KR910007021B1 - Fklter coefficient arithmetic method and device - Google Patents

Abstract

No content.

Description

Filter coefficient calculating device and calculation method of filter coefficient

1 is a graph showing frequency characteristics obtained by Hilbert transforming desired input frequency characteristics.

2 is a block diagram showing a filter coefficient calculating device according to a first embodiment of the present invention.

3 is a graph illustrating division of desired frequency characteristics.

4 is a graph showing sampling frequency transformation by zero insertion and band filter.

5 is a graph illustrating the characteristics of the filter of FIG.

6 is a graph illustrating the addition of the output of the filter of FIG.

7 is a block diagram showing a configuration of a FIR filter that can use the first embodiment.

8 is a block diagram showing a filter coefficient calculating device according to a second embodiment of the present invention.

9 is a block diagram showing an example of an FIR filter that can use the second embodiment of the present invention.

10 is a block diagram showing a filter coefficient calculating device according to a third embodiment of the present invention.

11 is a block diagram showing a filter coefficient calculating device according to a fourth embodiment of the present invention.

12 is a block diagram showing a filter coefficient calculating device according to a fifth embodiment of the present invention.

FIG. 13 is a block diagram modified from the fifth embodiment of FIG.

14 is a block diagram showing the configuration of a filter coefficient calculating device according to a sixth embodiment of the present invention.

15 is a graph for explaining sampling points of amplitude frequency characteristics.

Fig. 16 is a graph illustrating amplitude frequency characteristics and sampling points for obtaining the transfer function of the sixth embodiment.

17 is a flowchart showing a filter coefficient calculation algorithm.

18 is a block diagram showing a filter coefficient calculating device according to a seventh embodiment of the present invention.

19 is a graph for explaining a method for determining amplitude frequency characteristics.

20 is a block diagram illustrating a system with two FIR filters.

21 is a block diagram showing a filter coefficient calculation device according to an eighth embodiment of the present invention.

22 is a block diagram showing a filter coefficient calculating device according to a ninth embodiment of the present invention.

23 is a block diagram showing a filter coefficient calculating device according to a tenth embodiment of the present invention.

24 is a graph showing input characteristics and divided frequency characteristics in the tenth embodiment.

25 is a graph showing different input frequency characteristics.

Fig. 26 is a graph for explaining correction of divided frequency characteristics.

27 is a graph showing input frequency characteristics and split frequency characteristics according to the eleventh embodiment of the present invention.

28 is a block diagram showing a filter coefficient calculating device according to a twelfth embodiment of the present invention.

29 is a graph for explaining a display portion of a twelfth embodiment of the present invention.

30 is a perspective view showing a display portion of the twelfth embodiment;

Fig. 31 is a flowchart showing a method of calculating and setting a filter coefficient according to the present invention.

32 is a graph for explaining the generation state of amplitude frequency characteristics.

33 is a flowchart illustrating a filter coefficient calculation algorithm.

34 is a block diagram showing a filter coefficient calculating device.

35 is a blow chart showing filter coefficient calculation algorithm.

* Explanation of symbols for main parts of the drawings

1: Input means (circuit) 2: Division and extraction means (circuit)

3: Hilbert transform means 5, 34: inverse Fourier transform means (circuit)

6: interpolation means (circuit) 8: housework means (circuit)

9: transmission means (circuit) 20: linear phase transformation means (circuit)

22, 53: calculation circuit 31: amplitude input circuit

32: correction amplitude input circuit 33: phase calculation circuit

35: setting circuit 39: phase input circuit

41: group delay input circuit 42: integral circuit

56: setting switch 57: display unit

710: low pass filter 702: mid pass filter

703: high pass filter

According to the present invention, by integrating a finite number of coefficients (hereinafter referred to as FIR coefficient: FINITE IMPULSE RESPONES FACTOR) and a delayed signal, it is set to a transversal filter or a finite impulse response filter (hereinafter referred to as an FIR filter). A filter coefficient calculating device and a filter coefficient calculation setting method for obtaining a FIR coefficient that can realize a frequency characteristic of

FIR filters are used in various systems such as sound quality control devices for the purpose of arbitrarily obtaining frequency characteristics. The desired frequency characteristic realized by this is generally expressed as an amplitude frequency characteristic (power spectrum), which does not include phase information. Therefore, it is not possible to convert this to inverse puuri as it is to obtain the time function FIR coefficient. Therefore, the Hilbert transform is used as a known method for obtaining phase information from the powder spectrum. The Hilbert transform is used to obtain FIR coefficients that achieve desired amplitude frequency characteristics that do not interfere practically (eg, Reference 1: Acoustics Society, Electroacoustic Research Society (1985 EA85-44)).

This method is briefly described below with reference to FIG.

The Hilbert transform is a method of converting a given real function or an imaginary function into a plural function (Reference 2: Maedawa O Ohsa, the basis of digital signal processing). The method specified in Reference 1 assumes the desired frequency characteristic, i.e., the amplitude frequency characteristic indicated by the solid line 11 in FIG. 1A as a real function, and obtains a plurality of functions therefrom. In Fig. 1b, the real function is indicated by a solid line and the imaginary function is indicated by a broken line. If the amplitude frequency characteristic is found inversely from the plural functions obtained by the Hilbert transform, the deviation from the desired amplitude frequency characteristic is obtained. In Fig. 1c, after the Hilbert transform, the frequency characteristic is indicated by a solid line, and the desired frequency characteristic is indicated by a broken line. In order to reduce the autonomy, the desired frequency characteristic is increased by a certain ratio for each frequency, and the Hilbert transform is repeated again, and the sequential comparison is repeated several times to obtain a plural function that realizes the desired frequency characteristic to the extent that there is no practical problem. Next, the FIR coefficients are obtained by converting this to inverse puruli.

According to References 1 and 2, when the Hilbert transform is considered as a discrete time system, it is represented by the equations (1) and (2).

In formulas (1) and (2), H (K) is a plural function to be determined, P (m) is a real function, and U _N (Km) is represented by formula (2). In the means for performing the Hilbert transform, the value obtained by the expression (2) can be obtained in advance and stored as a data table. Since P (m) changes in Equation (1), an integration operation means is required to realize Equation (1), and at least N ^{2 integration} operations are required to perform Hilbert transformation of N points. It is difficult to reduce the time and the device size because N ² times of computations determine the conversion time of the Hilbert transform and the scale of the conversion device, and the Hilbert transform operation time and the device scale are determined according to the frequency band and the frequency resolution. .

In addition, Reference 2 described above specifies a method for obtaining phase information from a powder spectrum by obtaining a FIR coefficient that can realize desired frequency characteristics by performing linear phase transformation.

According to Ref. 2, if a phase of a discrete time system is considered in terms of a prior phase transformation, the transformation equation can be expressed as follows.

However, H (r) and Hi (k) are obtained real functions and imaginary functions, and A (k) represents amplitude frequency characteristics.

Even in the formulas (3) and (4), COS [-(N-1) / N × πk] and SIN [-(N-1 / N × πk] can be obtained as table data beforehand, At least N × 2 multiplications are required to perform the linear phase transformation, and N × 2 multiplications determine the conversion time of the linear phase transformation, and once the frequency band and frequency resolution are determined, the linear phase transformation operation time is determined by itself. There was a problem that this could be shortened.

The present invention has been developed to eliminate the above drawbacks according to the prior art.

Accordingly, it is an object of the present invention to provide a method for calculating a filter coefficient that can shorten an elongation time and provide a novel or advanced device with a simple structure.

In the present invention, the input means for inputting the desired frequency characteristics, the dividing means coupled to the input means for dividing the input frequency characteristics into a plurality of frequency bands, and to realize the frequency characteristics of each divided frequency band A filter coefficient calculating device for a transverse filter having a calculating means coupled to the dividing means for obtaining a filter coefficient is provided.

In addition, in the present invention, a first step of inputting an amplitude frequency characteristic having a first sampling frequency or a second sampling frequency into an input circuit, and using a relationship between the Hilbert transform based on the input amplitude frequency characteristic, Obtaining a transfer function by combining a sampling frequency of a species and a sampling frequency of a second species; and when the amplitude frequency characteristic of the input circuit is set to the first sampling frequency, the transfer at the second sampling frequency. If the impulse response is obtained by inverse Fourier transforming, and the amplitude frequency characteristic is set to the second type of sampling frequency in the input circuit, the transfer function is converted to the inverse Fourier at the first type of sampling frequency to obtain an impulse response. Filter coefficient calculation setting consisting of a step of obtaining and a step of setting the real part of the obtained impulse response to an external transverse filter as a filter coefficient It provides a way.

In addition, the present invention provides a method for inputting an amplitude frequency characteristic having a first sampling frequency or a second sampling frequency to an input circuit, a step of obtaining a transfer function based on the input frequency characteristic, and the input circuit. In the case where the amplitude frequency characteristic is set to the first type sampling frequency, the impulse response is obtained by converting the transfer function to the inverse Fourier at the second sampling frequency, and the amplitude frequency characteristic is sampled in the input circuit. In the case of setting the frequency, converting the transfer function to the inverse purifier at the first sampling frequency to obtain an impulse response, and setting the real part of the obtained impulse response to an external transversal filter as a filter coefficient. It is provided in the method of setting the filter coefficient calculation.

Embodiments of the present invention will be described below with reference to the drawings.

2 is a block diagram showing an FIR coefficient calculating device used in the FIR filter according to the first embodiment of the present invention. A microcomputer consisting of a central processing unit, a memory, and the like can be provided to configure an FIR coefficient calculating device. In Fig. 2, reference numeral 1 denotes input means for inputting a desired frequency characteristic to be realized by the FIR filter, and reference numeral 2 denotes a characteristic of a divided band for dividing the input frequency characteristic into a plurality of bands. Splitting and extracting means for extracting a point input from the input means (1) in order to display. In addition, the desired frequency characteristic of FIG. 3A is shown by the discrete distribution point on a curve, and the space | interval of the frequency is determined according to the frequency resolution of the low range of the desired frequency characteristic. 3b, 3c and 3d show three bands divided by the dividing and extracting means 2 of FIG. In FIG. 3, the horizontal axis is the frequency axis, and f _N1 , f _N2 , and f _N shown on the frequency axis are the highest frequencies in the divided bands, respectively, and are referred to as Nyquist frequencies in digital signal theory. 1/2 of the sampling frequency.

In Fig. 2, reference numeral 3 denotes Hilbert transform means for converting Hilbert into bands shown in Figs. 3b, 3c, and 3d, and reference numeral 4 denotes such that the frequency characteristic after Hilbert transform has no problem with the desired frequency characteristic. It is judged whether or not there is a coincidence, and when it does not match, the frequency characteristic evaluation and correction means input to the Hilbert converting means 3 so as to perform correction and again Hilbert transform. These means (3) and (4) are in accordance with the technique described in Ref.

(5) is an inverse Fourier transforming means for converting to inverse Fourier using the real and imaginary functions obtained by the frequency characteristic evaluation and correction means 4 to obtain the FIR coefficient for realizing the desired frequency characteristic. (6) is an interpolation means for performing interpolation between sample points. 4 is a graph illustrating an example of interpolation. As shown in FIG. 4A, the point represented by X lies between the values (indicated by 0) obtained by the inverse Fourier transform on the time axis. As shown in Fig. 4B, the X point on the time axis is interpolated and inserted into a value between the values (denoted by 0) obtained by the inverse Fourier transform. The FIR coefficients obtained by the interpolation means 6 are supplied to the low pass filter 701, the middle heat filter 702, and the high pass filter 703. In this case, the sampling frequency is tripled. 5, characteristics of the band pass filters 701, 702, and 703 are described.

(8) is FIR coefficient adding means for adding FIR coefficients obtained for each band. 6A, 6B, and 6C are pass filters 701, 702, and 703 of each band, which are means for sampling frequency-converted means, FIR coefficients for realizing desired frequency characteristics in each band obtained by passing a waveform; 6D adds all of these FIR coefficients, and the FIR coefficient which realizes the desired frequency characteristic of all the bands is calculated | required.

(9) shows FIR coefficient transmission means for transmitting the FIR coefficient obtained for the entire band to the FIR filter.

An example of the FIR filter using this embodiment is shown in FIG. In Fig. 7, reference numeral 10 denotes digital signal input means for inputting a digital signal, numeral 11 digital signal memory and delay means for storing and delaying an input signal for realizing a weak filter. For example, FIR coefficient holding means for holding the FIR coefficient transmitted through the coefficient transmission line 121, 13 is an accumulating means including a multiplication means 131 and an addition means 132, 14 is an FIR filter Digital signal output means for outputting the processed digital signal.

8 is a block diagram showing a filter calculation device according to a second embodiment of the present invention. The same reference numerals are used for the same means as in FIG. 1, and description thereof is omitted for simplicity. The feature of the second embodiment is that the FIR coefficients obtained for each band are transmitted directly to the FIR filter.

An example of an FIR filter using the second embodiment is shown in FIG. In the FIR filter of FIG. 9, the digital signal input by the digital input means 10 is supplied to the low pass filter 151, the mid pass filter 152, and the high pass filter 153 which respectively transmit predetermined bands. Then, the sampling means 161, 162 extracts sampling points so that the upper limit of each band becomes a sampling frequency that becomes the Nyquist frequency. Here, each FIR coefficient holding means 12 interpolates sampling points by interpolation means 171, 172 transmitted from each transmission means 9 in FIG. After the digital signal interpolated to the sampling point passes through the band filters 151 ', 152' and 153 'corresponding to each of the band filters 151, 152 and 153, the digital signal adding means 18 To reach. The added digital signal is output from the digital signal output means 14.

As described above, in order to display the characteristics of each band, the excessive FIR coefficient of each band is taken in. Next, since the number of data is not required for the low pass signal and the mid pass signal in the same manner as in the first embodiment, it is possible to perform the small-scale operation of the redundancy calculation shown in Eqs. (1) and (2) in a short time. .

FIG. 10 shows a third embodiment of the filter coefficient calculating device of the present invention, and the desired filter coefficient is obtained by linear phase deformation.

In FIG. 10, the same means as in the first embodiment shown in FIG. 2 are denoted by the same reference numerals, and description thereof is omitted.

In Fig. 10, reference numeral 20 denotes linear phase conversion means for performing band-linear conversion of each band shown in Figs. 3b, 3c, and 3d to obtain the real and imaginary functions. In order to find the FIR coefficient for realizing the desired frequency characteristic, the inverse Fourier transform means 4 converts the inverse Fourier using the real and imaginary functions obtained by the linear phase changing means 20. Sampling interpolation is performed in each interpolation means 6 and then delayed in each delay means 21 to center the linear phase coefficients in each band.

In addition, as an example of the FIR filter used in a present Example, the thing shown in FIG. 7 is used similarly to a 1st Example.

In addition, the fourth embodiment shown in FIG. 11 is configured to obtain the FIR coefficients of the respective bands and transmit them directly to the FIR filter.

One example of the FIR filter used in this embodiment is that shown in FIG.

As described above, since an excessive number of data is not required to display the characteristics of each band, arithmetic processing can be performed on a small circuit scale in a short time.

FIG. 12 shows a fifth embodiment of the present invention in which a desired filter coefficient is obtained by Hilbert transform. The difference from the third embodiment is that instead of the linear phase converting means 20, a calculation circuit 22 is used which finds a transfer function using a Hilbert transform relationship from each of the bands shown in Figs. 3b, 3c, and 3d. Is the point.

In the present embodiment, the FIR coefficient for realizing the desired amplitude frequency characteristics is obtained by converting the transfer function obtained from the calculation circuit 22 into the inverse purge in the inverse purge. It is almost as effective.

In addition, FIG. 13 shows a sixth embodiment of the present invention, in which FIR coefficients of respective bands can be directly transmitted to the FIR filter.

Next, a sixth embodiment of the present invention will be described with reference to FIG.

In Fig. 14, reference numeral 26 denotes an input circuit for inputting an arbitrary amplitude frequency characteristic for each sampling frequency of the second type, and reference numeral 27 denotes a first type under a linear phase condition based on the input amplitude frequency characteristic. A transfer function calculation circuit for obtaining a transfer function using a sample and a frequency, 28 is a conversion circuit for converting the transfer function obtained from the transfer function calculation circuit 27 to a reverse type of a first type sampling frequency, and 29 A setting circuit for setting the real part of the impulse response obtained by the inverse purge conversion circuit 28 as a filter coefficient, and 30 is an FIR filter which realizes the actually set amplitude frequency characteristic. In the input circuit 26, the desired amplitude frequency characteristic | H (w) | is set for each sampling frequency of the second type (see Fig. 15b), and in the transfer function calculating circuit 27, the following operation is performed. By doing so, the transfer function H (w) is obtained.

Here, w in the input amplitude frequency characteristic H (w) |

Is the sampling frequency of the second kind represented by " w " in cos, sin in the formulas (6) and (7),

It should be the first sampling frequency, denoted by.

In this case, when the sampling point N is an even number, H (w) is obtained using equations (6) and (7) until K = N / 2, and K in cos and sin from K = N / 2 + 1. The calculation is performed as K + 1. However, when N is large, the result is the same even if such an operation is not performed. Figure 16 shows the shape of this calculation. Fig. 16A shows the amplitude frequency characteristics set to the sampling frequencies of the second kind, and Fig. 16B shows the sampling frequencies of the first kind in cos and sins of Eqs. (6) and (7) on the unit circle of multiple planes. The correspondence between the values to be multiplied is indicated by an arrow.

From the transfer function H (w) obtained in this way, the inverse purge is transformed to the inverse purge of equation (10) using the first type of sampling frequency in the arithmetic circuit 28, and an impulse response to H (w) is obtained. Save

n

N-1, H(w)는 (5),(6),(7)식에 의해 구해진 값.)(0

n

N-1 and H (w) are values obtained by the formulas (5), (6) and (7).)

Here, again, the w of the complex function e ^{jwn in} the expression (10) is a value when the first type of sampling represented by the expression (9) is performed.

Next, only the real part of the impulse response h (n) obtained by the setting circuit 29 is taken, and this value is set in the FIR filter 30 as a filter coefficient. As a result, without using a window function, any amplitude frequency characteristic can be accurately realized without ripple.

Fig. 17 is a flowchart showing the filter coefficient calculation algorithm of the filter coefficient calculating device.

Next, a seventh embodiment will be described with reference to FIG.

In Fig. 18, reference numeral 31 denotes an amplitude input circuit for inputting desired amplitude frequency characteristics, and reference numeral 32 denotes an amplitude frequency for an object to be subjected to phase correction (e.g., correcting an inverse amplitude frequency characteristic such as a speaker). A resonant frequency input circuit for inputting an amplitude frequency characteristic for an object subjected to phase correction or phase correction, and automatically generating the reverse characteristic, (33) is an amplitude frequency obtained from the correction characteristic input circuit (32). A phase calculating circuit for converting the characteristic into a phase frequency characteristic from the relation of the Hilbert transform, and (34) transmits the amplitude frequency characteristic obtained from the amplitude input circuit 31 and the phase frequency polarity obtained from the phase calculating circuit 33. An inverse Fourier transform circuit that converts a function to an inverse Fourier, 35 is a set circuit that sets the impulse response obtained by the inverse Fourier transform circuit 34 as a filter coefficient, and 36 is actually set It is a FIR filter to realize the frequency characteristics.

In the amplitude input circuit 31, the desired amplitude frequency characteristic | H (w) | of w = 0 to π is set, and the amplitude frequency characteristic of w = π to 2π is automatically changed while resetting this input amplitude frequency characteristic. Generated (see also FIG. 19).

Next, the inverse amplitude frequency characteristic | H _I (w) | is input to the correction characteristic input circuit 32 about the object to which phase correction (for example, the phase frequency characteristic of a speaker etc. is corrected). This amplitude frequency characteristic is also inputted with an amplitude frequency characteristic of w = 0 to π, and the amplitude frequency characteristic of w = π to 2π is generated by repeating the input amplitude frequency characteristic.

In the phase calculating circuit 33, the amplitude frequency characteristic is converted into the phase frequency characteristic as follows using the relation of the Hilbert transform.

Hilbert transform is a method of obtaining a plural function based on a real function or an imaginary function originally assigned.

The periodic function h (n) of the period N (hereinafter, the sequence indicates the inter-disc system) is represented by the sum of the even function h _e (n) and the odd function h _o (n), which are periodic functions.

In addition, if the periodic causality rate is defined as

As, _e to h (n) it can be determined for h (n). Here, the properties of h _e (n) and h _o (n) show that:

The water coefficient for the puuri of h (n)

In other words, the water supply coefficient of H _e (n) is H _R (K), and the water supply coefficient of h _o (n) is jH ₁ (K).

From the above results, assuming that the sequence above is a finite period function of finite period N, this is the result of inversely transforming (usually called IDFT) of H _R (k) (k = 0 to N-1). Is assumed to be h _e (n).

From the above properties, the following can be seen.

Transfer function H '(k) to realize desired amplitude frequency characteristics | H' (K)

It becomes the following equation.

(However, the phase frequency characteristic is called θ (w)).

If we take the natural logarithm of both sides of (15), we get

Here, suppose that the result of transforming the inverse puuri of ιnH '(K) is h' (n), and the inverse purge transformation of ιn | H '(K) | is h _e ' (n). From the formula

It can be seen that.

In addition, when h '(n) obtained from Eq. (17) is converted into a fuuri, the sprinkling part of the obtained value corresponds to ιn | H' (K) | and the imaginary part corresponds to jθ (w).

Therefore, in summary, the phase frequency characteristics are obtained by operating the amplitude frequency characteristics as follows.

(1) by taking the natural logarithm of the amplitude frequency characteristics, (2) converting the value obtained in (1) to inverse puuri, and (3) calculating the equation (17) from the value obtained in (2) above. h '(n) is obtained, (4) h' (n) is converted into inverse puuri, and (5) h '(n) is converted into inverse puuri.

The phase frequency characteristics thus obtained operate so that the phase transition is minimized (phase frequency characteristics such as an enclosed sputter system indicate such characteristics).

If the amplitude frequency characteristic | H _I (w) | is calculated | required above, the phase frequency characteristic with respect to the object which has a binary amplitude frequency characteristic can be calculated | required according to the relationship of Hilbert transform.

Therefore, when a phase frequency characteristic having a binary amplitude frequency characteristic is obtained on the basis of the inverse amplitude frequency characteristic for the object to be subjected to phase correction, this becomes a phase frequency characteristic for performing phase correction. In other words, the inverse phase characteristic for the object to be subjected to phase correction can be obtained.

Next, the inverse Fourier transform circuit 34 converts the transfer function H (w) consisting of the amplitude frequency characteristic obtained from the amplitude input circuit 31 and the phase frequency characteristic calculated from the phase calculation circuit 33 into an inverse Fourier transform. By the impulse response method h (n) is obtained.

This method can obtain h (n) more accurately and simply than the method described in Reference 1.

The amplitude frequency characteristic of the impulse response thus obtained represents the amplitude frequency characteristic obtained from the amplitude input circuit 31, and the phase frequency characteristic of the impulse response represents the phase frequency characteristic calculated by the phase calculating circuit 33. This is described in detail as follows.

Assume a system with two FIR filters as shown in FIG. (37) inputs an arbitrary amplitude frequency characteristic, obtains the phase frequency characteristic θ11 (w) from the relation of the Hilbert transform in the binary amplitude frequency characteristic, and converts the transfer function having this frequency characteristic to the inverse fuuri The filter coefficients FIR filters A and 38 are the amplitude frequency characteristics obtained by adding the desired amplitude frequency characteristics | H ₁₂ (w) | and the inverse amplitude frequency characteristics of the amplitude frequency characteristics inputted by the FIR filter A (37). Based on the logarithmic amplitude frequency characteristics, the transfer function obtained under linear phase conditions (the amplitude frequency characteristic itself may be the transfer function, i.e., the phase term is zero). Is a FIR filter B that converts the result to inverse Fourier and obtains the value obtained as the FIR coefficient. In this configuration, the amplitude frequency characteristic of the entire system is the amplitude frequency characteristic | H ₁₂ (w) | input from the FIR filter B 38, and the phase amplitude frequency characteristic is θ ₁₁ (w). In practice, changes in the linear phase of the FIR filter B 38 are added, but when the phase correction is taken into account, their influence is ignored. That is, when considering the above phase, the group delay characteristic obtained by differentiating the phase frequency characteristic by the frequency is important, so that the change in the linear phase can be ignored. Therefore, on the contrary, for the whole system, a system composed of one FIR filter having a phase frequency characteristic obtained according to the relationship between an arbitrary amplitude frequency characteristic and a Hilbert transform includes one FIR filter having a linear phase and a Hilbert transform ( Corresponds to one FIR filter with a minimum phase shift system.

In this way, the impulse response h (n) obtained by the circuit 34 changed into the inverse purge has an arbitrary amplitude frequency characteristic and a frequency characteristic for performing correction.

Next, in the setting circuit 35, the impulse response h (n) obtained as described above is set as the filter coefficient to the FIR filter 36 which realizes the actually inputted frequency characteristic.

Incidentally, in the present embodiment, the phase correction is described as an example, but the phase frequency characteristic for an object having a certain amplitude frequency characteristic can be realized.

21 is a block diagram showing an eighth embodiment of the present invention. Numeral 39 denotes a phase characteristic input circuit for inputting arbitrary phase frequency characteristics. In Fig. 21, those having the same functions as those shown in Fig. 18 are denoted by the same numbers.

In the phase characteristic input circuit 39, arbitrary phase characteristics are input in the range of w = 0 to π. The characteristic of w = (pi) -2 (pi) produces | generates repeatedly changing the sign of the integer of a phase characteristic in the range of w = 0- (pi). This means that when the impulse response to be obtained is a real number, the amplitude frequency characteristic is an even function, and the phase frequency characteristic is an odd function.

The operation is similar to that of the seventh embodiment, and only the difference is calculated from the amplitude frequency characteristic for phase correction to which the phase frequency characteristic is given or directly inputted. However, in the present embodiment, the phase frequency characteristic is based on the minimum abnormality transition and the arbitrary phase frequency characteristic. As indicated in the seventh embodiment, the amplitude frequency characteristic to be realized is an arbitrary amplitude frequency characteristic input, and the phase frequency characteristic is a value obtained according to the relationship of the Hilbert transform. By converting this value to the inverse puruli to obtain the filter coefficient, the phase and amplitude could be set independently. From this, it can be seen that the phase frequency characteristics can be set to arbitrary characteristics.

In order to explain this, the part which calculates phase-frequency characteristic using the Hilbert transform relation of 7th Example is considered.

In the seventh embodiment, the phase frequency characteristic is obtained from the inverse amplitude frequency characteristic of the object to be subjected to phase correction. On the contrary, when setting the phase frequency characteristics, it is assumed that corresponding amplitude frequency characteristics can be obtained. In other words, when the obtained amplitude frequency characteristic is inputted, the set phase frequency characteristic is obtained from the Hilbert transform relationship.

Therefore, it is considered that an amplitude frequency characteristic corresponding to an arbitrary phase frequency characteristic exists.

In addition, explanation will be given with reference to FIG. 20 shown first. Numeral 37 denotes an FIR filter in which a filter coefficient corresponding to a transfer function obtained from amplitude frequency characteristics having arbitrary phase frequency characteristics set is set, and reference numeral 38 denotes an arbitrary amplitude frequency characteristic and FIR filter set. The inverse frequency frequency characteristic of the amplitude frequency characteristic obtained from the phase frequency characteristic of A (37) is added as the amplitude frequency characteristic, and the filter coefficient obtained from the transfer function obtained under the linear phase condition is called FIR filter B. .

With such a configuration, as in the seventh embodiment, the overall amplitude frequency characteristic is assumed to be the arbitrary amplitude frequency characteristic set in the FIR filter B 38, and the phase frequency characteristic is set as the phase frequency characteristic set in the FIR filter 37. Assume

In this way, it can be seen that the amplitude frequency characteristics and the phase frequency characteristics can be set independently by the configuration as shown in the eighth embodiment.

FIG. 22 is a block diagram showing the ninth embodiment of the present invention, in which those having the same functions as those shown in FIG. (41) is a group delay input circuit for inputting any group delay characteristic or group delay distortion characteristic, (42) is an integral for calculating the phase frequency characteristic by integrating the group delay characteristic inputted by the group delay input circuit 41; Circuit, the military zone characteristic γ (w) is

Is defined as Integrating this by the frequency w yields the phase frequency characteristic θ (w). This means that the phase frequency characteristic can be obtained based on arbitrary group delay characteristics. Therefore, as in the seventh and eighth embodiments, the amplitude frequency characteristics and the group delay characteristics (phase frequency characteristics) can be set independently.

In the seventh, eighth, and ninth embodiments, the impulse response obtained by the inverse Fourier transform circuit 34 is used as the filter coefficient for the FIR filter 36 as it is. Ripple may occur in the frequency characteristics realized by Therefore, in order to prevent the occurrence of ripple, the impulse response obtained by the inverse Fourier transform circuit 34 is multiplied by a window function having a hanning or hamming characteristic, and the result of the multiplication by the setting circuit 35 is multiplied by the coefficient of the FIR filter 36. You can do

Therefore, the setting circuit 35 can be configured to have such a window.

Next, a tenth embodiment of the present invention will be described with reference to FIG.

In Fig. 23, reference numeral 51 denotes an input circuit for dividing an arbitrary amplitude frequency characteristic into frequency bands corresponding to two sampling frequencies, and 52 denotes an input amplitude frequency characteristic corresponding to two different sampling frequencies. A division circuit for dividing into frequency bands, 53 is an arithmetic circuit for obtaining filter coefficients of the FIR filter corresponding to the frequency band of each sampling frequency, and 54 and 55 are FIRs having different sampling frequencies and the same number of filter coefficients. Filter.

The operation of this embodiment will be described below.

In the input circuit 51, an arbitrary amplitude frequency characteristic is divided into frequency bands corresponding to two sampling frequencies, and in the portion where the frequency bands overlap, the amplitude frequency characteristic is input by a small frequency resolution, and the frequency bands do not overlap. In the part, amplitude frequency characteristics are inputted by large frequency resolution. 24A shows the amplitude frequency characteristic input by the input circuit 51. FIG. The circle in Fig. 24a (○) table, and displays the predetermined amplitude value, f _1, f _2, is designated by two different sampling frequency, it becomes twice of f _2, f _1. The frequency division capability of the amplitude frequency characteristic input from the input circuit 51 is divided into this f _1/2 . At this time, if the frequency resolution of the frequency band up to f _1/2 that can input the amplitude frequency characteristics by the input circuit 51 is N, the number of filter coefficients of the FIR filter 54 corresponding to the frequency band is N. f is the ₁ / N, the frequency resolution of the amplitude frequency characteristic that can be entered in the band f _2/2 from the f _1/2 is a f ₂ / N. Here, when the frequency near the cutoff frequency of the low pass filter of the FIR filter 54 which actually processes the signal by the sampling frequency f ₁ is f ₃ , the settable amplitude frequency from f ₃ to f _1/2 is set. The frequency resolution of the characteristic is set to f ₂ / N as shown in FIG. In addition, the present embodiment may use such an input method.

The amplitude frequency characteristic input as described above is input to the dividing circuit 51, and is divided into two frequency characteristics 26b and 26d corresponding to two sampling frequencies and the number of filter coefficients of the FIR filters 54 and 55. . (X table and △ table are its amplitude values). Hereinafter, the division method of the amplitude frequency characteristic in the division circuit 52 will be described.

First, referring to FIG. 25, the division method of the amplitude frequency characteristic is explained. In the bands 0 to f _1/2 corresponding to the sampling frequency f ₁ , the amplitude frequency characteristic from 0 to f ₃ is the frequency resolution f _1. Since this is input from the input circuit 51 by / N, this value is used as it is. What matters in the band corresponding to the sampling frequency f ₁ f ₃ ~ f ₁ /, but the band up to two, since the amplitude frequency property inputted to the input circuit by the frequency resolution of f _2/2 in the band and the amplitude frequency From the value of the property, we need to find a new amplitude frequency point with a frequency resolution of f ₁ / N. There are many ways to obtain this new amplitude frequency point, but here we describe how to use linear interpolation. That is inputted in the input circuit _{(51) f 3 ~ f 1} / is a value from whichever of the amplitude frequency f ₃ to 2 A, if the next value B, the original sampling frequency, between A and B f ₁ The value C for is because the sampling frequency f ₂ is twice the f ₁

Can be obtained as Other values up to f _1/2 can be obtained as well. In addition, although the amplitude frequency characteristics of the frequency band corresponding to the sampling frequency f ₁ were obtained by performing linear interpolation, as described above, various methods, such as a method using a multi-order function, can be used. have.

Moreover, if with reference to claim 26 also describes the resolution method of the amplitude frequency characteristic and the frequency f ₄ (> f ₃₎ from the frequency f ₁ / for up to 2, bojeongreul performed such that the amplitude frequency characteristic value of f _1/2 Spirit You can also consider how to get the value and use it. The value of f ₄ is a frequency at which the signal control level of the low pass filter described above is a sufficient value (for example, a frequency 10 dB away from the level of the pass band). When the amplitude value input by the input circuit 51 at frequency f ₄ is set to D, the value E by the new sampling frequency f ₁ is

Can be obtained as By doing in this way, it is possible to further restrict the signal of the limited band of the low pass filter mentioned above. The condition that the amplitude frequency characteristic is made zero at f _1/2 is also a condition necessary when the calculation coefficient of the calculation circuit 53 is obtained under the condition of the linear phase.

Similarly, the same effect is obtained when the same correction is performed on the band of the sampling frequency f ₂ based on the frequency f ₅ which does not affect the signal to be processed.

It will be described next in claim 24 degrees and an amplitude frequency characteristics in the frequency range 0 ~ f _2/2, which with reference to claim 25 corresponds to the sampling frequency f _2.

Frequency value between frequencies f ₃ and f _2/2 can be used as this value since input by the input from circuit 51, the frequency resolution f ₂ / N. Since the input amplitude frequency characteristics cannot be used as it is different from the frequency resolution for the values from frequency 0 to f ₃ , it is necessary to calculate the value of this band by performing calculation. This calculation method is also considered in various ways, but here the method of extracting the input amplitude frequency characteristic is indicated. That is, since f ₂ is twice as high as f ₁ , the amplitude frequency characteristics having a sampling frequency f ₂ and the filter coefficient N from frequencies 0 to f ₃ have two amplitude frequency characteristics inputted by the frequency resolution f ₁ / N. It can obtain | require by extracting one each time (refer FIG. 24C). By extracting in this manner, the amplitude frequency characteristic of the frequency band corresponding to the sampling frequency f ₂ can be obtained from the amplitude frequency characteristic inputted by the frequency resolution f ₁ / N in the input circuit 1. Here, the extraction was performed to obtain the amplitude frequency characteristics of the frequency band corresponding to the sampling frequency f _2. However, as described above, other methods, for example, amplitudes of 0 to f ₆ (<f ₃ ) shown in FIG. The frequency characteristic can be considered zero or 0dB (-1,0).

Next, in the calculation circuit 53, the filter coefficients for the FIR filters 54, 55 are obtained using the two frequency characteristics divided by the division circuit 52. Here, to obtain the amplitude frequency characteristic of the repetition of the respective frequency bands _{(f 1/2 ~ f 1} , f 2/2 ~ the band f _2), and the amplitude frequency characteristic in the condition of linear phase by mistake, wherein (項) For example, we can perform an operation on transforming the inverse Fourure with zero imaginary terms and calculating the filter coefficients. The inverse Fourier transform is performed so that the imaginary term is not zero, and the imaginary term indicating the phase frequency characteristic or the group delay characteristic is provided. In addition, a filter coefficient calculation method that performs the minimum phase shift in accordance with the Hilbert transform relationship can be used. In this way, the filter coefficient can be obtained to realize the input amplitude frequency characteristic (or amplitude frequency characteristic, phase frequency characteristic or group delay characteristic) for the frequency band corresponding to each sampling frequency. The filter coefficient thus obtained is input to the FIR filters 54 and 55 corresponding to the respective frequency bands, and the amplitude frequency characteristic actually input to the input circuit 51 is realized.

Further, in the embodiment, in the input circuit 51, sampling of the amplitude frequency characteristics of the first type (frequency points are set every f ₁ / N from frequency 0 and frequency points are set every f ₂ / N from f ₃ ) Or sampling the second type (frequency points are set every f ₁ / N from frequencies f ₁ / N x 1/2 and frequency points are set every f ₂ / N from f ₃ ). Frequency division can be performed to obtain a filter coefficient corresponding to the frequency. In this case, the staggering of the frequency points occurs due to the difference in the frequency bands, but by taking a method such as linear interpolation or polynomial function approximation, frequency division which excludes this can be performed, and the filter coefficients of each frequency band can be obtained. have. In addition, in the embodiment, the amplitude frequency characteristic may be input and other frequency characteristics such as a phase frequency characteristic or a group delay characteristic may be input.

27 is a graph showing an input method of amplitude frequency characteristics of an input circuit and a frequency division method of a division circuit according to an eleventh embodiment of the present invention. The configuration of the eleventh embodiment is the same as that of the tenth embodiment of FIG. 23, and only the input method and the division method are different.

The amplitude frequency characteristics (Fig. 27a) that can be input from the input circuit 51 are different from the sampling frequencies f ₁ and f ₂ that can be processed over the entire band (in this figure, of f ₁ / 2N). Can be set by frequency division capability. In the next division circuit 52, in order to obtain a frequency point capable of processing the actual signal, the input frequency point is _one of two input frequency points for the band from the input frequency point to the frequency corresponding to the sampling frequency of f ₁ . Arithmetic operations are performed to extract dogs (Fig. 27b), and for a frequency band corresponding to the subsample frequency of f ₂ , arithmetic operations are performed to extract one of four input frequency points (Fig. 27c). Split amplitude frequency characteristics.

Otherwise, the same processing as in the tenth embodiment is performed. As in the present embodiment, even in the input circuit 51, even if the frequency characteristic can be set by a frequency resolution different from the frequency resolution that can be processed, the division circuit 52 can perform a proper operation so that it can actually be processed. Since the frequency characteristic can be obtained, the filter coefficients for the FIR filters 54 and 55 can also be obtained.

In the present embodiment, the case where the first type of sampling is performed in the division circuit 52 is shown, but the case of the second type of sampling is also considered.

In the tenth and eleventh embodiments, when the first type of sampling is performed, it is usually necessary to multiply the filter coefficient obtained by the calculation circuit 53 by the window function, but the second type of sampling is performed. In this case, it is not necessary to multiply the filter coefficient by the window function.

A twelfth embodiment of the present invention will be described below with reference to FIG.

FIG. 28 shows a block diagram of a filter coefficient calculating device for an FIR filter in a twelfth embodiment of the present invention.

In Fig. 28, reference numeral 50 denotes an input unit for selecting a frequency value and an amplitude value by moving it up, down, left, and right, 56 is a setting switch for setting a frequency value and an amplitude value, 57 is an amplitude frequency characteristic, A cross for indicating the position of the current frequency value and an amplitude value, and a display portion for displaying the frequency value and the amplitude value numerically, (58), an operation circuit for obtaining a filter coefficient based on the set amplitude frequency characteristic, (59) Is a transmission circuit for transmitting the obtained filter coefficient to an external FIR filter.

29 shows a display portion of the filter coefficient calculating device according to the present embodiment.

In FIG. 29, 60 is a graph indicating the horizontal axis as frequency and a new axis as amplitude, 61 is a set amplitude frequency characteristic, and 62 is a cross configured to move in response to the movement of the input unit 50. , (63) is a numerical representation of the frequency value and amplitude value corresponding to the position of the current cross 62.

A frequency value and an amplitude value are set and displayed in the position of the point A of the cross 62 in the 29th. The horizontal axis is logarithmic, and the vertical axis is displayed in decibels.

30 shows the configuration of the input unit 50 of the filter coefficient calculating device in the embodiment. 64 is a main body, 65 is a bead, 66 and 67 is a roller.

In FIG. 30, the beads are rotated by rotating the beads directly or by moving the bodies 64 with the main body 64 up or down. The rotation of the beads is transmitted to the rollers 66 and 67, and the rotations of the rollers 66 and 67 are electrically or mechanically converted, and the conversion is performed by the cross 62 of the display portion 57. It corresponds to movement. By moving the input unit 50 of FIG. 28, the cross 62 displayed on the display unit 57 moves in response to the movement, and the frequency value corresponding to the point A with a circle that flashes the center of the cross 62. The amplitude value is displayed numerically. The cross 62 is moved to a position corresponding to the frequency value and amplitude value to be set, and the setting switch 56 is pressed. For this reason, the desired frequency value and amplitude value corresponding to the position of the cross 62 are set. In the display unit 57, the preset frequency value and amplitude value are stored, and this value and the newly set value are connected by a straight line or a curve, and when the line is already displayed at the same value, the line is erased and the new line Is displayed and the value is set.

The amplitude frequency characteristic set as described above is set in the calculation circuit 58, and a complex function including phase information is calculated according to the Hilbert transform or the method based on the linear phase, and the FIR filter is converted to the inverse purge. The filter coefficient of is obtained. The filter coefficient obtained here is transmitted and set to the FIR filter externally by the transmission circuit 59.

In this embodiment, the filter is a FIR filter. However, even in a parametric type filter formed by an analog circuit, the filter coefficient obtained by the calculation circuit is only a parameter (frequency, 0, gain) of each filter. The operation is the same as in this embodiment. The point A may be displayed in a cross table or the like in addition to the dongle table.

Hereinafter, the calculation setting method of the filter coefficient concerning this invention is demonstrated concretely, referring FIGS. 31-33.

Fig. 31 is a block diagram of an apparatus for calculating filter coefficients using the calculation setting method of the present invention. In Fig. 31, when an arbitrary amplitude frequency characteristic is input for each of the second types of sampling frequencies by the input circuit, the first calculation circuit 72 takes the natural logarithm of the inputted amplitude frequency characteristics and samples the first type. Assuming that is performed, a characteristic of W = pi to 2 pi is generated. Also, the amplitude frequency characteristic is usually input having W = 0 to π.

The characteristics obtained by using the first type of sampling frequency in the first calculation circuit 72 are converted to the inverse Fourth by the first inverse Fourier transform circuit 73, and the second operation circuit 74 makes the first one. The value obtained by the conversion circuit 73 is set to he (n) in the inverse purge of 1, and the calculation of h '(n) = he (n) u (n) is performed. The value obtained by the second arithmetic circuit 74 is transformed to the puuri by the Fourier transform circuit 75 at a sampling frequency of the first type. In the third arithmetic circuit 76, using the phase term of the imaginary part of the value obtained by the Fourier transform circuit 75, H ' _R (K) = | H' (K | cos (θ (w ), H ' ₁ = | H' (K) | · sin (θ (w) is calculated by performing sampling of the second type, and the transfer function is obtained.The transfer function obtained by the third arithmetic circuit 76 The FIR filter converts the real part of the impulse response obtained by the second inverse Fourier transform circuit 77 to the first type of sampling frequency and obtained by the second inverse Fourier transform circuit 77 as a filter coefficient. It is set at 79 via a setting circuit 78.

Specifically, the desired amplitude frequency characteristic H (w) in the input circuit 71 is set for each sampling frequency of the second type (see FIG. 15B). The calculation circuit 72 calculates the following equation. If the number of sampling points is N and the value obtained by the calculation circuit 72 is H'e (K),

In the prior art, this calculation is

It was done by. FIG. 32A shows the relationship between H'e (K) and | H (w) | according to this embodiment, and FIG. 32B shows relationship between H'e (K) and | H (w) | according to the prior art. Indicates.

Next, in the first inverse Fourier transform circuit 73, the inverse Fourier transform of H'e (K) expressed by the following equation is performed at the first type of sampling frequency.

In the prior art, it is usually calculated as W = 2 · π / N · (K + 1/2).

Next, in the second calculation circuit 74, a value obtained according to Expression (21) is used.

Is calculated, and this value is transformed to the Fourier by the Fourier transform circuit 75. In other words,

Is calculated.

In the prior art, the portion is usually calculated as W = (2 · π / N) (K + 1/2).

Next, in the third calculation circuit 76, the following equation is calculated using the imaginary part H ' _I (K) of H' (K) obtained by the Fourier transform circuit 75.

only,

In this case, the calculation corresponding to the second type of sampling is performed.

Next, in the second inverse purge conversion circuit 77,

Convert to reverse puuri.

At this time, again using the sampling frequency of the first kind

n

N-1로서 계산이 행하여진다.However, W = 2 · π / N · K, 0

n

The calculation is performed as N-1.

Here, in the expression (27), W of the plural function e ^jwn is a value when the first kind of sampling is performed, and is not the second sampling frequency.

Next, the real part of the impulse response h (n) obtained by the setting circuit 78 is derived, and this value is set in the FIR filter 79 as a filter coefficient. H (n) set here is theoretically zero above N / 2, but in actual calculations, it is generally not zero due to an error. Therefore, when the output of the setting circuit 78 is equal to or above N / 2 of h (n), Configure to be zero.

As described above, by using the second type of sampling frequency, the portion that must be calculated using the first type of sampling frequency is accurately realized without any ripple.

33 is a flowchart illustrating the above-described operation of the filter coefficient calculating device.

In addition, although the case where the amplitude frequency characteristic is inputted as the sampling frequency of the 2nd type so far was demonstrated, when the amplitude frequency characteristic was inputted by the sampling frequency of the 1st type, each calculation displayed above is the 1st type, the 1st. The same effect is obtained by reversing the two sampling frequencies.

Hereinafter, the calculation setting method of the filter coefficient of this invention is demonstrated concretely, referring FIG. 34 and FIG.

34 is a block diagram of a filter coefficient calculating device using the filter coefficient calculating method according to the present invention. In FIG. 34, when an arbitrary amplitude frequency characteristic is input from the input circuit 81 for each sampling frequency of the second type, the inverse Fourier transform circuit 82 makes the input amplitude frequency characteristic a transfer function. Convert to inverse fuuri with one sampling frequency. In the setting circuit 83, the real part of the impulse response obtained by the inverse Fourier transform circuit 82 is set in the FIR filter 84 as a filter coefficient.

Specifically, in the input circuit 81, the desired amplitude frequency characteristic | H (w) | is provided for each sampling frequency of the second type (see FIG. 15B), and this amplitude frequency characteristic is transmitted. Let H (w) be.

It is done.

The transfer function H (w) thus obtained becomes a linear phase. This is understood from the following equation.

From the nature of the transformation

In the equations (31) and (32), the impulse response obtained under the condition that the amplitude frequency characteristic is the transfer function H (w) is given by H _R (w) = | H (w) | · cos (αw) and Hi (w) = Delay by m for the impulse response obtained from the transfer function determined by-(H (w) | .sin (αw). Therefore, it can be seen from this equation that the transfer function H (w) is linearly phased. Inverse Fourier transform circuit 82 uses the first type of sampling frequency based on this transfer function H (w) The inverse purge of equation (33) is transformed to obtain an impulse response corresponding to H (w).

nN-1, H(w)는 (28)식, (29)식, (30)식에 의해서 구한값)(0

n N-1 and H (w) are values obtained by the formulas (28), (29) and (30))

Here, the value is a case where the first kind of sampling represented by w = (2π / N) · k and k = 0 to (N-1) is ^performed in the complex function e ^jwn of the formula (31), and w = (2π / N) (k + 1/2), (k = 0 to (N-1)), which is not the second type of sampling frequency.

In this case, when the second type of sampling frequency is used, the window function is not multiplied, and a ripple occurs in the amplitude frequency characteristic realized.

Next, the setting circuit 83 derives the real part of the impulse response h (n) obtained by the inverse Fourier transform circuit 82, and this value is set in the FIR 84 as a filter coefficient, which is inputted here. Arbitrary amplitude frequency characteristics can be accurately realized without ripples.

35 is a flowchart illustrating the above operation performed in the above filter coefficient calculating apparatus.

Claims (31)

Input means (1) 51 for inputting a desired frequency characteristic, dividing means (2) 52 coupled to the input means for dividing the input frequency characteristic into a plurality of frequency bands (FIG. 5); And a calculating means (53), coupled to said dividing means, for obtaining filter coefficients for realizing the frequency characteristics of each divided frequency band. The filter coefficient calculating device according to claim 1, wherein the desired frequency characteristic is one of an amplitude frequency characteristic, a phase frequency characteristic, a group delay frequency characteristic, and a group delay distortion frequency characteristic. The filter coefficient calculating device according to claim 1, wherein said input means (1) (51) inputs frequency characteristics by frequency resolution corresponding to the number of said filter coefficients. The frequency portion corresponding to the plurality of different sampling frequencies has a frequency resolution smaller than the resolution corresponding to the number of filter coefficients for each frequency band of the input means (1, 51). Inputting a frequency characteristic, and said dividing means (2) (52) divides a frequency characteristic having a small frequency resolution and a frequency characteristic having a large frequency resolution. The frequency corresponding to the number of filter coefficients for each frequency band according to claim 1, wherein the input means (1) (51) is located near a boundary of a portion where frequency bands corresponding to a plurality of different sampling frequencies overlap. A frequency characteristic having a frequency resolution larger than the resolution is input, and the dividing means (2) (F2) is based on the input frequency characteristic having the large frequency resolution among the frequency characteristics having the small frequency resolution at which the frequency bands overlap. A Peter coefficient computing device characterized by dividing other frequency characteristics and frequency characteristics with large frequency resolution. 2. A value according to claim 1, wherein the input means (1) (51) inputs a frequency characteristic having a frequency resolution irrespective of the number of filter coefficients, and the input frequency characteristic has a frequency resolution corresponding to the number of filter coefficients. A filter coefficient calculating device for performing the calculation so that the. The first type of sampling according to claim 1, wherein the input means (1) 51 selects each frequency resolution (sampled frequency point every 15a degrees) determined from the frequency 0 based on the sampling frequency and the filter coefficient. Filter coefficient computing device, characterized in that for performing. The method of claim 1, wherein the input means (1) (51) is a frequency

A filter coefficient calculating device, characterized in that the second type of sampling is performed by selecting a sampling frequency point every 15b as determined based on the sampling frequency and the number of filters. 2. The apparatus according to claim 1, wherein the input means (1) (51) inputs frequency characteristics irrespective of sampling of the first and second types, and the dividing means (2) (52) is input by the input means. And a sampling point of a first kind or a second kind on the basis of the obtained frequency characteristic. The input means (1) (51) according to claim 1, wherein the input means (1) (51), when the division means (2) (52) set frequency characteristics in accordance with the first type of sampling, And a value obtained by multiplying the filter coefficient obtained by the window function having a hanning or hamming characteristic as a filter coefficient. 3. The method of claim 2, corresponding to the splitting means (2) 52, the trans member as more high-band cut-off frequency (f ₅₎ frequency of the band-pass filter for frequency dividing a signal input to the stand filter, said transport member to stand up filter (the frequency of 1/2 of the sampling frequency) band Nyquist frequency of the filter coefficient calculation device, characterized in that the partition is corrected so that (f _2/5) toward the frequency characteristic is zero. 2. The dividing means (2) according to claim 1, wherein the dividing means (2) comprises a Nyquist frequency (half of the sampling frequency) of a frequency band corresponding to one of a plurality of different sampling frequencies (f ₁ ) and (f ₂ ). (f _1/2), the filter coefficient calculation device, characterized in that for correcting the correction so that the frequency characteristic with respect to zero (f _2/2). 2. The first converting means (22) according to claim 1, wherein said calculating means (53) performs first Hilbert transform (3) or linear phase transform (20) on amplitude frequency characteristics of each band divided by said dividing means. Filter coefficient calculation device characterized in that it comprises a. 14. The calculating means according to claim 13, wherein said calculating means (53) is provided with an inverse Fourier changing means (5) for performing inverse Fourier transform on the frequency characteristic converted by said first converting means (22). Filter coefficient calculation device, characterized in that. 2. The apparatus according to claim 1, further comprising: input means (1) (51) for inputting a desired amplitude frequency characteristic, and division and extraction means (2) for dividing a band of the input frequency characteristic into a plurality of bands and extracting a sample point; 52, first changing means 22 for performing Hilbert transform or linear phase transform on amplitude frequency characteristics of each band divided by the dividing and extracting means (2) (52), and the first Second or second conversion means (5) for performing inverse Fourier transform on the frequency characteristic by means of the conversion means (22) and the filter coefficients obtained by the second conversion means (5) directly or indirectly. Filter coefficient calculating device having a transmission means (9). The sampling frequency converting means (6) according to claim 1, wherein the sampling frequency converting means (6) for matching the sampling frequencies of the filter coefficients obtained by the inverse Fourier transform means (5) with respect to each divided frequency band, A plurality of band pass filter means 701, 702 and 703 each outputting only a signal of a predetermined band from the output of (6), and an adding means 8 for adding the filter coefficient obtained for each band, A filter coefficient calculating device, characterized in that for transmitting the added filter coefficient by the transmission means (9). Input means (1) consisting of amplitude input means (31) for inputting an arbitrary amplitude frequency characteristic and phase characteristic input means (39), (32) (33), (41) (42) for inputting an arbitrary phase frequency characteristic (1). ), (31) (39), (31) (33), (31) (41) (42), and inverse puuri for obtaining the impulse response corresponding to the transfer function having the input amplitude and phase frequency characteristics And a conversion means (5) (34) and setting means (35) for setting the impulse response obtained by the inverse Fourier transform as a filter coefficient to the external filter means. 18. The correction means according to claim 17, wherein the input means (31) (32) (33) includes correction characteristic input means (32) for inputting arbitrary correction amplitude frequency characteristics and correction input by the correction specific input means (32). A phase calculating means 33 for obtaining a phase frequency characteristic according to the relation of the Hilbert transform on the basis of the amplitude frequency characteristic, the amplitude frequency characteristic obtained from the amplitude input means 31 and from the phase calculating means 33 A filter coefficient calculating device, characterized in that a filter coefficient is obtained by converting a transfer function having the obtained phase frequency characteristic to an inverse Fourier transform means (34). 18. The method according to claim 17, wherein the phase characteristic input means (31, 41) (42) comprises a group delay input means (41) for inputting a group delay frequency characteristic or a group delay distortion frequency characteristic, and a group delay frequency characteristic or group delay. An integrating means 42 for calculating the phase frequency characteristic by integrating the distortion frequency characteristic, the transfer function having an amplitude frequency characteristic obtained from the amplitude input means 31 and a phase frequency characteristic obtained from the integration means 42; A filter coefficient calculating device for converting the inverse Fourier to the inverse Fourier transform means (34) to obtain a filter coefficient. The method according to claim 17, wherein the phase characteristic input means (39), (32) (33), (41) (42) inputs only phase frequency characteristics in the range of w = 0 to π (w: each frequency), And repetitively changing the sign of the input phase frequency characteristic to generate a phase frequency characteristic in the range of w = π˜2π, and setting this as the overall phase frequency characteristic. 18. The amplitude input means (31) according to claim 17, wherein the amplitude input means (31) inputs only amplitude frequency characteristics in the range of w = 0 to π (w: each frequency), and retraces the input amplitude frequency characteristics to allow w = π to 2π. A filter coefficient computing device, characterized in that an amplitude frequency characteristic is generated in the range of to be the total amplitude frequency characteristic (Fig. 19). 18. The apparatus as set forth in claim 17, further comprising: window means (35) for multiplying the impulse response obtained by the change means (5) (34) by a window function having a hanning or hamming characteristic. A filter coefficient computing device using the impulse response obtained by multiplying a function as a filter coefficient. 18. The method according to claim 17, wherein the input means (1), (31) (39), (31) (32) (33), (31) (41) and (42) are amplitude frequencies according to the second type of sampling frequency. Inputting the characteristic, the inverse Fourier transforming means (5) 34 obtains the input amplitude frequency characteristic, the transfer function calculated by the first type of sampling frequency under the condition of the linear phase, and the calculated transfer function By converting to the inverse fuuri by the first kind of sampling frequency. Filter coefficient computing device, characterized in that for obtaining the impulse response. 18. The display device according to claim 17, wherein the horizontal position of the first display correspondingly moving the movement represents a frequency value, and the vertical position of the first display represents display means (57) representing an amplitude value. A filter coefficient calculating device having setting means 56 for setting a second display indicating a position. 25. The apparatus as set forth in claim 24, wherein said setting means (56) memorizes a preset frequency value and amplitude value, sets a new frequency value and amplitude value, and then displays two positions corresponding thereto by a straight line or a curve. Filter coefficient calculation device. 25. The apparatus as set forth in claim 24, wherein said setting means (56) erases the already displayed line and sets a new value when the already displayed line has the same frequency value and amplitude value as the vertically displayed line. Filter coefficient calculation device. 25. The filter coefficient calculating device according to claim 24, wherein said setting means (56) displays the set frequency value and amplitude value numerically. 25. The filter coefficient calculating device according to claim 24, wherein the first table has a cross shape, and the second table has a circle or cross shape. 18. The filter coefficient according to claim 17, wherein the filter coefficient to be obtained is a filter coefficient for an FIR filter or an IIR filter configured by a digital circuit or a parametric filter type parameter (frequency, 0, gain) configured by an analog circuit. Filter coefficient calculation device characterized in that. Sampling frequency of the first kind and the second kind using the process of inputting the amplitude number characteristic into the input circuit by the first or second sampling frequency and the Hilbert transform based on the input amplitude frequency characteristic Calculating a transfer function by combining a sampling frequency of?, And when the amplitude frequency characteristic is set to the first type of sampling frequency in the input circuit, the transfer function is converted into an inverse In the case where the impulse response is obtained and the amplitude frequency characteristic is set to the second type of sampling frequency in the input circuit, a step of obtaining an impulse response by converting the transfer function to an inverse Fourier at the first type of sampling frequency; Filter coefficient calculation setting method comprising the process of setting the real part of the impulse response obtained by the change to the external transversal filter as the filter coefficient. method. Inputting an amplitude frequency characteristic to an input circuit according to a sampling frequency of the first or second type, obtaining a transfer function based on the input amplitude frequency characteristic, and applying the amplitude frequency characteristic to the input circuit, In the case of setting the sampling frequency of the species, the impulse response is obtained by converting to the inverse pushey using the transfer function at the sampling frequency of the second species, and when the amplitude frequency characteristic is set to the sampling frequency of the second species in the input circuit. Converting the transfer function into an inverse purge at the first type of sampling frequency to obtain an impulse response; and setting a real part of the impulse response obtained by changing to the inverse puuri as a filter coefficient to an external transversal filter. Method for setting calculation of filter coefficients.

Images