JP5652515B2 - Signal processing device - Google Patents

Description

  The present invention relates to a signal processing apparatus that performs various operations on an input signal, and more particularly to a technique for extending a low frequency component.

  A speaker having a small diameter may not be able to physically output a low frequency component. Therefore, by generating a harmonic of the low frequency component and adding it to the original signal, it is possible to use a phenomenon called missing fundamentals so that the low frequency component that is not physically output is output. A technique for listening is used (see, for example, Patent Document 1).

Japanese National Patent Publication No. 11-509712

  The harmonics can be generated by a method such as clipping an input signal or rectifying, but these two methods have a problem that only odd-order or even-order components are generated. On the other hand, in the case of integral calculation, both odd-order and even-order components can be generated, but there is a problem that response is poor.

  Accordingly, an object of the present invention is to provide a signal processing apparatus capable of generating odd-order and even-order harmonics in real time.

  The signal processing apparatus according to the present invention includes an input unit that inputs an audio signal, an acquisition unit that acquires an audio signal from the input unit, a power calculation unit that performs a power operation, and an output signal of the power operation unit as an audio signal of the input unit Adding means for adding to and outputting. The power calculation unit performs, for example, the third power and fourth power calculation on the audio signal acquired by the acquisition unit. According to the third and fourth quadrature formulas of cos, the first order (fundamental) component and third order (third harmonic) component of the input signal are generated by the third power calculation, and the zeroth order (DC) component of the input signal by the fourth power calculation, A secondary (second overtone) component and a fourth order (fourth harmonic) component are generated. Therefore, by adding these signals, odd-order and even-order harmonics can be generated in real time. Note that the DC component may be removed by a subsequent high-pass filter or the like.

  The effect of the missing fundamental is to detect the difference frequency of the harmonic component (bass that does not actually exist), so even if only the even-order component (or odd-order component) is added to the original signal alone, The difference frequency is twice the original frequency, and the fundamental (fundamental) perception is somewhat inferior. However, in the present invention, since both odd-order and even-order harmonics are added to the original signal, the difference frequency matches the fundamental tone, and it is possible to strongly sense the volume of bass.

  In the above-described invention, an aspect including a compression unit that performs dynamic range compression of the audio signal acquired by the acquisition unit and generates a compressed signal is also possible. In this case, the power calculation unit performs the multiplication by multiplying the compressed signal and the audio signal acquired by the acquisition unit. For example, in the case of a cube operation, the compressed signal is squared and multiplied by the audio signal acquired by the acquisition unit. If the signal is simply raised to the power, the amplitude value is also raised, so that the dynamic range becomes a very large signal. However, by raising the compressed signal to the power, the dynamic range of the signal after the power can be suppressed. Note that if only the compressed signal is raised to the power, the dynamic range may be excessively compressed or the sense of distortion may increase. Therefore, it is used as a signal to be multiplied to some extent by the original signal (audio signal acquired by the acquisition unit). It is an aspect.

  The dynamic range compression as described above may be performed on the signal after the power operation. In this case, the adding means adds the signal after the power calculation subjected to dynamic range compression to the audio signal input to the input unit and outputs the result. If the level of the input signal is low and the signal level after the power operation is too low, the low-frequency expansion effect cannot be obtained (the effect of missing fundamentals is not felt in the sense of hearing), or the input signal level is too high There is a possibility that the level of the signal after the power calculation is too high and may be recognized as a distorted sound. However, by adding the signal after the power calculation after dynamic range compression to the original audio signal, Sound pressure can be suitably compensated.

  Furthermore, in the dynamic range compression, the input signal is smoothed (eg, moving average) on the time axis, a gain adjustment value is calculated based on the smoothed signal, and the input gain gain value is input. It is good also as an aspect which adjusts the level of a signal. If the level of the input signal suddenly rises or falls sharply, the dynamic range may change drastically, and distortion sound due to compression may be noticeable. Even if there is a sudden rise or fall in the signal level, the dynamic range compression mode does not change abruptly, so that the distortion sound caused by the compression can be made inconspicuous.

The signal processing device is provided between the level detection unit for detecting the level of the audio signal acquired by the acquisition unit, the power calculation unit, and the addition unit, and extracts and outputs a low frequency range of the audio signal And a control unit that performs control to enable or disable the low-frequency extraction function of the low-frequency extraction unit, and the control unit has a level of the audio signal detected by the level detection unit. When the value is equal to or higher than a predetermined value, the low frequency extraction function of the low frequency extraction unit is validated, and when the level of the audio signal detected by the level detection unit is less than a predetermined value, the low frequency extraction function of the low frequency extraction unit May be invalidated .

  As the level of the audio signal increases, the level of the harmonic generated by the power calculation unit also increases. However, in this configuration, when the level of the audio signal is large, the harmonic does not easily pass through the low-frequency extraction unit.

  The audio signal may be composed of audio signals of a plurality of channels including a center channel, and the level detector may detect the level of the audio signal of the center channel.

  In this configuration, when the level of the center channel including a large amount of speech is high, that is, when speech is detected, it is difficult for harmonics to pass through the low-frequency extraction unit. Therefore, the lines are less susceptible to harmonics.

  According to the present invention, odd-order and even-order harmonics can be generated in real time.

It is the block diagram which showed the structure of the speaker apparatus. It is the block diagram which showed the structure of the signal processing part. It is the block diagram which showed the structure of the harmonic generation part. It is the figure which showed an example of the DRC table. It is the block diagram which showed the structure of the dynamic range compression part. It is the figure which showed the change (change of the signal of a time-axis) of a signal when a harmonic component is added.

  An embodiment according to a signal processing apparatus of the present invention will be described. FIG. 1 is a block diagram showing a configuration of a speaker device incorporating a signal processing device of the present invention. The speaker device includes an input unit 1, a signal processing unit 2, and a sound emitting unit 3.

  The input unit 1 includes a plurality of channels (C: center channel, L: front left, R: front right, SL: surround left, SR: surround right, SBL: surround back left, SBR: surround back right, LFE: subwoofer) Audio signals are input. The number of channels is not limited to 7.1 ch shown in the figure, and may be an example in which a larger number of channels are input, or may be monaural.

  The signal processing unit 2 corresponds to the signal processing device of the present invention, performs low-frequency extension processing on the audio signal Sin input from the input unit 1, and outputs the audio signal Sout to the sound emitting unit 3. The sound emitting unit 3 performs a sound field applying process or the like on the audio signal Sout input from the signal processing unit 2, amplifies the sound, and then emits sound from a speaker (not shown). The speaker may be provided independently for each channel, or may be a speaker array in which a large number of small-diameter speaker units are arranged. In the case of a speaker array, the audio signal of each channel is distributed to all (or some) speaker units and controlled by delay, so that the emitted sound is beamed and reflected directly to the listener or reflected from the wall surface. You may make it reach.

  FIG. 2 is a block diagram illustrating a configuration of the signal processing unit 2. The notation such as “Lch in” shown on the left side in the figure indicates the input of the audio signal of each channel, and indicates the audio signal (Sin) input from the input unit 1. The notation such as “Lch out” shown on the right side in the figure indicates the output of the audio signal of each channel, and indicates the audio signal (Sout) output to the sound emitting unit 3.

  The acquisition unit 11 acquires the audio signal of each channel, adds these audio signals, and outputs them to the LPF 12. At this time, the acquisition unit 11 divides the added audio signal by the added number of channels and normalizes the divided audio signal. Note that the addition levels between channels may be the same, or the levels of some channels may be raised before addition to emphasize some channels. Further, the present invention is not limited to an example in which the audio signals of all channels are added, but may be an example in which audio signals of some channels are added.

  The LPF 12 is a low-pass filter that attenuates a band component equal to or higher than the cutoff frequency of the audio signal output from the acquisition unit 11 and outputs the attenuated band component to the HPF 13. The LPF 12 is realized by, for example, a secondary IIR filter.

  The HPF 13 is a high-pass filter, and attenuates a band component below the cutoff frequency of the audio signal output from the LPF 12 and outputs the attenuated band component to the harmonic generation unit 14. The HPF 13 is also realized by, for example, a secondary IIR filter.

  A band pass filter is realized by the LPF 12 and the HPF 13, and only a partial band of the audio signal is extracted and output to the harmonic generation unit 14 as a signal Sa. Note that the passband as the bandpass filter is set according to the bass band to be expanded. For example, when the lowest frequency that the speaker can output is 100 Hz, the cutoff frequency of the LPF 12 is set to 100 Hz. Further, the cutoff frequency of the HPF 13 may be set according to the auditory characteristics (for example, 20 Hz).

  The harmonic generation unit 14 generates a harmonic of the signal Sa output from the HPF 13 and outputs the harmonic as the signal Sb to the LPF 15. As shown in FIG. 3, the harmonic generation unit 14 includes an absolute value calculation unit (ABS) 141, a moving average calculation unit 142, a gain calculation unit 143, a level adjustment unit 144, a third power calculation unit 145, and a fourth power calculation unit. 146.

  The input signal (signal output from the HPF 13) Sa is input to the absolute value calculation unit 141, the level adjustment unit 144, the third power calculation unit 145, and the fourth power calculation unit 146, respectively.

  The absolute value calculator 141 calculates the absolute value of the amplitude of the input signal Sa and outputs it to the moving average calculator 142. The moving average calculation unit 142 calculates a moving average of absolute values (for example, a moving average of several samples to several hundred samples). Thereby, a component similar to the envelope is extracted. In addition to the moving average, other calculations may be used as long as the time axis is smoothed. For example, a root mean square calculation may be used.

  The gain calculation unit 143 considers the absolute value and the moving average value as an envelope, and calculates a gain value according to a predetermined table or function. For example, the gain value according to the dynamic range compression (DRC) table shown in FIG. 4 is calculated. In the example of FIG. 4, when the level on the input side is large (in the example of the figure, −8 dB or more), the output component is reduced and the level of the input side is small (in the example of the figure, from −8 to −21 dB). In this case, the output component is increased. On the other hand, if the level on the input side is too small (−21 dB or less in the example in the figure), it is regarded as a noise component, and conversely, the level is adjusted to be lowered. Of course, the DRC method is not limited to this example, and other methods may be used.

  The level adjustment unit 144 adjusts the level of the input signal Sa with the gain value calculated by the gain calculation unit 143, and outputs the level to the third power calculation unit 145 and the fourth power calculation unit 146 as a compressed signal comp (Sa). Note that the compression signal comp (Sa) is a signal whose gain is controlled according to a signal smoothed on the time axis by a moving average, and even if the input signal has a sudden level increase or level decrease, Since the range does not change abruptly, the distortion sound due to compression can be made inconspicuous.

A cube power calculation unit 145 and a fourth power calculation unit 146 respectively perform a cube calculation and a cube operation on the input signals. Thereby, the harmonics of the odd-order component and the even-order component of the input signal Sa are generated in real time. That is, the triple angle and quadruple angle formulas for cos are expressed as cos 3θ = 4 cos ³ θ- ³ cos θ and cos 4θ = 8 cos ⁴ θ-8 cos 2θ + 1, respectively. Next (fundamental) component, third order (third harmonic) component are generated, and the fourth order (direct current) component, second order (second harmonic) component, and fourth order (fourth harmonic) component of the input signal are generated by the fourth power calculation. . Therefore, by adding and outputting the output signals of the third power calculator 145 and the fourth power calculator 146 by the adder 147, the harmonic generator 14 includes continuous harmonic components from the 0th order to the 4th order. A signal (output signal Sb) can be output.

  However, when the input signal is simply raised to the power, the amplitude value is also raised to the power, so that the signal has a very large dynamic range. Therefore, in the present embodiment, the dynamic range is suppressed by raising the compressed signal comp (Sa) to the power by the cube calculating unit 145 and the fourth calculating unit 146. That is, the cube calculation unit 145 performs the cube calculation by squaring the compressed signal comp (Sa) and multiplying this by the input signal Sa. The fourth power calculation unit 146 performs the fourth power calculation by multiplying the compressed signal comp (Sa) to the third power and multiplying this by the input signal Sa. Thereby, the dynamic range of the signal after the power can be suppressed. Note that if only the compressed signal comp (Sa) is raised to the power, the dynamic range is excessively compressed, so that distortion sound may be noticeable. Therefore, the cube calculation unit 145 and the fourth power calculation unit 146 use the input signal Sa as a multiplication signal to some extent (one in the above example) so that the increase in the dynamic range is suppressed but the compression is not excessive. .

  In the above method, the third-order component and the fourth-order component are at a lower level than the first-order component and the second-order component and do not include the fifth-order or higher component. The next or higher component may be calculated. However, the harmonic generation unit 14 according to the present embodiment prevents the harmonic component from being heard as a distorted sound by suppressing the generation of the fourth-order component and suppressing the levels of the third-order and fourth-order components. However, the low-frequency expansion as a missing fundamental is conspicuous.

  The output signal Sb generated as described above is input to the DRC 18 via the LPF 15, the HPF 16, and the LPF 17.

  The LPF 15 is a low-pass filter for removing high-order components generated at the time of generating harmonics, and attenuates a band component equal to or higher than the cutoff frequency (for example, 200 Hz) of the signal output from the harmonic generation unit 14 and outputs it to the HPF 16. To do. In particular, when a non-integer multiple component is input to the input signal Sin (for example, when the input signal has two waves of cos ω1 and cos ω2 and n · ω1 ≠ ω2), the harmonic generation unit 14 has an unnecessary high frequency. There is a possibility that a second harmonic component (for example, a cross-modulation component such as cos (ω1 + ω2)) may occur, and the LPF 15 has a function for removing these unnecessary higher harmonic components. The LPF 15 is also realized by a secondary IIR filter.

  The HPF 16 is a high-pass filter for mainly removing the DC component and non-integer multiple components on the low frequency side, and attenuates the band component below the cutoff frequency (for example, 50 Hz) of the audio signal output from the LPF 15 to reduce the LPF 17. Output to. The HPF 16 is also realized by, for example, a secondary IIR filter.

  A band pass filter is realized by the LPF 15 and the HPF 16, and unnecessary components are removed from the signal Sb including the harmonic component.

  The LPF 17 is not essential in the present invention, but is a low-pass filter for fine adjustment. The combination of the LPF 17 and the preceding LPF 15 adjusts the cut-off frequency and the Q value to gently attenuate the high frequency side. It can be made sharp or steep. Also, for example, when speech such as speech is included in the audio signal acquired by the acquisition unit 11, speech of speech is adjusted by adjusting the frequencies of the LPF 15 and LPF 17 so that harmonic components are not added to the speech bandwidth of the speech. Can be made easier to hear. Whether or not speech of lines is included may be determined from the level of the Cch signal. For example, a Cch signal level detection unit may be provided, and the processing of the LPF 17 may be turned on if the level of the Cch signal is equal to or higher than a predetermined value, and the LPF 17 may be turned off if the level of the Cch signal is lower than the predetermined value.

  Next, as shown in FIG. 4, the DRC 18 includes an absolute value calculation unit (ABS) 181, a moving average calculation unit 182, a gain calculation unit 183, and a level adjustment unit 184. That is, it has the same configuration and function as the absolute value calculation unit (ABS) 141, moving average calculation unit 142, gain calculation unit 143, and level adjustment unit 144 of the harmonic generation unit 14 shown in FIG. Therefore, detailed description of each component is omitted in FIG.

  The DRC 18 performs dynamic range compression of the signal Sc output from the LPF 17 and appropriately controls the level of the harmonic component. That is, if the level of the input signal is low and the level of the signal Sc including the harmonic component is too low, the effect of low-frequency band expansion cannot be obtained (the effect of missing fundamentals is not felt in the sense of hearing), or the level of the input signal Is too high, the level of the signal Sc including the harmonic component may be too high to be recognized as a distorted sound. However, the signal Sc is converted into a dynamic range compressed signal Sd and added to the original signal Sin. Thus, the low-frequency sound pressure can be suitably compensated. Further, the signal Sd is a signal whose gain is controlled according to a signal smoothed on the time axis by moving average, and even if there is a sudden level rise or level drop in the input signal, the dynamic range compression mode Does not change abruptly, so that distortion sound due to compression can be made inconspicuous.

  Returning to FIG. 2, the signal Sd output from the DRC 18 is input to the adder 19. The adder 19 adds the signal Sd input from the DRC 18 to the original signals (Cin, Lin, Rin, SLiin, SRin, SBLin, SBRin, LFEin) of each channel. Note that if the same signal Sd is input to each channel, only the harmonic component becomes too large, so the adder 19 divides the signal Sd by the number of channels to be added. Note that the addition level between channels may be the same, and for some channels, the level of the signal Sd (the signal after division) is increased and added to emphasize some channels. Good. Further, the present invention is not limited to the example of adding to the audio signals of all channels, but may be an example of adding to the audio signals of some channels.

  For Lin and Rin, after the signal Sd is added, the frequency characteristics are adjusted by the PEQ 21. The PEQ 21 is not essential in the present invention, but increases the level of a predetermined band in order to emphasize harmonic components. In particular, when a non-integer multiple component is input to the input signal Sin, a non-integer multiple component of harmonics may remain in the passbands of the LPF 15 and the HPF 16 and may be included as unnecessary distortion sound. In this case, it is necessary to lower the level of the signal Sd so that the distorted sound does not stand out. Therefore, the PEQ 21 adjusts the frequency characteristic so as to emphasize the harmonic component by increasing the level of the predetermined frequency component (or lowering the other band). However, if the signal Sa input to the harmonic generation unit 14 is a single frequency component, the processing is not necessary. For example, when a single frequency component detection unit is provided and a single frequency component is input, The level control of the signal Sd and the function of the PEQ may be stopped.

  FIG. 6 shows signal changes (time-axis signal changes) when harmonic components are added by the above-described method. FIG. 2A shows a sine wave signal (50 Hz as an example). The output signal Sout when this sine wave is input to the signal processing unit 2 as the input signal Sin is shown in FIG. As shown in FIG. 5B, the output signal Sout is asymmetric between positive and negative. That is, it means that even order components are included in addition to odd order components. Therefore, according to the signal processing unit 2 of the present embodiment, it is possible to generate harmonic components that are continuous from the lower order, and the effect as a missing fundamental compared to an aspect in which only odd-order components or only even-order components are generated. Can be emphasized more. In addition, since harmonics can be generated by exponentiation of signals on the time axis, it is not necessary to perform calculations on the frequency axis, and odd-order and even-order harmonics can be generated in real time. In addition, when a processing delay that can be perceived by the auditory sense occurs due to the ability of the signal processing unit (the computing ability of the DSP or CPU), the original signal Sin is subjected to delay processing to correct a time lag. May be.

DESCRIPTION OF SYMBOLS 1 ... Input part 11 ... Acquisition part 12 ... LPF
13 ... HPF
DESCRIPTION OF SYMBOLS 14 ... Harmonic wave generation part 141 ... Absolute value calculating part 142 ... Moving average calculating part 143 ... Gain calculating part 144 ... Level adjusting part 145 ... Third power calculating part 146 ... Fourth power calculating part 147 ... Adder 15 ... LPF
16 ... HPF
17 ... LPF
18 ... DRC
182 ... Moving average calculator 183 ... Gain calculator 184 ... Level adjuster 19 ... Adder 2 ... Signal processor 21 ... PEQ
3 ... Sound emission part

Claims (2)

An input section for inputting an audio signal;
An acquisition unit for acquiring an audio signal input to the input unit;
A power calculation unit that performs a power calculation on the audio signal acquired by the acquisition unit;
Adding means for adding the signal after the power operation calculated by the power operation unit to the audio signal input to the input unit;
A level detection unit for detecting the level of the audio signal acquired by the acquisition unit;
A low-frequency extraction unit that is provided between the power calculation unit and the adding means, and extracts and outputs a low frequency of the audio signal;
A control unit that performs control to enable or disable the low-frequency extraction function of the low-frequency extraction unit; and
With
The control unit enables the low-frequency extraction function of the low-frequency extraction unit when the level of the audio signal detected by the level detection unit is equal to or higher than a predetermined value, and the level of the audio signal detected by the level detection unit is When it is less than a predetermined value, the low frequency extraction function of the low frequency extraction unit is invalidated .
Signal processing device.
The audio signal is composed of audio signals of a plurality of channels including a center channel,
The level detection unit detects the level of the audio signal of the center channel;
The signal processing apparatus according to claim 1.

Images