JPH03152599A - Digital filtering device for sound - Google Patents

Abstract

PURPOSE:To obtain a sound with a few folded noise not depending on the pitch of the sound to be reproduced by controlling a cutoff frequency for reproducing data in an LPF digital filtering circuit means based on the relation of an original sound with a reproducing pitch. CONSTITUTION:The subject device is equipped with a sampling data storage means 11 to store data in which the original sound is sampled, a reproducing pitch designation means 1 to instruct the pitch of the reproducing sound variably, a reproducing data generating means 12 to generate the reproducing data having the same pitch as that of a designated reproducing sound, the LPF digital filtering circuit means 13 to perform the low-pass filtering of the reproducing data, and a characteristic control means to control the characteristic of the circuit 13. In other words, the cutoff frequency of the LPF digital filter circuit means 13 is controlled variably for the pitch of the reproducing sound designated variably based on the reproducing pitch and the pitch of the original sound stored in the sampling data storage means 11. Thereby, it is possible to reduce the folded noise due to the reproducing processing of the reproducing data generating means.

Description

[Detailed description of the invention] [Technical field of invention] The present invention relates to a digital filter device for sound.

[Prior art and its problems] Digital filter devices for sound used in audio devices, electronic musical instruments, etc. are known. One of the problems inherent in reproducing sound information recorded by encoding such as PcM is the problem of aliasing noise. This can be avoided on the reproduction side by lowering the cutoff frequency of the LPF filter, but simply lowering the cutoff frequency also removes signal components contained in the sound itself, resulting in a poorer tone. therefore,
The usual countermeasure is to band-limit the signal captured during sampling on the recording side to within the Nyquist frequency.
Additionally, oversampling technology has recently been put into practical use, in which the rate of playback sampling is higher than that of recording sampling.

However, when playing back PCM musical sound data at a pitch different from the recorded j aliasing noise, which cannot be ignored, is more likely to occur.
A satisfactory solution to this problem has not yet been found.

[Object of the Invention] Therefore, an object of the present invention is to provide a sound digital filter device that can reduce the problem of aliasing even when reproducing sampled sound data at a pitch different from that of the original sound.

[Structure 1 of the Invention] In order to achieve the above object, the present invention includes a sampling data storage means for storing data obtained by sampling the IX sound, and a reproduction pitch specifying means for variably instructing the pitch of the reproduced sound. , a reproduction data generation means for accessing the sampling data storage means and generating reproduction data having a specified reproduction sound pitch; and an LPF for low-pass filtering the reproduction data from the reproduction data generation means.
comprising digital filter circuit means and characteristic control means for controlling the characteristics of the LPF digital filter circuit means,
A sound characterized in that the characteristic control means has a cutoff frequency determining means for variably determining the cutoff frequency of the LPF digital filter circuit means for the reproduced sound based on the pitch of the original sound and the pitch of the reproduced sound. A digital filter device is provided.

According to this configuration, for the pitch of the reproduced sound that is variably specified, the LPF is
Since the cutoff frequency of the digital filter circuit means is controlled to be variable, it is possible to reduce the problem of aliasing noise caused by the reproduction processing of the reproduction data generation means.

- In the configuration example, the cutoff frequency of the LPF digital filter circuit means is decreased in proportion to or substantially proportional to the reproduction pitch as the reproduction pitch decreases from the original sound pitch. This also takes into consideration the naturalness of musical tones, and has the effect of maintaining a natural timbre by ensuring that the ratio of harmonic components to the fundamental wave of musical tones does not change significantly depending on the playback pitch of musical tones. There is.

[Example] Hereinafter, the present invention will be described in detail with reference to the drawings.

FIG. 1 shows a functional block diagram of a digital filter device for sound according to the present embodiment. This digital filter device for sound is applied to an electronic musical instrument, and includes a performance input means l (for example, a keyboard, a performance sequencer, etc.) Performance information is inputted from the computer, and the performance input decoding means 2 decodes the performance information.

A sampling sound storage means 11 is provided as a source of digital sound data, and information such as the performance pitch decoded by the performance input decoding means 2 is passed to the reproduction data generation means 12. In response to this, the reproduction data generation means 12 reads the sampled sound data from the sampling sound storage means 11, generates reproduction data having the specified performance pitch by interpolation, etc., and supplies it to the LPF digital filter circuit means 13. .

Further, the performance pitch information decoded by the performance input decoding means 2 is also passed to the first cutoff frequency parameter calculation means 4. This means 4 is LPF digital filter circuit means 1
The first cutoff frequency f in the frequency response characteristic of No. 3
This is to generate cI or the parameters that determine it.

That is, according to this embodiment, the sampling sound storage means 11
Taking into account the nature of natural musical sounds in which the sound is played back at a pitch different from the original sound, the spectrum changes more or less depending on the pitch, and the problem of aliasing noise (aliasing) caused by interpolation in data pitch playback, we developed a method. 4 receives the pitch 3 of the original sound sampled in the sampling storage means, and determines a first cutoff frequency fcl that matches the performance pitch obtained from the performance input decoding means 2.

Furthermore, this embodiment relates to providing the LPF digital filter circuit means 13 with a second cutoff frequency fc2 lower than the first cutoff frequency fc+, and this second cutoff frequency fc2 is lower than the first cutoff frequency fc+. It has a function of making the first cutoff frequency dependent on the performance pitch given by the frequency parameter calculation means, and also has means for introducing temporal changes in the musical tone spectrum. That is, data C of the basic ratio between the first and second cutoffs specified by the user-programmable ratio specifying means 5 and the time after note-on detected by the performance input decoding means 2 (note time ) is inputted together with the first cutoff frequency fc1 into the second cutoff layer wavenumber parameter calculation means 7, where the information of the LPF digital filter circuit means 13 is inputted to the second cutoff layer wavenumber parameter calculation means 7. Second
The cutoff frequency or the parameters that determine it are calculated.

The first cutoff frequency fcl calculated in this way
and the second cutoff frequency fc2 are supplied together with the resonance parameter R designated by the user programmable resonance designation means 8 to the filter coefficient calculation means 9, where they are used to directly control the LPF digital filter circuit means 13. The filter coefficient set, i.e.
Information used in digital performance operations within the LPF digital filter circuit means 13 is generated.

The generated filter coefficient set is transferred to the LPF digital filter circuit means 13 via the transfer means lO, and upon receiving this transfer information, the LPF digital filter circuit means 1
3 performs low-pass filtering processing on the musical tone signal from the reproduction data generating means 12.

The results are sent to an external sound system (not shown).

FIG. 2 is a hardware block diagram of an electronic keyboard instrument having the features described in FIG. 1. CPU20 is ROM3
It operates according to the program stored in RAM 40.
is used as working memory. The input device M50 includes a keyboard 50A as a performance input means and a numeric keypad 50B as an input means for ratio data and resonance data.The state of the numeric keypad 50B and the state of the keyboard 50A are C.
It is periodically monitored by the PU 20, and the results, messages, etc. are displayed on the display device i60.

For example, ratio designation data and resonance data from the numeric keypad 50B are stored in corresponding areas in the RAM 40 and used for filtering control, which will be described later.

When a key press is detected from the keyboard 50A, the CPU
20 compares the performance pitch with the original sound pitch of the sampled sound data in the PCM sound source 70 (this is part of the constant data in the ROM 30), calculates its relative height, and performs a predetermined calculation. LPF digital filter circuit (
Waveform signal processing device! 180) is calculated. Further, when a key press is detected, the CPU 20 transfers the performance pitch data to the PCM sound source 90 to cause the PCM sound source 90 to generate PCM playback data having the performance pitch, and also counts the note time from the key press detection and periodically PCM sound source 9 uses an amplifier envelope to control the volume of playback data.
Send to 0.

In parallel with this series of PCM sound source processing, the CPU 20
Waveform signal equipment including PF digital filter calculation circuit!
180 (DSP) to enable real-time level dynamic low-pass filtering of the PCM playback data in the device 80. That is, in response to key press detection,
Part of LPF digital filter circuit! After determining the cutoff frequency determining parameter, the CPU 20 periodically creates a time variation parameter (filter envelope) for the filter, and stores this time variation parameter and the ROM 40.
A parameter for determining the second cutoff frequency in the LPF digital filter circuit is determined from the ratio data from the numeric keypad 50B stored in and the first cutoff frequency determination parameter that has already been calculated and saved. , generates a set of filter coefficients suitable for processing by the waveform signal processing device 80, and sends this coefficient set to the waveform signal processing device 80 to control the device 80 in real time.

As a result, the waveform signal processing device 80 generates PCM playback data in a format in which the first cutoff frequency depends on the performance pitch and the second cutoff frequency depends on the first cutoff frequency, the designated ratio, and the time-varying parameter. to low-pass filter.

The filtered digital musical tone signal is outputted as an audio signal through a D/A converter 90.

Figure 3 shows the LPF realized by the waveform signal processing device shown in Figure 1.
0 shows the logical configuration of the digital filter circuit.
As shown in the figure, this LPF digital filter circuit is a 5th-order Lohas filter in which first-order, second-order, and second-order digital filter sections are cascaded. The first-order digital filter section has input multipliers 101 that multiply the PCM playback data input from the PCM sound source 70 by 1, and a coefficient multiplier 102 that multiplies its output and the output of the 1-sample delay element 124 by btt. an input adder 114 that adds the feedback signal and inputs the result to the delay element 124 and the output adder 115; and a coefficient multiplier 102 that multiplies the input addition result and the output of the delay element 124 by b12. an output adder 115 forming the output of the primary section.

A secondary section cascaded to this primary section has an input multiplier 104 that doubles the primary section output signal;
Coefficient multiplier 105 which multiplies its output by 21 times the feedback signal of the secondary section, that is, the delayed output of delay element 125 which delays the output of input adder 116 by one sample.
Addition $ which takes the sum of the output and the output of the coefficient multiplier 106 which multiplies b22 the 2-sample delayed signal from the delay element 126 which further delays the output of the delay element 125 by 1 sample and outputs the delayed signal.
The input adder 116 adds the signals from 117, and the input addition result is added to b23 times the 1-sample delayed signal (coefficient multiplier 107) and b24 times the 2-sample delayed signal (coefficient multiplier !08). and an output adder 118 that adds to the result (adder 119).

The last secondary section that is connected to this intermediate secondary section has a coefficient of 3
input multiplier 109, input adder 120, two 1-sample delay elements 127, 128, feedback coefficient b3
1 coefficient multiplier 110, feedback coefficient b32 coefficient multiplier 111, feedback signal forming adder 12
1, coefficient multiplier b33 of forward coefficient b33. It consists of a coefficient multiplier b34 for the forward coefficient b34, an adder 123 for forming a delayed forward signal, and an output adder 122 for forming the final output of the digital filter.

The frequency response characteristic of the digital filter with the configuration shown in Figure 3 is 1
Since it is a product of the f-characteristic of the next section, the f-characteristic of the intermediate quadratic section, and the f-characteristic of the final quadratic section, it is possible to make the coefficients between sections independent, but in reality, the desired Due to the design practice of approximating the cutoff characteristic of It was a characteristic form.

In contrast, the present invention aims to provide a digital filter with a plurality of cutoff regions (cutoff region and transient region between the passband and stopband), and in this embodiment, B in FIG. As shown, a composite characteristic with two cutoffs is considered as the LPF characteristic. therefore,
two cutoff frequencies, i.e., a first cutoff frequency fcl, which mainly has the function of preventing aliasing noise;
It has a second cutoff frequency fc2 which is lower than the first cutoff frequency fcl and mainly plays a function of tone processing, and the range below the second cutoff frequency fC2 basically becomes a pass band.

In order to provide this composite LPF characteristic to the 5th-order digital filter shown in FIG. 4, it is preferable to define the coefficients of each part as follows.

However, R = resonance coefficient 0≦R<1fs = sampling frequency fc2 = fcl XC As seen from equation (1), the A parameter depends on the first cutoff frequency fcl, while the AI parameter
It depends on the cutoff frequency fc2. And the A parameter is the feed pack coefficient b2 of the intermediate quadratic section.
1. b22, input coefficient 2 and final quadratic section feedback coefficient b31. b32, the input coefficient is concerned with the determination of 3, so that these filter sections apply the first cutoff frequency to the signal. Also, the AI parameters are the coefficient b11 of the primary section. ! -Kl, so this first-order section has an effect on the second cut-off frequency of the signal. The resonance coefficient R is also incorporated into and acts on the intermediate quadratic section.

Figure 5 shows the change in frequency response characteristics when different resonance coefficients R are given to characteristic type B in Figure 4.The resonance effect occurs approximately at the first cutoff frequency, and emphasizes the signal component of

FIG. 6 is a diagram illustrating the relationship between the first cutoff frequency fc1 and the performance pitch, and this relationship is given by calculations shown in the flow described later.

Generally, when the pitch of a natural musical tone becomes lower, its harmonic components also move to lower frequencies. On the other hand, due to the generation of performance pitch (playback pitch) data from sampled sound data such as PCM data, spectral components (noise) that are not present in the original sound occur, and especially when reproduced by simple sampling sound data interpolation. When generating data, the effect is significant, and aliasing (aliasing noise) occurs in high frequencies (
Components (higher than the Nyquist frequency) are folded back to the lower frequencies, causing unnatural musical distortion. Considering these points, it is desirable to lower the first cutoff frequency for aliasing prevention when the playback pitch is lower than the original sound.Following this approach, in Figure 6, the playback pitch is 90% or less than the original sound pitch. In this case, the first cutoff frequency fc1 is lowered in proportion to the reproduction pitch. At the playback pitch of 90%, which is the break point, the first
The cutoff frequency fcl is 90% of half the sampling frequency fs (Nyquist frequency).

As mentioned above, another feature of the embodiment is that the second cutoff frequency for tone processing changes depending on the first cutoff frequency for preventing aliasing noise. The cutoff region formed between the first cutoff frequency and the second cutoff frequency influences the timbre spectrum, and if the ratio between the two is constant, the pitch feeling will change depending on the change in the playback pitch. It is thought that the timbre feeling basically remains unchanged. Here, it is meaningful to specify the user-programmable ratio of both cutoff frequencies from the input device. Furthermore, since the spectrum of a musical tone changes over time, it is desirable to have a means for realizing this, and in the embodiment, this is expressed as a time-varying parameter EGF (for example.

This can be stored in the ROM 30 in FIG. 2 as data corresponding to each tone color.

FIG. 7 is an example of the time-varying parameter EGF, and the 8th
The figure shows the change in the second cutoff frequency fc2 with respect to time when the time-varying parameter EGF and the user programmable ratio are set to 0.8, 0.8, and 0.4, respectively, and the gSl cutoff frequency is shown by a horizontal dotted line. fcl
This is shown together with In this Figure 8, fc2
is assumed to be given by f C2 = f clXCXEGF (the same applies to the flow described later).

The operation of the embodiment will be described below according to the flows shown in FIGS. 9 to 13. Both flows periodically
Executed by 0.

In the numeric keypad scan flow (9th S), the CPU 20 checks the input from the numeric keypad 50B (9-1), and calculates the ratio C1.
Alternatively, if the resonance parameter H is input, it is stored in the RAM 40 (9-2).

In the keyboard scan flow (Figure 1θ), CPU2
0 reads the status of the keyboard 50A (10-1),
If there is a change in the state (lO-2), it is necessary to determine whether it is a key press or a key release (1O-3), and when a key release is detected, simply set the gate flag GT to the value "0" indicating that the key is being released. and (10-
7) Exit the flow, but when detecting a key press, calculate the relative height FT of the performance pitch to the original pitch, and set the gate flag GT to the value "l" indicating that the key is being pressed (
10-5), and transfers this relative pitch data to the PCM sound source 90 and sets its internal PCM interpolation playback logic (lO-4), whereby the interpolation playback logic (not shown) of the PCM sound source 90 is set. PCM waveform memory (rj
gJ (not shown) is accessed to generate a string of musical sound waveform data having performance pitches. Furthermore, CPU20 is ω-GEN
Call the routine of IO-6 (Fig. 11) and determine whether the performance relative pitch FT' is 90% or more of the original pitch (1l-1), and if so, set the parameter ω by ω=π×0.3. 1l-2), if it is less than 90%, ω=
The parameter ω is calculated by πXPT. Here, the parameter ω has been mentioned above, and is therefore a parameter that determines the first cutoff frequency fcl of the LPF digital filter. Therefore, according to the flow shown in Figure 11,
The characteristics of the first cutoff frequency depending on the performance pitch as shown in FIG. 6 are given. Moreover, this realizes the first cutoff frequency calculation means 4 described in FIG. 1.

In addition, 70 of filter parameter calculation (filter control)
- (FIG. 12), the CPU 200 calculates the time i a harameter (filter Envelope parameter) Calculate the current value of EGF 12-1
), the parameter ωl is calculated by ωl=ωXCXEGF (12-2), where the parameter ω1 exists, and is therefore a parameter that determines the second cutoff frequency fe2 of the LPF digital filter. As a result, the second cutoff frequency parameter calculation means 7 described in FIG. 1 is realized, and the time-dependent and ratio-dependent properties of the second cutoff frequency as illustrated in FIG. 8 are provided. . Then, the CPU 20 calculates the A parameter and the A1 parameter from the first cutoff frequency parameter ω and the second cutoff frequency parameter ωl.
2-3), from this and the user-specified resonance parameter R stored in the RAM 40, as shown in equation (1),
Calculate a set of filter coefficients shown in each coefficient multiplier in FIG. 3 (12-4), transfer the set to the signal processing device 180 (12-5), and signal processing device W! 80 enables dynamic real-time filtering processing.

Also, in the amplifier envelope (volume control) flow, CP
U20 uses the value of gate flag GT and time data from note-on or time data from note-off, etc.
Amplifier envelope EG that controls the volume of M playback data
Calculate A (13-1) and transfer it to PCM sound @70.

[Modifications] This concludes the description of the embodiments, but various modifications and changes are possible within the scope of the present invention.

For example, the order of the LPF digital filter may be other than 5, and even if the performance pitch (playback pitch) changes after note-on, the LPF digital filter
Characteristics including the cutoff frequency of the digital filter circuit can be sequentially controlled.

Furthermore, the signal processing device 80 and the PCM sound source 70 may be configured to be capable of polyphonic support by time division multiplexing.

[Effect of the invention] Finally, the effects of the invention described in the claims will be described.

According to claim Jjil, when reproducing data having a reproduction pitch different from the pitch of the S original sound based on the original sound sampling data stored in the sampling data storage means, the LPF digital filter is used based on the relationship between the a sound pitch and the reproduction pitch. Since the cutoff frequency for the reproduced data in the circuit means is controlled, good sound with little aliasing noise can be obtained regardless of the pitch of the reproduced sound.

Furthermore, in claim 2, the cutoff frequency is controlled to be approximately proportional to the reproduction pitch for reproduction pitches lower than the original sound pitch, so that it is possible to maintain the order of harmonics included in the reproduction sound over a wide pitch range. This allows you to obtain reproduced sound with good timbre reproducibility.

[Brief explanation of the drawing]

FIG. 1 is a functional block diagram of a sound digital filter according to an embodiment of the present invention, FIG. 2 is a hardware block diagram when the embodiment of FIG. 1 is embodied in an electronic keyboard instrument, and FIG. Figure 4 is a diagram showing the frequency response characteristics of the digital filter circuit, Figure 5 is the frequency response characteristic of the digital filter circuit for different resonance parameters. Diagram showing. FIG. 6 is a graph showing the relationship between the performance pitch and the first cutoff frequency of the digital filter circuit, and FIG. 7 is a graph showing an example of the temporal characteristics of the time-varying parameter EGF. FIG. 8 is a graph showing the time dependence of the second cutoff frequency of the digital filter circuit when the time-varying parameters of FIG. 7 are used under several ratio settings. FIG. 9 is a flowchart of numeric keypad scan. Figure 10 is a flowchart for keyboard scanning, Figure 11 is a flowchart for calculating parameters that determine the first cutoff frequency of the digital filter circuit, and Figure 12 is a flowchart for determining the parameters that determine the second cutoff frequency of the digital filter circuit. Flowchart of calculation of filter parameters included. FIG. 13 is a flowchart for calculating an amplifier envelope for a sound source. 4th...! Cutoff frequency parameter calculation means, 7... Second cutoff frequency parameter calculation means, 9... Filter coefficient calculation means, lO.
...Transfer means, 13...LPF digital filter circuit means. 1st Fig. 5 o λ force device Fig. Low round end -r7vtoτhui voice hirorumoshi (
R: Restoration Assist Figure 9 Key Scan Figure 10 Key Additional Scan Figure 11 (1)! lt Snake EGF: To the filter engine〇-Pra°-Yu (OSEG
F: &? ) Figure 12 Fidano V Fumeta Fisho. 13. Loaf to Fig. 7>7#work>

Claims (2)

[Claims] (1) Sampling data storage means for storing data obtained by sampling the original sound, playback pitch designation means for variably instructing the pitch of the playback sound, and having the pitch of the playback sound specified by accessing the sampling data storage means. comprising: reproduction data generation means for generating reproduction data; LPF digital filter circuit means for low-pass filtering reproduction data from the reproduction data generation means; and characteristic control means for controlling characteristics of the LPF digital filter circuit means; A sound characterized in that the characteristic control means has a cutoff frequency determining means for variably determining the cutoff frequency of the LPF digital filter circuit means for the reproduced sound based on the pitch of the original sound and the pitch of the reproduced sound. digital filter device. (2) In the sound digital filter device according to claim 1, the cutoff frequency determining means sets the cutoff frequency substantially proportional to the pitch of the reproduced sound when the pitch of the reproduced sound is lower than the pitch of the original sound. What is claimed is: 1. A digital filter device for sound, comprising means for calculating.