JP4335752B2 - Pseudo stereo signal generation apparatus and pseudo stereo signal generation program - Google Patents

Description

  The present invention relates to a pseudo stereo signal generation technique for generating an artificial stereo signal by inputting an audio signal.

  Currently, audio playback devices equipped with stereo speakers are widely used. If there is a stereo speaker, it is possible to reproduce a stereo signal normally used in a music CD or the like with a good sound image. However, music content recorded in old times and MIDI sound sources that are widely used in electronic musical instruments are often monaural signals, and even when monaural audio signals are played back with stereo speakers, a good sound image is played back. I can't.

  As a pseudo-threatio technology that expands the sound image of a monaural audio signal like a stereo signal, it is described in the book “Handbook of Recording Engineering” by John Argle, published by Stereo Sound Co., Ltd. The Lorisen method shown in FIG. 5 is well known. In the method of Lorisen, pseudo-stereo is made using the fact that the relative characteristics of the peak and the dip can be obtained depending on whether the delay signal is added to or subtracted from the original sound. Thereby, a difference occurs between both ears, and a stereo space in which the sound image is widened compared to monaural can be realized.

However, since the same process is performed over all frequencies in the Laurizen method, there is a problem that the presented sound image is very unclear and unnatural.
Japanese Patent Application Laid-Open No. 7-95696 discloses a method of dividing a signal band and performing different processing in each band. FIG. 6 illustrates the principle of this technology. According to the publication, “a plurality of localization circuits that localize the sound images of the direct sound and the reflected sound are provided, and the localization circuit gives a level difference to the low-frequency component divided by the LPF (low-pass filter). The middle frequency component divided by the band pass filter) is localized in the vertical direction by the notch filter, and the wide frequency component divided by the HPF (high pass filter) is transferred by the FIR (Finite Impulse Response) filter. The sound image is localized by performing a simulation ".

  According to this publication, a good stereo sound image can be obtained by inputting a monaural sound source, but it is necessary to generate direct sound and reflected sound. When applied to a music signal containing reflected sound, excessive reflected sound is generated, and there is a problem that hearing deterioration occurs. In addition, there is also a problem that the apparatus scale is increased by providing a plurality of sound image localization means.

Japanese Unexamined Patent Publication No. 7-95696 By John Argle, translated by Masaki Sawaguchi, "Handbook of Recording Engineering" Stereo Sound, 1988.6, P84-86

  In view of the above problems, an object of the present invention is to provide a pseudo stereo signal generation apparatus that can generate a good sound image for any monaural audio signal and that has a small apparatus scale.

The pseudo stereo signal generation device according to the present invention includes:
A low-pass filter that extracts the low-frequency component of the input signal, and a high-pass filter that also extracts the high-frequency component of the input signal,
Decorrelation means for decorrelating the output signal from the low-pass filter with respect to the input signal to obtain a low-band decorrelation signal;
A high-frequency decorrelation means for decorrelating and outputting the output signal of the high-pass filter with respect to the input signal;
The first adder that outputs the low- pass filter , the low- band decorrelation means, and the output signal from the high-band decorrelation means in the positive phase and the output signal from the low-pass filter in the positive phase, And a second adder for adding the output signal from the low-band decorrelation signal and the output signal from the high-band decorrelation means in reverse phase and outputting the result.

The pseudo stereo signal generation program according to the present invention is
Computer to generate a pseudo-stereo signal from the audio signal,
A low-pass filter that extracts the low-frequency component of the input signal, and a high-pass filter that also extracts the high-frequency component of the input signal,
Decorrelation means for decorrelating the output signal from the low-pass filter with respect to the input signal to obtain a low-band decorrelation signal;
High-pass decorrelation means for decorrelating and outputting the output signal of the high-pass filter with respect to the input signal,
Gain adjusting means for adjusting the gain of the low-band decorrelation signal, the output signal from the low-pass filter, and the output signal from the high-band decorrelation means , respectively;
The output signal from the gain-adjusted low-pass filter is output in positive phase, and the gain-adjusted low-frequency decorrelation signal and the output signal from the gain-adjusted high-frequency decorrelation unit are added in positive phase and output. 1 adder,
The output signal from the gain-adjusted low-pass filter is output in the positive phase, and the gain-adjusted low-band decorrelation signal and the output signal from the gain-adjusted high-band decorrelation means are added in reverse phase and output. It functions as an adder of 2.

According to the present invention, the low-pass signal after passing through the low- pass filter and the high-pass signal after passing through the high-pass filter are decorrelated, so that the power difference between the original signal and the signal obtained by adding / subtracting the decorrelated signal is reduced. The sound image biased to either the left or right is not generated. In addition, components with concentrated power in the low frequency range such as vocals are localized with the sound image spreading slightly near the center, and other high frequency components can present a natural surround stereo image. There is an effect. Furthermore, since it can be configured with only two filters, one decorrelation means, several multipliers, and an adder, it can be realized with a very small device scale.

Embodiment 1 FIG.
FIG. 1 is a diagram illustrating the principle of a pseudo stereo signal generation apparatus according to the present invention.
First, the audio signal 102 is input to the pseudo stereo signal generation device 101. The input audio signal 102 is input to a low-pass filter 103 (hereinafter referred to as LPF) and a high-pass filter 104 (hereinafter referred to as HPF) to obtain a low-frequency signal 105 and a high-frequency signal 106. The low frequency signal 105 is input to the decorrelation means 107 and the multiplier 108. The high frequency signal 106 is input to the signal enhancement unit 111.
The decorrelation means 107 generates a signal having no correlation with the input signal 105 to obtain the low-band decorrelation signal 109, and outputs the low-band decorrelation signal 109 to the multiplier 110. Here, the low-band decorrelation signal 109 is used as a low-band surround signal. The low-frequency signal 105 that is not decorrelated is used as a central localization signal.

  In the signal enhancement means 111, the input high frequency signal 106 is subjected to signal enhancement processing so as to be enhanced on human hearing to generate a high frequency surround signal 112, and the high frequency surround signal 112 is multiplied by a multiplier 113. Output to. Here, the signal emphasis processing in the signal emphasizing means 111 may be any processing such as adding reverberation or adding a head acoustic transmission characteristic from the lateral direction. Alternatively, this means may be omitted.

  The three multipliers 108, 110, and 113 adjust the gains of the input signals, and output them to the two adders 117 and 118. In the two adders 117 and 118, the central localization signal 114 is in-phase, and the low-frequency surround signal 115 and the high-frequency surround signal 116 are added in opposite phases in the adder 117 and the adder 118, respectively, to generate pseudo stereo signals 119 and 120. .

  Here, the gain adjustment will be supplemented. Increasing the gain of the high-frequency surround signal in the multiplier 113 increases the sense of spread of the sound image. In the multiplier 110, if the gain of the low-frequency surround signal is increased, the sound image of vocals, drums, etc. spreads in the case of a music signal. However, if the gain is increased too much, the sound image will be blurred. Therefore, lowering the gain than the high-frequency surround signal is preferable for hearing. When the gain of the central localization signal is increased in the multiplier 108, the overall sound image becomes narrower. These gains may be fixed, but by making adjustments possible from the outside, the user can create a desired sound image.

  Further, as a preferable property of the LPF 103 and the HPF 104, when the frequency characteristics of the LPF and the HPF are added, it is preferable that the characteristics become as flat as possible over the entire frequency.

  According to the present invention, by decorrelating the low-frequency signal 105 after passing through the LPF 103, there is no difference in the power of the signal obtained by adding / subtracting the original low-frequency signal and the low-frequency uncorrelated signal, so No biased sound image is produced. In addition, components with concentrated power in the low frequency range such as vocals are localized with the sound image spreading slightly near the center, and other high frequency components can present a natural surround stereo image. There is an effect. In addition, since it can be configured with only two filters, one decorrelation means, several multipliers, and an adder, it can be realized with a very small device scale.

Embodiment 2. FIG.
Generally, LPF and HPF have a transition band. If attention is paid to this transition band, the output signals of LPF and HPF may be in phase in this transition band. In particular, the phases are aligned when LPF and HPF are realized by the same FIR filter. In the first embodiment, in the transition band, the high-frequency / low-frequency signals having the same phase are added in the opposite phase by the adder 118, so that the signal component is attenuated only for one of the signals. As a result, the sound image of the signal in the transition band is biased to the left or right. In order to prevent this, in the first embodiment, the HPF output signal may be input to the high-band decorrelation unit and then input to the signal enhancement unit.
According to the present embodiment, it is possible to prevent the deviation of the sound image in the transition band of the filter, and it is possible to present a better sound image.

Embodiment 3 FIG.
In the first embodiment, the decorrelation unit 107 is arbitrary. However, the decorrelation unit suitable for the present invention is characterized by a sum signal obtained by adding the input signal and the output signal from the decorrelation unit, and subtraction. It is desired that the difference signal has almost no difference in signal power. For example, the Hilbert transform may be applied as the decorrelation means. That is, if the input signal is subjected to Hilbert transform and a signal whose phase is advanced by 90 degrees is obtained, the difference between the power of the sum signal and the difference signal of the input signal and the signal obtained by Hilbert transform becomes zero. This property will be described with reference to FIG. In FIG. 2, A is a map of a single spectrum on the complex plane. Advance A phase by 90 degrees to obtain B. When A + B and AB are obtained, it can be seen that their absolute values are equal.
Further, when the Hilbert transform is used, discontinuity may occur at the boundary between the blocks on which the Hilbert transform is performed. However, a smooth connection may be made by applying a cross spectrum method or the like.

Alternatively, a phase characteristic that advances the phase by 90 degrees may be approximately realized by an all-pass filter. In this case, discontinuity in the output signal is eliminated in principle. Further, since it can be realized by cascading low-order all-pass filters, it is possible to contribute to the reduction of the device scale.
This embodiment has an effect that a simple decorrelation unit suitable for the present invention can be realized.

Embodiment 4 FIG.
FIG. 3 is a diagram showing another embodiment according to the present invention. The difference from the first embodiment is that a band pass filter (band pass filter, hereinafter referred to as BPF) for extracting a mid band component of the input signal and a mid band decorrelation means are added.

  First, the audio signal 102 is input to the pseudo stereo signal generation device 101. The input signal 102 is input to the LPF 103, the BPF 301, and the HPF 104 to obtain the low frequency signal 105, the mid frequency signal 302, and the high frequency signal 106. The low frequency signal 105 is input to the decorrelation means 107 and the multiplier 108. Midband signal 302 is input to midband decorrelation means 303 and multiplier 304. The high frequency signal 106 is input to the signal enhancement unit 111.

  The decorrelation means 107 generates a signal having no correlation with the input signal 105 to obtain the low-band decorrelation signal 109, and outputs the low-band decorrelation signal 109 to the multiplier 110. Here, the low-band decorrelation signal 109 is used as a low-band surround signal. The low-frequency signal 105 that is not decorrelated is used as a low-frequency central localization signal.

  The mid-band decorrelation unit 303 generates a signal having no correlation with the input signal 102 to obtain the mid-band decorrelation signal 305 and outputs the mid-band decorrelation signal 305 to the multiplier 306. Here, the mid-band decorrelation signal 305 is used as a mid-band surround signal. The mid-range signal 302 that is not decorrelated is used as a mid-range central localization signal.

  In the signal enhancement means 111, the input high frequency signal 106 is subjected to signal enhancement processing so as to be enhanced on human hearing to generate a high frequency surround signal 112, and the high frequency surround signal 112 is multiplied by a multiplier 113. Output to. Here, the signal enhancement processing by the signal enhancement means 111 may be arbitrary, such as adding reverb or adding a head acoustic transmission characteristic from the lateral direction. Alternatively, this means may be omitted.

  The five multipliers 108, 110, 113, 304, and 306 adjust the gains of the input signals, and output them to the two adders 117 and 118. In the two adders 117 and 118, the low-frequency central localization signal 114 and the mid-frequency central localization signal 307 are in phase, and the low-frequency surround signal 115 and the mid-frequency surround signal 308, and the high-frequency surround signal 116 are the adder 117 and the adder 118. Then, the pseudo-stereo signals 119 and 120 are generated by adding them in opposite phases.

Here, as a preferable property of LPF, BPF, and HPF, when the frequency characteristics of LPF, BPF, and HPF are added, it is preferable that the characteristics become as flat as possible over all frequencies. In addition, although BPF is used in the above, more detailed operations can be performed by performing decorrelation processing in each band using multiple BPFs that do not overlap the passbands and performing different gain adjustments for each band. You can go.
According to the present embodiment, different gain adjustments are possible in the two bands, the low band and the middle band, so that there is an effect that it is possible to adjust a finer sound image.

Embodiment 5 FIG.
Embodiments 1 to 4 all target monaural audio signals, but can also target stereo audio signals. FIG. 4 is a diagram showing another embodiment in which the present invention is adapted to a stereo audio signal.

First, the left and right audio signals 402 and 403 are input to the pseudo stereo signal generation device 401. The left and right signals 402 and 403 are input to an adder 404 and an adder 405 that performs subtraction processing to obtain a sum signal 406 and a difference signal 407. The sum signal 406 is configured in the same manner as the pseudo stereo signal generation apparatus 101 shown in the first embodiment, and is input to the pseudo stereo converting means 416 and the multiplier 408 operating in the same manner. The pseudo-stereo means 416 performs pseudo-stereo processing on the input signal 406 to obtain the left and right pseudo stereo signals 119 and 120 whose sound images are enlarged, and outputs them to the adders 412 and 413.
Multipliers 408 and 409 adjust the gain of the input signal, and output sum signal 410 and difference signal 411 to adders 412 and 413. Here, the difference signal 411 is output so as to be in opposite phases on the left and right.

The adders 412 and 413 add the three input signals to generate a left output signal 414 and a right output signal 415, respectively.
Here, the gain adjustment will be supplemented. When the gain of the difference signal 407 is increased in the multiplier 409, the phase difference between the left and right signals increases, and the surround effect is improved. In this embodiment, the sum signal 406, which is the central localization component, can also be pseudo-stereo and the sound image can be enlarged, so that a more effective surround effect can be obtained.
According to the present embodiment, there is an effect that it is possible to enlarge and present a sound image not only for a monaural signal but also for a stereo signal.

  Note that the processing described in the first to fifth embodiments can also be realized by operating a computer by a program.

By applying the present invention to an audio playback apparatus equipped with a stereo speaker, a monaural audio signal can be pseudo-converted into a stereo signal and played back, and a stereo space in which the sound image is widened compared to monaural is realized. Can do.
In addition, a stereo space in which a sound image spreads with a small recording amount can be realized by performing a long-time recording mode in monaural on a DVD recorder or the like and applying the present invention at the time of reproduction to perform pseudo-stereo reproduction.

1 is a configuration diagram of a pseudo stereo signal generation apparatus according to Embodiment 1 of the present invention. It is a figure for demonstrating the property of Hilbert transformation. It is a block diagram which shows another embodiment of this invention. It is a block diagram which shows another embodiment of this invention. It is a block diagram of the pseudo | simulation stereo signal generation apparatus by the method of Laurizen. It is principle explanatory drawing of the technique which divides | segments a signal zone | band and performs a different process in each zone | band.

Explanation of symbols

  101, 401: pseudo stereo signal generator, 102: audio signal, 103: low-pass filter, 104: high-pass filter, 105: low-frequency signal, 106: high-frequency signal, 107: decorrelating means, 108, 110, 113: Multiplier, 109: Low-band decorrelation signal, 111: Signal enhancement means, 112: High-frequency surround signal, 114: Central localization signal, 115: Low-frequency surround signal, 116: High-frequency surround signal, 117, 118: Addition 119, 120: pseudo stereo signal, 302: midband signal, 303: midband decorrelation means, 304, 306: multiplier, 305: midband decorrelation signal, 402, 403: audio signal, 404, 405, 412, 413: adder, 406: sum signal, 407: difference signal, 408: multiplier, 411: difference signal, 416: pseudo-stereoscopic means.

Claims (8)

In an apparatus for generating a pseudo stereo signal from an audio signal,
A low-pass filter that extracts the low-frequency component of the input signal;
Similarly, a high-pass filter that extracts the high-frequency component of the input signal,
Decorrelation means for decorrelating the output signal from the low-pass filter with respect to the input signal to obtain a low-band decorrelation signal;
A high-frequency decorrelation means for decorrelating and outputting the output signal of the high-pass filter with respect to the input signal;
A first adder that outputs the output signal from the low-pass filter, the output signal from the low-band decorrelation means, and the output signal from the high-band decorrelation means in positive phase, respectively, and the low-pass filter And a second adder for adding and outputting the output signal from the high-frequency decorrelation means in reverse phase. Stereo signal generator. Gain adjustment means for separately adjusting the gain of the low-frequency decorrelation signal, the output signal from the low-pass filter, and the output signal from the high-frequency decorrelation means , and outputting them to both adders is provided. 2. The pseudo stereo signal generation device according to claim 1. 2. A signal emphasizing unit that adds reverb to the output signal from the high-frequency decorrelation unit or adds a head-related sound transmission characteristic from the lateral direction and outputs the resultant signal to each adder. Or the pseudo stereo signal generation device according to 2. Pseudo stereo signal generating apparatus according to any one of claims 1, characterized in that a configuration for processing using Hilbert transform as the decorrelation means 3. Pseudo stereo signal generating apparatus according to any one of 3 claims 1 characterized by using the all-pass filter as the decorrelation means. A bandpass filter that extracts the mid-frequency components of the input signal;
Midband decorrelation means for decorrelating the output signal from the bandpass filter to the input signal to obtain a midband decorrelation signal;
Further comprising gain adjustment means for adjusting the gain of the output from the mid-band decorrelation signal and the band pass filter,
The first adder further adds, in positive phase, the output signal from the band-pass filter whose gain has been adjusted and the output signal from the mid-band decorrelation means whose gain has been adjusted,
The second adder is configured to further add the output signal of the gain-adjusted band-pass filter in the positive phase and the gain-adjusted output signal of the mid-band decorrelation means in the negative phase. The pseudo stereo signal generation device according to claim 2. When the input signal is a stereo signal, an adder that obtains a sum signal and a difference signal by adding and subtracting the left and right signals of the input stereo signal, and a gain adjusting means that adjusts the gain of the difference signal and the sum signal, respectively.
The pseudo-stereo signal generating means configured by the pseudo-stereo signal generation device according to claim 1, wherein a pseudo-stereo signal is obtained by inputting a sum signal;
A third adder that outputs one of the outputs of the pseudo-stereoizing means, the gain-adjusted sum signal and the difference signal, respectively, added in positive phase;
And a fourth adder for adding and outputting the other output of the pseudo-stereoizing means and the gain-adjusted sum signal in the positive phase and the gain-adjusted difference signal in the opposite phase, respectively. A pseudo stereo signal generating device. Computer to generate a pseudo-stereo signal from the audio signal,
A low-pass filter that extracts the low-frequency component of the input signal, and a high-pass filter that also extracts the high-frequency component of the input signal,
Decorrelation means for decorrelating the output signal from the low-pass filter with respect to the input signal to obtain a low-band decorrelation signal;
High-pass decorrelation means for decorrelating and outputting the output signal of the high-pass filter with respect to the input signal,
Gain adjusting means for adjusting the gain of the low-band decorrelation signal, the output signal from the low-pass filter, and the output signal from the high-band decorrelation means , respectively;
The output signal from the gain-adjusted low-pass filter is output in positive phase, and the gain-adjusted low-frequency decorrelation signal and the output signal from the gain-adjusted high-frequency decorrelation unit are added in positive phase and output. 1 adder,
The output signal from the gain-adjusted low-pass filter is output in the positive phase, and the gain-adjusted low-band decorrelation signal and the output signal from the gain-adjusted high-band decorrelation means are added in reverse phase and output. A pseudo-stereo signal generation program for functioning as an adder of 2.

Images