initial reflection sound processing means 8 Rear reverberation sound addition means 9 Adder 10 Cross Talk cancellation processing means 11 Amplitude amplification means 41 Viewer 42FL, 42FR Speaker

Claims (2)

An audio localization control device for reproducing a multi-channel audio signal by arranging left and right speakers for stereo reproduction at a position closer to the viewer than a normal reproduction position,
Input level detecting means for detecting the input level of the multi-channel audio signal every predetermined time;
It is determined whether or not the current input level detected by the input level detection means is greater than a predetermined value by a predetermined value or more. If not, the gain and / or delay time of the rear reverberant sound is determined as the current input level. A rear reverberation control means for setting the gain and / or delay time of the rear reverberation sound to a value according to a level obtained by adding the previous input level and a predetermined value when large,
A sound image localization control device. An audio localization control program for reproducing a multi-channel audio signal by arranging left and right speakers for stereo reproduction closer to the viewer than a normal reproduction position,
An input level detecting step for detecting an input level of the multi-channel audio signal at predetermined time intervals;
It is determined whether or not the current input level detected by the input level detection means is greater than a predetermined value by a predetermined value or more. If not, the gain and / or delay time of the rear reverberant sound is determined as the current input level. A rear reverberation control step for setting the gain and / or delay time of the rear reverberation sound to a value corresponding to a level obtained by adding the previous input level and a predetermined value when the value is large,
A sound image localization control program to be executed by a computer.

Images