JP2671648B2 - Digital data interpolator - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital data interpolating device for interpolating digital data generated in time series.

[0002]

2. Description of the Related Art Heretofore, in electronic musical instruments, in order to process an analog amount such as an amplitude, a waveform or a pitch of a musical tone to be produced, digital processing for handling various time-series digital data is performed.

FIG. 9 is a block diagram showing an example of this type of electronic musical instrument. When a key is pressed on the keyboard a, the key detecting section b generates a key code and a key-on, and a pitch data generating section c.
Generates pitch data corresponding to the key code, and outputs this pitch data to the waveform generator e via the interpolation circuit d.

On the other hand, the bend wheel f is operated to change the pitch of the musical sound, and the bend wheel detector g generates pitch control data according to the operating state of the bend wheel f and interpolates the pitch control data with the interpolation circuit h. Is output to the pitch data generating section c via.

The waveform generator e generates waveform data according to the input pitch data and outputs it to the multiplier j via the interpolator i, and the envelope generator k generates envelope data corresponding to the amplitude of the musical tone. And outputs it to the multiplier j through the interpolation circuit m.

Then, the multiplier j multiplies the waveform data by the envelope data to obtain the waveform data whose amplitude is controlled. This waveform data is the D / A converter n
Is converted into an analog signal, and a musical tone is generated by the sound system o.

The pitch data, pitch control data, waveform data, and envelope data are time-series digital data, which are interpolated by the interpolating circuits d, h, i, m. For this kind of digital data interpolation, for example, a digital low-pass filter as disclosed in Japanese Patent Laid-Open No. 61-107298 is used.

[0008]

However, since the digital low-pass filter as described above outputs the data value following the value of the input digital data as the interpolation data, the filter characteristic matches the characteristic of the input data. There is a problem that it cannot be interpolated properly unless it is a thing.

For example, when the time constant of the filter is too long for the time-series data with a large change, the interpolation data H does not follow the actual data E well as shown in FIG. 8 (A). Further, as shown in FIG. 8B, when the time constant of the filter is short with respect to the time series data E having a small change, the stepwise interpolation data H is obtained. It is an object of the present invention to appropriately interpolate digital data generated in time series even if the characteristics of the digital data change.

[0010]

SUMMARY OF THE INVENTION The digital data interpolating apparatus of the present invention, which has been made to solve the above-mentioned problems, comprises a change rate calculating means for obtaining a change rate of time-series digital data and a time constant. An interpolating circuit that is composed of a variable digital filter, and that interpolates input digital data with characteristics according to the time constant and outputs interpolated data,
Time constant setting means for setting the time constant of the interpolation circuit according to the change rate obtained by the change rate calculating means.

[0011]

In the digital data interpolating apparatus of the present invention, the time constant of the digital filter in the interpolating circuit is set according to the rate of change of the digital data obtained by the rate-of-change calculating means. When the rate of change of the digital data is large, the followability of the interpolation data is good, and when the rate of change is small, the interpolation data is smooth without following the change point of the digital data.

[0012]

1 is a block diagram showing an interpolating device according to an embodiment of the present invention. This interpolator is applied to an envelope generator in an electronic musical instrument, and the envelope generator 1 time-sequentially outputs envelope data e (i) of a digital value corresponding to the amplitude of a musical tone to be generated at a predetermined frequency f _1. Output to. For example, when a musical tone of a piano is generated by this electronic musical instrument, numerical data obtained by sampling the envelope value as shown in FIG. 5 at a predetermined cycle is output as envelope data e (i).

In this case, the envelope value rises in the attack portion A immediately after the key is depressed, is attenuated at a relatively fast speed in the decay portion D from its maximum amplitude, and is gradually attenuated in the sustain portion S following the decay portion D. In addition, when muting a tone by releasing the key, etc.
Alternatively, it is rapidly attenuated from the middle of the sustain block S. As described above, the change rate (absolute value) of the envelope data e (i) is large in the attack portion A, the decay portion D and the release portion R, and is small in the sustain portion S.

On the other hand, waveform data of a musical tone to be generated is output at a constant frequency f ₀ as time-series digital data by interpolating data read from a waveform memory by a device (not shown). The frequency f ₀ of the data is higher than the frequency f _{1 of the} envelope data e (i) output from the envelope generator 1. Further, by multiplying the waveform data by the envelope value, a digital tone signal whose amplitude is controlled can be obtained.

Therefore, the interpolation circuit 2 interpolates the envelope data e (i) at the same frequency f ₀ as the waveform data to generate interpolation data having the same resolution as the waveform data. In addition,
The frequency f _{1 of the} envelope data and the sampling frequency f ₀ of the waveform data are, for example, f ₁ = 1 kHz and f ₀ = 50.
It is set to be kHz.

The interpolation circuit 2 includes a subtracter 21, a multiplier 22,
The envelope data e (i) from the envelope generator 1 is composed of an adder 23 and a register 24.
Is input to the subtractor 21 as the minuend data, and the multiplier 2
The subtraction output of the subtracter 21 is multiplied by the filter coefficient τ by 2, and the multiplication output is input to the adder 23. Further, the addition output of the adder 23 is input to the register 24,
The data latched in the register 24 is input to the subtractor 21 as reduction data and to the adder 23 as addition data.

The interpolation circuit 2 operates at the frequency f ₀ of the waveform data, and as shown in FIG. 2, the envelope data e (i) latched by the envelope generator 1 at the frequency f ₁ is fed to the frequency f _0. Of the envelope data e (i, k), and outputs interpolation data h (i, k) as shown in the figure to the envelope data e (i, k). Becomes

Here, the time constant of this low-pass filter is determined by the filter coefficient τ set in the multiplier 22 from the time constant controller 3. The filter coefficient τ is 0 <
It is a value in the range of τ <1, and when the value of the filter coefficient τ changes to the “0” side, it becomes like the one-dot chain line L1 in FIG. 2, and when it changes to “1” side, it becomes like the one-dot chain line L2.

On the other hand, the time constant controller 3 includes an envelope change rate calculator 31, an n-stage shift register 32, and an adder 3.
3, subtractor 34, register 35, and time constant conversion device 3
6 and outputs a filter coefficient τ for setting the time constant in the interpolation circuit 2.

The envelope change rate calculation device 31 calculates the difference ΔS (i) between the current data and the previous data of the envelope data e (i) from the envelope generator 1.
= E (i + 1) -e (i) is calculated, and this difference ΔS (i) is input to the n-stage shift register 32 and the adder 33 as the change rate of the envelope data e (i).

The output of the n-stage shift register 32 is input to the subtractor 34 as subtractive data, and the addition output of the adder 33 is input to the subtractor 34 as minuend data.
Further, the subtraction output of the subtractor 34 is input to the register 35, and the data latched in the register 35 is input to the adder 33 and the time constant conversion device 36.

The time constant controller 3 operates at the same frequency as the frequency f _{1 of} the envelope data e (i) output from the envelope generator _1, and the n-stage shift register 3
2 sequentially delays the sequentially input difference ΔS by n stages and sequentially outputs it as subtraction data to the subtractor 34, and the adder 33 adds the difference ΔS and the data delayed by one stage by the register 35, and adds this addition value The data is output to the subtractor 34.

Therefore, the data latched in the register 35 is the sum S of the n consecutive change rates of the difference ΔS sequentially output from the envelope change rate calculating device 31, and this sum S is the frequency f ₁ The values are sequentially input to the constant conversion device 36.

The time constant converter 26 divides the sum S of the input change rates by the number n of shift stages of the n-stage shift register 22 to obtain the average change rate of n sections of the envelope data e (i). S / n is sequentially obtained, and the filter coefficient τ corresponding to the average change rate S / n is sequentially output to the multiplier 21 at the frequency f ₁ . Here, the relationship between the filter coefficient τ and the average change rate S / n is set in advance. When the average change rate S / n is large, the filter coefficient τ is increased, and when the average change rate S / n is small, Reduce the filter coefficient τ.

For example, the relation between the change L of the envelope data and the best filter coefficient τb for this is τb =
Alternatively, the filter coefficient τ for the average change rate S / n may be obtained by τ = R (S / n) as a function R of R (L). A simple function of this function R (x) is a bit shift function.

As described above, the envelope data e (i)
Since the filter coefficient τ is set to an appropriate value in accordance with the change rate of the interpolation data h (i ,,) when the envelope data e (i) changes greatly as shown in FIG.
k) follows well, and when the change of the envelope data e (i) is small as shown in FIG. 3 (B), the interpolation data h (i,
It is possible to prevent k) from becoming stepwise.

An envelope data input to the interpolation circuit 2 is provided between the envelope generator 1 and the interpolation circuit 2 by providing, for example, an n / 2-stage shift register that operates at the same frequency f ₁ as the time constant control unit 3. The e (i) may be delayed by a predetermined amount (n / 2 stages in this case). By doing so, the interpolation circuit 2 determines the envelope data e (i) that is currently being input.
Envelope data e (i) of approximately n / 2 steps before and after including (i)
The interpolation can be performed based on the average change rate S / n of

Although the above-described embodiment applies the present invention to the interpolation of envelope data, it can also be applied to the other interpolation circuits described with reference to FIG. Also,
In the above embodiment, the sum S of the change rates is simply taken, but when the sum is taken, each may be multiplied by the weight.

Further, the time constant control unit 3 of the above embodiment is adapted to generate the filter coefficient τ by the n average change rates S / n of the change rates ΔS of the envelope data e (i). As shown in FIG. 4, the adder 3
7, a multiplier 38 and a register 39 constitute a low-pass filter, the change rate ΔS from the envelope change rate calculation device 31 is averaged by this low-pass filter, and the time constant conversion device 36 filters according to the averaged change rate. The coefficient τ may be generated.

FIG. 6 is a block diagram showing an embodiment in which the present invention is applied to waveform switching interpolation. The waveform memory 4 stores, for example, waveform data E ₀ , E ₁ , ... Of different waveforms corresponding to the attack portion A, the decay portion D, the sustain portion S, and the release portion R of the envelope shown in FIG. 5, respectively. Has been done.

Each waveform data E of this waveform memory 4₀, E
₁,… Is for one cycle where the number of data sampling points is m
It is the data of (cycle T).
One cycle of waveform data is repeatedly read at cycle T
You. And should be pronounced corresponding to each part of the envelope
When changing a musical tone, the waveform data to be read is, for example, E ₁
To E_TwoIt is switched as follows.

However, if the waveform data is simply switched, the waveform abruptly changes and an unnatural musical tone is generated. Therefore, for example, at time t shown in FIG.
Data values D1 and D2 of t + mT, t + 2mT, ..., T + 5mT, ... Are interpolated for each phase for data values of the same phase of the waveform data E ₁ and the waveform data E ₂ . .

That is, the time-series data to be interpolated in this embodiment is a data string of in-phase data of the waveform data that is sequentially read, and the waveform interpolation is performed by interpolating the data strings corresponding to the respective phases. It is something to do. Note that such a waveform switching interpolation technique is disclosed in, for example, Japanese Patent Laid-Open No. 61-107298.

In FIG. 6, when a key is pressed on the keyboard 5, the key detection circuit 5a generates a key code KC and a key-on signal KON, and the address generator 6 outputs waveform data E ₀ , E ₁ ,
Sample points in ... and the address data AD1 corresponding to (phase) (0) ~AD (m- 1), each waveform data E _0, E _1, address data AD2 to specify ...
(0), AD2 (1), ... Are generated.

Then, these address data AD1,
Predetermined waveform data E is output from the waveform memory 4 by the AD 2, and the read waveform data E is input to the low pass filter 7 and the time constant controller 8.

The low-pass filter 7 is composed of a subtractor 71, a multiplier 72, an adder 73 and an m-stage shift register 74, and the waveform data E whose number of sample points is m.
With respect to, each of the multipliers 72 functions as an independent filter for a plurality of data strings composed of in-phase data.
Data is interpolated for each data string according to the filter coefficient τ set to.

For example, data that changes abruptly in the data sequence of the same phase, such as the change from the data sequence of D1 to the data sequence of D2 in FIG. 7, is changed to a data sequence having a gradually changing period T. Since this function works similarly for the data sequence of the same phase corresponding to each phase, the interpolation is performed so that the waveform gradually changes (see Japanese Patent Laid-Open No. 61-107298).

The waveform data interpolated by the low-pass filter 7 is added to the envelope generator 1 by the multiplier 9.
The envelope data from 0 is multiplied, and thereby the waveform data whose amplitude is controlled is converted into an analog signal by the D / A converter 11, and the sound system 12 generates a musical sound.

In the time constant control unit 8, the waveform change rate calculation device 81 calculates the change rate ΔS in the in-phase data sequence of the waveform data of the number of sample points m, and the adder 82, the multiplier 83 and the m-stage shift register. Reference numeral 84 functions as a low-pass filter that is independent of the rate of change ΔS in each in-phase data string. Then, the time constant conversion device 85 outputs the filter coefficient τ corresponding to the averaged difference ΔS, similarly to the time constant conversion device 36 of the above embodiment, for each phase of the waveform data.

As described above, in this embodiment, since the filter coefficient τ is set to an appropriate value in accordance with the rate of change of the in-phase data in the waveform data, when the change in the waveform is large, the change in the waveform is well followed. Thus, the switching interpolation can be performed, and the smooth switching interpolation can be performed even when the change in the waveform is small.

[0041]

As described above, according to the present invention, an interpolation circuit for interpolating digital data input in time series is configured by a digital filter having a variable time constant, and the digital data input in time series is converted. The rate of change is calculated, and the time constant corresponding to this rate of change is set in the digital filter of the interpolation circuit to interpolate the digital data.Therefore, when the rate of change of the digital data is large, the followability of the interpolated data is good. When the rate of change is small, smooth interpolation data is obtained without following the changing point of digital data. Therefore, even if the characteristics of the digital data change, the interpolation can be appropriately performed.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an interpolation device according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating a relationship between envelope data and interpolation data according to an embodiment.

FIG. 3 is a diagram showing a state of interpolation data in an example.

FIG. 4 is a diagram showing an example of a low-pass filter used in another embodiment.

FIG. 5 is a diagram showing an example of an envelope shape in the embodiment.

FIG. 6 is a block diagram showing another embodiment in which the present invention is applied to waveform switching interpolation.

FIG. 7 is a diagram showing an example of waveform data in the same example.

FIG. 8 is a diagram showing a state of interpolation data in a conventional interpolation circuit.

FIG. 9 is a block diagram of an electronic musical instrument according to the present invention.

[Explanation of symbols]

1 ... Envelope generator, 2 ... Interpolation circuit, 3, 8 ... Time constant control unit, 4 ... Waveform memory, 7 ... Low-pass filter.

Claims (1)

(57) [Claims] 1. A change rate calculating means for obtaining a change rate of time-series input digital data and a digital filter having a variable time constant, and the input digital data is interpolated with a characteristic according to the time constant. And an interpolation circuit for outputting interpolation data, and a time constant setting means for setting a time constant of the interpolation circuit according to the change rate obtained by the change rate calculating means. .