JPH03124110

Patent Translate
Powered by EPO and Google
Notice
This translation is machine-generated. It cannot be guaranteed that it is intelligible, accurate,
complete, reliable or fit for specific purposes. Critical decisions, such as commercially relevant or
financial decisions, should not be based on machine-translation output.
DESCRIPTION JPH03124110
[0001]
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a
digital convolution filter having the same characteristics and ease of use as an analog voltage
complex filter. [Prior Art] In an analog music synthesizer or the like, a VCF (voltage complicated
filter) as shown in FIG. 9 is actively used. Here, each unit filter 1-1. 1−2. ... 1-〇 consists of a
first-order low-pass filter configured to be able to vary the cutoff frequency using, for example, a
passive circuit of CR, and the low-pass transfer function has a cutoff frequency fC fC = □... (2) 2
Further, the feedback circuit 3 and the subtraction circuit 4 are cascaded unit filters 1-1. 1-2. The
output of the final stage 1-n of 1-n is negatively fed back to the first stage. The gain β of the
feedback circuit 3 is related to the resonance near the cutoff frequency fC of vCF. In digital music
synthesizers and the like, digital filters of FIR (finite impulse response) type or IIR (infinite
impulse response) type are used as ones corresponding to such VCF (digital companded filter) .
However, these digital filters have many multiplier coefficients to be set at the same time, and the
relationship between these coefficients and the filter characteristics is complicated, which makes
the control difficult. The present inventors can control the characteristics using a digital firstorder low-pass filter as disclosed in JP-A-61-18212 instead of the unit filter in the analog filter of
FIG. An attempt was made to construct a digital complex filter. An example is shown in FIG. 10. In
the figure, the sign "+" adds the data input to the unmarked or ten-marked input end and adds
the data input to the unmarked input end. An adder or subtractor to subtract, M is a multiplier
that multiplies an input signal by a constant value (hereinafter referred to as a coefficient), Z ′
′ 1 is a period of one cycle (sampling period) of sampling pulse T delay Delay circuit. Also, the
code attached to the upper side of each multiplier indicates the coefficient by which the signal is
multiplied in that multiplier. In the figure, the digital first-order low-pass filter 1-1.1-2, which is a
unit filter, uses the addition in the characteristic equation of the analog first-order low-pass filter
as an adder, the subtraction as a subtracter, the multiplication as a multiplier, and the integration
08-05-2019
1
as an accumulator. The low-pass transfer function is expressed by αH (Z)-,-(1-CE) z -'°°°°° (3).
Further, the cutoff frequency fc is represented by α (= aT) if it is sufficiently smaller than 1
(where f 3 is a sampling frequency). That is, this digital first-order low-pass filter has almost the
same frequency characteristics as the analog filter, and has an advantage that the relationship
between the coefficient α of the multiplier and the filter characteristic is simple and easy to
handle like the analog filter. In the digital complex filter of FIG. 10, the frequency characteristic
can be arbitrarily set as shown in FIG. 11 by setting the coefficient α of the multiplier M and the
coefficient β of the multiplier 3. Here, the cutoff frequency depends on the coefficient α of the
multiplier M, as can be understood from equation (4). The coefficient β of the multiplier 3
corresponds to the gain of the feedback circuit 3 in the analog VCF, and relates to the resonance
in the vicinity of the cutoff frequency fc of the filter. [Problems to be Solved by the Invention] By
the way, in the digital compandal filter of FIG. 10, if β is set to a large value and α is brought
close to 1, the half of the sampling frequency fs is When oscillation occurs and α approaches 1
further, the oscillation amplitude further increases. Since the filter shown in FIG. 6 is a system of
finite word length, the input signal is blocked due to this oscillation, and the signal level of the
output is disadvantageously reduced. That is, for the input signal sample waveform as shown in
FIG. 12A, the output signal sample waveform is such that the oscillation waveform is
superimposed as shown in FIG. 12B, and the maximum amplitude is determined by the finite
word length. , The part beyond it is limited in amplitude and is clipped as a waveform, and the
amplitude of the output signal is reduced. The present invention has been made in view of the
problems in the above-described conventional example, and it is an object of the present
invention to provide a digital complex filter having the same characteristics and ease of use as an
analog voltage complex filter. Do. [Means for Solving the Problems] In order to achieve the above
object, the present invention cascade-connects a plurality of digital first-order low-pass filters
(unit filters) in which the same cut-off frequency is arbitrarily set. And adding the addition in the
characteristic equation of the analog first-order low-pass filter as the digital first-order low-pass
filter in a digital compandal filter configured to feed back the output of the cascade circuit to the
input end using a multiplier and an adder / subtractor. Using a filter that substitutes subtraction
for subtraction, multiplication for multiplication, and integration for accumulation, and the
transfer gain decreases at half the sampling frequency for signal processing. Inserting a
characteristic insertion filter in a closed loop consisting of the plurality of unit filters and the
feedback circuit It is characterized in that.
As the insertion filter, a band pass filter or another digital low pass filter having a cutoff
frequency higher than the cutoff frequency of the first order low pass filter and lower than a half
frequency of the sampling frequency can be used. Here, the subtractor includes one configured
to perform subtraction equivalently by an inverter and an adder. [Operation] In the above-
08-05-2019
2
described configuration, the operation of the portion composed of the plurality of unit filters and
the feedback circuit is substantially the same as the analog VCF in FIG. 10 except for the problem
of the oscillation of fs / 2. In the insertion filter, the transfer gain is reduced at a half of the
sampling frequency fs at which the unit filter is operated. Thus, the loop gain of the closed loop
is reduced by the presence of the insertion filter. Thereby, the oscillation of f5 / 2 is prevented or
the amplitude is reduced. Meanwhile, the cut-off frequency of the insertion filter is set higher
than the cut-off frequency of the unit filter. For this reason, the adverse effect on the
characteristics as a digital complex filter is prevented. Further, the present invention is
configured to be equivalent to the analog VCF except for the insertion filter, so that the frequency
characteristics are almost the same. That is, the cutoff frequency fC can be controlled by giving
the unit filter a multiplication coefficient instead of the control voltage of the analog VCF, and
changing the coefficient of the feedback multiplier can change the resonance characteristic of the
filter. . Furthermore, since a digital first-order filter similar to that disclosed in the
aforementioned Japanese Patent Application Laid-Open No. 61-18212 is used as a unit filter, the
relationship between the multiplier coefficient and the filter characteristics, particularly the
cutoff frequency fc is simple and handled easy. C Effect] As described above, according to the
present invention, it is possible to realize a digital combined filter having the same characteristics
as the analog VCF and having the same ease of use as the analog filter. The present invention will
be described in detail based on the following examples. Note that common or corresponding
parts are denoted by the same reference numerals throughout the drawings. FIG. 1 shows the
configuration of a digital complex filter according to an embodiment of the present invention.
The filter shown in the figure comprises a unitary filter 1-1.1-2 which is a digital first-order lowpass filter, a multiplier 3 and a subtractor 4, and a second low-pass filter 5 as an insertion filter
as a feature of the present invention. .
The digital first-order low-pass filters 1-1 and 1-2 are substantially the same as those described
in the embodiment of JP-A-68-18212, and the Laplace transfer function H (S) of the analog firstorder low-pass filter , +8... (5), and this 2 閏 H (z) is appropriately simplified according to
necessity, and then converted into a circuit. These Laplace transfer functions and S-2
transformations are known. As the s-z conversion to be adopted, a <s-z conversion based on a
differential approximation of differential which performs the conversion of s-aml-z-'exp (aT)
and (s-a + Jb) (s-a + j) b) matched 2-conversion in which conversion is performed by a
conversion pair of b) = (s−a) 2 + b2−4−2 e ′ ′ cos (b T) z − ′ + e ′ ′ ′ ′ z− 2 is
preferable. In the case of <s-z conversion based on differential approximation of differential, it is
the simplest. When S is replaced with 1-z- 'and aT is replaced with α to perform this S-Z
conversion on the Laplace transfer function, αH (z) = 1-,-, +. -----(6) This equation can be
expressed in a circuit as shown in FIG. 2 using the delay circuit z-1, the multiplier α and the
adder-subtractor. This circuit is corrected as follows, since the coefficient has to be obtained by
division 1 / (1 + α), which may cause a processing delay. That is, the difference 1-z "'between the
current data and the data one sampling period before means a derivative, and the derivative (1-z-
08-05-2019
3
') alpha of the constant alpha is zero. Taking this into consideration, the above equation can be
rewritten as αt-(ta) z-'(6). If this equation is expressed in a circuit, it becomes as shown in FIGS.
FIG. 4 shows a circuit example of a first-order low-pass filter obtained by matching 2 conversion.
The filters in FIGS. 1 to 4 are determined according to the coefficient α given by the value in the
range of 0 to 1 and the cutoff frequency is 0. And, α is! In a sufficiently smaller range, the
accuracy of the above conversion or approximation is high, and the cut-off frequencies fC and α
are approximately proportional to each other (where fs is a sampling frequency). Therefore, by
changing this coefficient α, the cutoff frequency can be changed as shown in FIG. The fact that
the coefficient α is approximately proportional to the cutoff frequency fc in this way means that
the filter can be easily controlled. The filter shown in FIGS. 2 to 3 includes 0UT1 as an output
terminal and 0UT2 for generating an output obtained by delaying this 0UTI by one sampling
period by the delay circuit z-1.
In FIG. 1, the output terminal of the filter 1 uses ou'rt, but the output terminal of the filter 2 uses
0 UT2. This is because when the closed loop (delay free loop) not including the delay circuit is
formed, the normal calculation operation is not performed, so the unit filter 1-1.1-2, the second
low pass filter 5, the multiplier 3 And the subtracter 4 to include the delay circuit z-1. In FIG. 1, in
the second low-pass filter 5, the unit filter 1- is used except that the coefficient of the multiplier
M is set to α'-172 and the cutoff frequency fQ is set to approximately f, / 4. It is configured
exactly the same as 1.12. FIG. 5 shows the frequency characteristic of the second low pass filter
5. This filter 5 reduces the loop gain at the frequency f3 / 2 of the closed loop consisting of the
unit filter 1-1.1-2, the multiplier 3 and the subtracter 4, and prevents the oscillation of the
frequency f3 / 2 or Reduce the amplitude of the waveform. The multiplier 3 receives the output
0UT2 of the outputs of the unit filter 2 via the second low pass filter 5, and multiplies the output
0UT2 by a predetermined coefficient β. The subtractor 4 subtracts the output of the multiplier 3
from the input sample waveform signal and inputs it to the unit filter 2. By changing this
coefficient β, the filter characteristics can be changed as shown in FIG. FIG. 6 shows a
configuration example in the case of applying the digital combined filter of the present invention
to a sound source of an electronic musical instrument. In the figure, reference numeral 61
denotes a digital waveform sound source consisting of a memory storing, for example, each
sample point data of natural musical instrument sound, 62 a digital condural filter of the present
invention, and 63 a based on the output of the digital condural filter 62. It is a tone forming
means for forming a tone. In the figure, as shown in FIG. 7, the digital compandal filter 62 has its
cutoff frequency controlled by the parameter .alpha. The filter resonance characteristic is
controlled by the control signal (in particular, corresponding to the timbre data). By these
controls, the tone color of the generated tone can be controlled. [Modification] The present
invention can be appropriately modified and implemented without being limited to the abovedescribed embodiment. For example, although the example in which the second low pass filter 5
is inserted between the unit filter 1-2 and the multiplier 3 has been described above, the
insertion position of the low pass filter 5 may be anywhere within the closed loop. .
08-05-2019
4
For example, as shown in FIG. 8, the same effect can be obtained by inserting the circuit into the
ring opening main circuit of the unit filter 1-1 and the subtracter 4. However, when the
coefficient β is reduced, oscillation is less likely to occur even if α is brought close to 1, so it is
preferable that the filter be less susceptible to the influence of the inserted filter. For this
purpose, the position of the filter to be inserted is preferably in a feedback system, such as
between the unit filter 1-2 and the multiplier 3 or between the multiplier 3 and the subtractor 4.
In the above, the second low-pass filter 5 fixed the characteristics, but as shown in FIG. 8, the
characteristics may be made variable by the coefficient γ (= α ′). Further, the filter to be
inserted is not limited to the examples shown in FIGS. 1 and 8 and may be, for example, a band
pass filter as long as the gain is lowered near the high frequency f, / 2. . Furthermore, control of
each of the coefficients (parameters) may be logarithmic control. In this case, multiplication can
be processed by addition, and processing can be simplified.
[0002]
Brief description of the drawings
[0003]
FIG. 1 is a circuit diagram showing the configuration of a digital complex filter according to an
embodiment of the present invention, FIGS. 2 to 4 are circuit diagrams showing an example of
the configuration of a unit filter in FIG. 1, and FIG. 1 is a graph showing an example of the
frequency characteristic of the second low pass filter in FIG. 1, FIG. 6 is a circuit diagram of an
electronic musical instrument sound source showing one application example of the present
invention, and FIG. FIG. 8 is a circuit diagram showing a modification of the digital complex filter
of FIG. 1, FIG. 9 is a circuit diagram of a conventional analog VCF, and FIG. Fig. 11 is a circuit
diagram of the digital convolard filter examined prior to Fig. 11, a frequency characteristic
diagram of the digital convolard filter in Figs. 1 and 10, and Figs. 12A and 12B are Figs. Is a
diagram illustrating input and output waveforms of the digital Condo rolled filter 0 Figure.
1-1.1-2: Unit filter (digital first-order low-pass filter) 3, M: multiplier 4 · 1+ "two-subtractor 5:
second low-pass filter (insertion filter) Fig.
08-05-2019
5