JPH09181545A - Frequency dependent digital limiter - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a digital limiter in which bleeding does not exist, a limiter operation is guaranteed and the degradation of an S/N is not generated. SOLUTION: After an input signal is passed through a reverse characteristic filter 1 having the frequency characteristic to be the reverse characteristic to the frequency characteristic of a desired amplitude compression, an amplitude compression type nonlinear processing which does not depend on the frequency is performed in a digital soft crump part 2 and the signal is inputted in the forward characteristic filter 3 having the frequency characteristic of the desired amplitude compression. At the time of the high level of the input signal, the frequency characteristic of the forward characteristic filter 3 becomes, predominant. At the time of the low level, the characteristic becomes a flat frequency characteristic. In an over-sampling part 4, a resampling is performed by the frequency which is (n) times the sampling frequency. In a decimation part 6, the signal is returned to the digital signal of an original sampling frequency f0 and the digital signal is outputted.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a frequency-dependent digital limiter in which the amplitude compression characteristic changes depending on the frequency. It can be used to prevent overmodulation or to reduce sound quality deterioration due to overmodulation prevention in FM modulation.

[0002]

2. Description of the Related Art FIG. 8 is a block diagram of a first conventional example. In the figure, 51 is a first band division filter, and 52 is a second band division filter.
Band division filter, 53 is a quasi-instantaneous compression circuit, and 54 is R
An MS detector, 55 is a gain controller, 56 is a variable gain amplifier, and 57 is an adder. According to this conventional technique, quasi-instantaneous compression is performed on a component of a specific frequency band of an input signal so that a frequency characteristic becomes flat for a low level signal.

The input signal is input to the first and second band division filters 51 and 52, and the signals in the frequency band subjected to the quasi-instantaneous compression are added via the second band division filter 52 and the quasi-instantaneous compression circuit 53. The output from the first band-division filter 51 is added to the output by the output device 57. The quasi-instantaneous compression circuit 53 inputs the input signal to this circuit to the variable gain amplifier 56, and at the same time, outputs the effective value to the RMS detector 54.
The variable gain amplifier 56 is detected via the gain controller 55, and the gain is suppressed to a lower value as the input level increases. The circuit shown in the figure may be an analog circuit, or a digital signal processing processor (hereinafter referred to as D
It can also be realized by a digital circuit using (for example, SP).

However, since the RMS detector 54 rectifies an input signal and obtains an effective value through a low-pass filter, it has a varying high-level signal component and a constant low-level signal component, for example, in the input signal. , Hiss noise and the like are included, there is a problem in that the gain control determined by the high-level signal component causes fluctuations in the steady low-level signal component, which easily causes deterioration called breathing. Also, for transient input signals, there is a problem that the gain cannot be controlled due to the influence of the time constant of the low-pass filter, and the limiter operation cannot be performed. Occur.

9 and 10 are explanatory views of a second conventional example, FIG. 9 is a block diagram, and FIG. 10 is a circuit diagram of a main part block thereof. In the figure, 1 is an inverse characteristic filter, 3 is a forward characteristic filter, 61 is a diode clamp circuit, 62 is an operational amplifier, 63 is an input side resistor R _S , 64 and 67 are first voltage dividing resistors R _A , 65 and 68. Is a second voltage dividing resistor R _B , 66,
69 is a feedback diode D _i . As shown in FIG.
In this second conventional example, an input signal is passed through an inverse characteristic filter 1 which has an inverse characteristic to the frequency characteristic of the amplitude compression amount, and is then instantaneously compressed by the diode clamp circuit 61. The frequency characteristic is flattened with respect to a low level signal by performing the instantaneous compression in which the amplitude compression amount changes depending on the frequency through the forward characteristic filter 3 as follows.

As shown in FIG. 10, the diode clamp circuit 61 includes an operational amplifier 62 and a first voltage dividing resistor R _A 6
4, 67, a second voltage dividing resistor R _B 65, 68, and a feedback diode D _i 66, 69 are used as a basic configuration of a diode limiter, and four feedback paths having different voltage dividing ratios and different resistance values are provided. I have.

The input-side resistor R _S 63 is connected to the negative input terminal of the operational amplifier 62, and is connected to the cathode side of four feedback diodes such as the feedback diode D _i 66 and the feedback diode D _i 69. The anode side of the book feedback diode is connected. The output voltage of the operational amplifier 2 is raised in the positive direction becomes conductive and the feedback diode D _i 66 to _{V · (R A / R B} ) , a second voltage dividing resistors R _A 64 serves as a feedback resistor. Similarly, the output voltage of the operational amplifier 2 is increased to the negative direction becomes conductive and the feedback diode D _i 69 to _{-V · (R A / R B} ), a second voltage dividing resistors R _A 67 is as a feedback resistor work. Since four return paths are provided, the compression circuit has positive and negative symmetry.

In the analog instantaneous compression type in which an analog signal is nonlinearly processed by a diode clamp circuit as shown in FIGS. 9 and 10, the inverse characteristic filter 1 and the forward characteristic filter 3 are used.
Are connected in series, the inverse characteristic filter 1 in the previous stage is used.
The signal in the frequency band whose level has been lowered by is amplified by the forward characteristic filter 3 in the subsequent stage, but since the noise level of the analog signal has a constant value, the S / N ratio deteriorates.

[0009]

SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned circumstances, and does not suffer from deterioration such as breathing as in the conventional quasi-instantaneous compression circuit, and can be applied to an impulse-like input signal. It is an object of the present invention to provide a frequency-dependent digital limiter that guarantees a limiter operation and does not cause deterioration of the S / N ratio like an analog instantaneous compression circuit.

[0010]

According to a first aspect of the present invention, in a frequency-dependent digital limiter,
An inverse characteristic filter, an oversampling section for inputting the output of the inverse characteristic filter, an amplitude compression type non-linear processing section for inputting the output of the oversampling section, and a decimation section for inputting the output of the amplitude compression type non-linear processing section. And a forward characteristic filter for inputting the output of the decimation unit, and an amplitude compression characteristic according to the frequency characteristic of the forward characteristic filter.

According to a second aspect of the present invention, in a frequency-dependent digital limiter, an inverse characteristic filter, a plurality of cascaded digital soft clamp units for inputting the output of the inverse characteristic filter, and the cascade connection are provided. A forward characteristic filter for inputting the output of the digital soft clamp section of the plurality of stages, the digital soft clamp section, an amplitude compression type non-linear processing section for inputting the output of the oversampling section, It is characterized in that it has a decimation section for inputting the output of the amplitude compression type non-linear processing section and has an amplitude compression characteristic according to the frequency characteristic of the forward characteristic filter.

According to a third aspect of the present invention, in the frequency dependent digital limiter according to the first or second aspect, the amplitude compression type non-linear processing section adds an input and an odd power term of the input. It is characterized by having means.

[0013]

FIG. 1 is a block diagram of a first embodiment of the present invention. In the figure, the same parts as those in FIG. Reference numeral 2 is a digital soft clamp unit, 4 is an oversampling unit, 5 is an amplitude compression type nonlinear processing unit, and 6 is a decimation unit. This embodiment is a frequency-dependent digital limiter that performs amplitude compression nonlinear processing in the digital domain.

An input signal such as a digital voice signal is passed through an inverse characteristic filter 1 having a frequency characteristic opposite to the desired frequency characteristic of amplitude compression, and then input to a digital soft clamp section 2 where a digital signal is input. By the signal processing, amplitude compression type non-linear processing independent of frequency is performed. This output is input to the forward characteristic filter 3 having a desired amplitude compression frequency characteristic, and this output becomes a digital signal output. When the input signal is at a high level, the frequency characteristic of the forward characteristic filter 3 becomes dominant, and when it is at a low level, the frequency characteristic is flat.

The digital soft clamp unit 2 comprises an oversampling unit 4, an amplitude compression type non-linear processing unit 5 and a decimation unit 6 which are connected in cascade. In the oversampling unit 4, when f ₀ is the sampling frequency of the input signal and n is an integer of 2 or more, re-sampling is performed at a higher sampling frequency of nf ₀ . For example, the number of sampling points is increased and 0 bit is inserted here. The amplitude compression type non-linear processing unit 5 processes the signal at a sampling frequency nf ₀ that is n times higher. The decimation unit 6 thins out the signal-processed data to restore the original digital signal of the sampling frequency f ₀ and outputs the digital signal.

Since the amplitude compression type non-linear processing section 5 realizes instantaneous compression, the breathing phenomenon disappears and the sound quality deterioration is solved. Further, the uncertainties of the limiter operation are also solved, which is effective in reducing overmodulation. Since the instantaneous compression is realized in the digital domain, it is possible to secure a significantly higher S / N ratio than the analog type even with the same instantaneous compression. On the other hand, it generates harmonic distortion and intermodulation distortion by the nonlinear processing, higher component than the Nyquist frequency f _0/2 in this generates aliasing. However, in order to perform nonlinear processing in the frequency nf ₀ performs pre oversampling, it is possible to suppress the occurrence of aliasing noise for easily removing higher than the Nyquist frequency f _0/2 component. As for the components in the frequency band of the input signal among the harmonic distortion and the intermodulation distortion, as a result of conducting a trial listening experiment, it is relatively difficult for the musical sound having a complicated spectrum to perceive these distortions. I know.

In the above description, the inverse characteristic filter 1 and the forward characteristic filter 3 are digital filters,
For example, it is realized by a DSP, but it may be realized by an analog filter. When a digital filter is used, the filter characteristics may be realized after oversampling as in the digital soft clamp unit 2. In this case, the oversampling unit 4 may be further moved to the stage before the inverse characteristic filter 1 and the decimation unit 6 may be moved to the stage subsequent to the forward characteristic filter 6.

FIG. 2 is a characteristic diagram of an example of an input / output characteristic curve in the amplitude compression type non-linear processing section. The curve shown is
When x is an input signal, y is an output signal, and a is a negative real number, it is a function curve of y = ax ³ + x. This solid line portion is used as the input / output characteristic of the amplitude compression type nonlinear processing unit. . The input / output characteristic indicated by the solid line is an amplitude compression characteristic in which the absolute value of the output signal increases as the absolute value of the input signal increases, but the increase rate decreases. Any function consisting of an arbitrary number of odd power terms and the sum of input signals can be similarly used as the input / output characteristic of the amplitude compression type nonlinear processing unit.

This characteristic is, for example, that the soft
It is a characteristic that can simulate the input / output characteristics of a clamper, a limiter, a compressor, etc., and can also be called a saturation characteristic. The broken line portion of FIG. 2 is a range in which the absolute value of the output signal decreases as the absolute value of the input signal increases,
It is okay to use the function with a slightly wider range within the broken line.

FIG. 3 is a block diagram of an example of the amplitude compression type non-linear processing section. This block diagram is an example of the amplitude compression type nonlinear processing unit 5 of FIG. In the figure, 11 is a FIR low-pass filter, 12 is a third-order power multiplier, 13 is a fifth-order power multiplier, 14 is an amplifier, 15 is an amplifier, 16 is an adder, 1
Reference numeral 7 is an FIR low pass filter. First, the Nyquist frequency f _0/2 or less of the frequency band, or, FIR that passes only a frequency band lower than this input signal (Fi
By inputting the output signal of the oversampling unit 4 of FIG. 1 to the low-pass filter 11 of the Impulse Response, the generation of aliasing noise is prevented in advance. If the output signal of the oversampling unit 4 does not include the component of f _0/2 or more, the FIR
It is not necessary to provide the low pass filter 11.

The output of the FIR low-pass filter 11 is input to an adder 16, a third-order power multiplier 12, and a fifth-order power multiplier 13, respectively, and the third-order power multiplier 12 and the fifth-order power multiplier 13 are respectively desired by amplifiers 14 and 15. It is amplified with an amplification factor that realizes the characteristic of, and is added to the output of the FIR low-pass filter 11 in the adder 16. Note that the power multiplier may be only the third-order power multiplier 12, or a higher-order odd-order power multiplier may be connected between the FIR low-pass filter 11 and the adder 16. Third power multiplier 1
In 2,5-order power unit 13, since the harmonics are generated, the output of the adder 16, the Nyquist frequency f _0/2 or less frequency bands or passes only the frequency band lower than this input signal FIR By passing it through the low-pass filter 17, generation of aliasing noise is prevented. FI
The output of the R low pass filter 17 is output to the decimation unit 6 of FIG.

The FIR low-pass filters 11 and 17
Does not have to be the FIR type, but by using the FIR type, phase distortion can be eliminated and correct amplitude limiting operation can be performed. When the phase distortion occurs, the waveform may be deformed and a peak higher than the clip level which is the limit value of the amplitude level may occur.

FIG. 4 is a block diagram of a concrete example of the amplitude compression type non-linear processing section. This specific example is a more specific example of the amplitude compression type non-linear processing unit 5 in FIG. In the figure, the same parts as those in FIGS. Reference numerals 21 and 22 are multipliers, and 23 is a control unit. Only the third power 12 is used as the odd power, and the FI
When the output of the R low-pass filter 11 is x and the output of the adder 16 is y, y = ax ³ + x
It is assumed to have the input / output characteristics shown in FIG.

The third-order power multiplier 12 is realized by squaring the output x of the FIR low-pass filter 11 by the multiplier 21, and then multiplying the output x by the multiplier 22. 3
The output of the next exponentiator 12 is amplified by the amplifier 14, and the control unit 23 controls the amount of attenuation, whereby the coefficient a can be changed to adjust the amplitude compression characteristic. The amplifier 24 adjusts the gain. These signal processings can be executed by the DSP.

Although the embodiment of the frequency-dependent digital limiter of the present invention has been described above, as a concrete example of use of this limiter, an FM used for premastering of an optical video disk (hereinafter referred to as LD). The audio signal limiter will be described. First, the FM audio signal processing unit of the LD cutting system will be briefly described.

FIG. 5 shows the FM of the LD cutting system.
It is a schematic block diagram of an audio signal processing unit. In the figure, 31 is a CX noise reduction encoder, 32 is a pre-emphasis unit, 33 is an FM band limiter, 34 is an FM modulator,
Reference numeral 35 is an adder, and 36 is an LD cutting machine.

An audio signal from a master tape or the like (hereinafter,
The FM audio signal) is 2: 1 compressed in decibel with respect to a signal having a level higher than -28 dB regardless of frequency by the CX encoder 31 which is not directly related to the present invention, based on the CX noise reduction method. It Subsequently, the high frequency side is raised in the pre-emphasis unit 32, the maximum amplitude level is limited to a predetermined value in the FM band limiter 33, the FM modulation is performed in the FM modulator 34, and the FM modulation is performed in the adder 35. Video signal,
Then, the audio signal (hereinafter referred to as the EFM audio signal) is added to the EFM-encoded signal and supplied to the LD cutting machine 38 in the frequency division multiplex format to make the LD master master.

In the LD, digitally recorded E
Two types of audio signals are recorded: an FM audio signal and an FM audio signal that is recorded by analog FM modulation. Note that, in reality, both have left and right stereo channels, but they are omitted in the figure. On the LD reproducing device side, it is necessary to prevent a level difference from occurring when the decoded or demodulated audio signals are switched to each other. Therefore, at the time of recording, the EFM audio signal and the FM audio signal are
It is necessary to adjust the level.

On the other hand, in the market, the volume feeling at the time of reproduction affects the evaluation of the sound quality. Therefore, it is desirable that the EFM audio signal and the FM audio signal be encoded or FM-modulated at the highest level and recorded.

However, since the maximum value of the absolute value of the amplitude level is limited in the FM band limiter 33 so that the FM audio signal satisfies the band limit standard of the FM modulation signal in the LD, the limit level is set. There is a problem that the exceeded audio signal is hard clipped abruptly and becomes noise during reproduction depending on the spectrum of the musical tone signal. Moreover, in the pre-emphasis unit 32 in the preceding stage, the high-frequency emphasis of the audio signal is performed as the standard of the analog FM modulation signal in the LD. Therefore, the higher the frequency component of the audio signal is, the hard clip is generated in the FM band limiter 33. There is a problem that it easily occurs.

Therefore, at the stage of premastering for creating a sound source such as a master tape, the FM audio signal is passed through a dedicated FM audio signal limiter to create a master tape. By using this master tape,
Also in the LD cutting system of FIG. 5, the band limitation and pre-emphasis standards are satisfied, and the FM band limiter 33 does not generate a hard clip even when the audio signal becomes high level. In addition, this FM
It is also possible to connect the audio signal limiter directly to the input section of FIG.

FIG. 6 is a block diagram of the FM audio signal limiter. In the figure, 41 is a pre-emphasis section, 42a
42h is the first to eighth digital soft clamp parts,
43 is a de-emphasis unit, 44 is a double oversampling unit, 45 is a cubic power adder, and 46 is a 1/2 decimation unit. In this FM audio signal limiter, the higher the signal component on the high frequency side of the audio signal emphasized in the pre-emphasis unit 33 of FIG.
Amplitude compression processing is performed to reduce the gain to a maximum of 20 dB when a full level signal is input.

This FM audio signal limiter is based on the frequency-dependent limiter of FIG. 1 as a basic structure and is an embodiment of the frequency-dependent limiter of the present invention. FIG.
As the forward characteristic filter 3 of 3, the filter having the characteristic opposite to that of the pre-emphasis unit 33 in FIG. 5 is used. That is, since the characteristics are the same as those of the de-emphasis section provided on the analog FM demodulation side, this filter is referred to as a de-emphasis section 43. Further, since the inverse characteristic filter 1 of FIG. 1 has an inverse characteristic of the forward characteristic filter 3, it has a characteristic similar to that of the pre-emphasis unit 33 of FIG. 5, so this filter is referred to as a pre-emphasis unit 41. Note that the frequency characteristic of the de-emphasis unit 43 does not have to be completely opposite to the frequency characteristic of the pre-emphasis unit 33 of FIG.
The frequency characteristic of 3 may be substantially the same as the frequency characteristic of the pre-emphasis unit 33 in FIG.

In the concrete example of the FM audio signal limiter, the double oversampling unit 44 and the 1/2 decimation unit 46 are used as the oversampling unit 4 and the decimation unit 6 in FIG.
The third-order power adder 45 similar to that of the concrete example of FIG. 4 is used as the amplitude compression type non-linear processing unit 5, and the digital soft clamp unit has a multi-stage configuration, and the first to eighth digital soft clamp units are used. 42a to 42h are connected in cascade.

Of course, the digital soft clamp section 2
High-order power multiplier, high-magnification oversampling section,
It is also possible to form a single stage by using the decimation section. However, in the signal processing by the third-order polynomial like the third-order power adder 45, the gain reduction of about 2.5 dB is the limit, but when using the higher-order polynomial, in order to prevent the aliasing noise, High-order oversampling is required, which makes it difficult to implement with a DSP. In this embodiment, since the digital soft clamp unit has a multi-stage configuration, it is possible to realize a maximum gain reduction of 20 dB with double oversampling.

FIG. 7 is a characteristic diagram of the FM audio signal limiter shown in FIG. In the figure, the horizontal axis is frequency [Hz], the vertical axis is amplitude [dB], and the input level is 0 dB, −10 d.
The output level (solid line) and the total harmonic distortion THD (broken line) at B, -20 dB, and -30 dB are shown.
The measurement was performed by performing a sine wave sweep at each input level. 50 Hz for a small input signal of -20 dB
From 0 to 15kHz, the characteristics are flat, but 0dB
15kH from around 2kHz for large input signals
It can be seen that there is an attenuation of about 20 dB up to the vicinity of z, and the digital soft clamp section is operating.

[0037]

As is apparent from the above description, according to the first aspect of the invention, the inverse characteristic filter, the oversampling section for inputting the output of the inverse characteristic filter, and the output of the oversampling section are inputted. Since it has an amplitude compression type nonlinear processing unit, a decimation unit that inputs the output of the amplitude compression type nonlinear processing unit, and a forward characteristic filter that inputs the output of the decimation unit, the instantaneous amplitude according to the frequency characteristic of the forward characteristic filter is obtained. The compression characteristic is realized, the breathing phenomenon is eliminated, the sound quality deterioration is solved, and the uncertainties of the limiter operation are also solved.

Further, since the instant compression is realized in the digital domain, there is an effect that a remarkable S / N ratio can be secured as compared with the analog type. Among harmonic distortion and intermodulation distortion, it is possible to suppress the occurrence of aliasing noise for easily removing higher than the Nyquist frequency f _0/2 component.

According to the second aspect of the present invention, an inverse characteristic filter, a plurality of cascaded digital soft clamp sections for inputting the output of the inverse characteristic filter, and a cascaded plurality of stages of digital soft clamps. It has a forward characteristic filter for inputting the output of the unit, and the digital soft clamp unit inputs the output of the oversampling unit, the amplitude compression type nonlinear processing unit for inputting the output of the oversampling unit, and the output of the amplitude compression type nonlinear processing unit. Since it has the decimation section, the same effect as the invention described in claim 1 is obtained, and the gain reduction of the amplitude compression type non-linear processing section is small, and even if the order of oversampling is small, a large gain is obtained as a whole. The effect is that reduction can be realized.

According to the third aspect of the present invention, since the amplitude compression type non-linear processing section has means for adding the input and the odd power terms of the input, the width compression type non-linear processing is performed by a simple polynomial. There is an effect that it can be realized.

[Brief description of the drawings]

FIG. 1 is a block diagram of a first embodiment of the present invention.

FIG. 2 is a characteristic diagram of an example of an input / output characteristic curve in an amplitude compression type non-linear processing unit.

FIG. 3 is a block diagram of an example of an amplitude compression type non-linear processing unit.

FIG. 4 is a block diagram of a specific example of an amplitude compression type non-linear processing unit.

FIG. 5 is a schematic block diagram of an FM audio signal processing unit of the LD cutting system.

FIG. 6 is a block diagram of an FM audio signal limiter.

7 is a characteristic diagram of the FM audio signal limiter of FIG. 6;

FIG. 8 is a block diagram of a first conventional example.

FIG. 9 is a block diagram of a second conventional example.

FIG. 10 is a circuit diagram of a main part block of a second conventional example.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Inverse characteristic filter, 2 ... Digital soft clamp part, 3 ... Forward characteristic filter, 4 ... Oversampling part, 5 ... Amplitude compression type non-linear processing part, 6 ... Decimation part, 12 ... 3rd power multiplier, 13 ... 5th order Exponentiator, 14,
15 ... Amplifier, 16 ... Adder, 21, 22 ... Multiplier, 2
3 ... Control unit, 41 ... Pre-emphasis unit, 42a to 42
h ... 1st to 8th digital soft clamp parts, 43 ...
De-emphasis section, 44 ... Double oversampling section, 45 ... Third power adder, 46 ... 1/2 decimation section.

Claims (3)

[Claims] 1. An inverse characteristic filter, an oversampling section for inputting the output of the inverse characteristic filter, an amplitude compression type non-linear processing section for inputting the output of the oversampling section, and an output of the amplitude compression type non-linear processing section. A frequency-dependent digital limiter characterized by having a decimation unit for inputting and a forward characteristic filter for inputting the output of the decimation unit, and having an amplitude compression characteristic according to the frequency characteristic of the forward characteristic filter. 2. An inverse characteristic filter, a plurality of cascaded digital soft clamp sections for inputting an output of the inverse characteristic filter, and an order of inputting outputs of the cascaded plural stages of digital soft clamp sections. The digital soft clamp unit has a characteristic filter, and the digital soft clamp unit includes an oversampling unit, an amplitude compression type nonlinear processing unit for inputting an output of the oversampling unit, and a decimation unit for inputting an output of the amplitude compression type nonlinear processing unit. And a frequency-dependent digital limiter having an amplitude compression characteristic according to the frequency characteristic of the forward characteristic filter. 3. The frequency-dependent digital limiter according to claim 1, wherein the amplitude compression type non-linear processing section has means for adding an input and an odd power term of the input.