JPH0573065A - Musical sound signal generation device - Google Patents

Abstract

PURPOSE:To improve the quality of a musical sound by reducing return noises by performing FM modulation processing with high accuracy. CONSTITUTION:The musical signal generation device is equipped with a modulation signal supply means 2 which supplies a modulation signal of frequency in the audio band according to performance information, a carrier signal generating means 3 which generates a carrier signal as a signal to be modulated, a modulation arithmetic means 8 which converts the modulation signal and carrier signal to a sampling frequency of high order and composes a musical sound by the FM modulation processing, and an interpolating means (14) which resamples the FM output of the sampling frequency of high order of the modulation arithmetic means with a sampling frequency of low order and outputs a musical sound composite signal.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a tone signal generator for frequency-modulating a reference waveform signal to generate a tone color tone signal containing a harmonic component.

[0002]

2. Description of the Related Art Conventionally, as a musical tone generator of an electronic musical instrument or the like, a tone signal of a frequency modulation system has been used in which a sinusoidal waveform signal of a predetermined frequency is frequency-modulated to obtain a musical tone signal of rich tone color containing many harmonic components. Generators are known. For example, the device disclosed in JP-A-60-263997 corresponds to this. In this device, the modulated wave data is sequentially input to the shift register, and the FM modulation process is performed by a method in which the read tap is selected according to the modulated wave.

[0003]

By the way, since the technique described in the above-mentioned conventional publication simply performs frequency modulation,
There is a high possibility that a considerable amount of aliasing noise (appropriately referred to as alias) will be included in the output frequency-modulated signal, and there is a problem that the quality of a musical sound may deteriorate. That is, although there is a simple description regarding the sampling frequency conversion in the technique described in the conventional publication, when the reference waveform includes a high frequency component, if the calculation cycle is the same as the sampling frequency of the input signal, the modulation Furthermore, there was a problem that the high frequency component increased, and this became an alias and was annoying. In general, the modulation data is a continuous amount, but in the conventional device, the modulation data is sampled intermittently as shown in FIG.
One of, for example, and outputs the result Y _m + ₁ or Y _m. Therefore, compared with Y ₀ which is the original calculation result, α
There was a problem that an error was generated and noise was heard.

The present invention has been made in view of the above circumstances, and it is an object of the present invention to provide a musical tone signal generator capable of performing FM modulation processing with high precision to reduce aliasing noise and improve the quality of musical tones. Has a purpose.

[0005]

In order to solve the above-mentioned problems, a tone signal generator according to the present invention comprises a modulation signal supplying means for supplying a modulation signal according to performance information and a carrier wave as a modulated signal. A carrier signal generating means for generating a signal, a modulation calculating means for converting the modulated signal and the carrier signal into a high-order sampling frequency, and synthesizing a musical tone by FM modulation processing, and an FM having a high-order sampling frequency in the modulation calculating means. And an interpolating means for re-sampling the output at a low-order sampling frequency and outputting a tone synthesis signal.

[0006]

According to the present invention, both the modulated signal and the carrier signal are converted into the higher order sampling frequency, the tone synthesis is performed by the FM modulation process, and the FM output of the higher order sampling frequency is changed to the lower order sampling frequency. It is resampled and a musical sound synthesis signal is output. Therefore, the FM modulation processing is performed with high accuracy by converting both the modulated wave and the modulated wave into a high sampling frequency, reducing aliasing noise and improving the quality of a musical sound.

[0007]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a diagram showing the configuration of an embodiment of the present invention. In the figure, 1 is a control unit, which is a pitch, tone color,
It receives information necessary for performance such as key depression information and outputs various signals for controlling each circuit. Specifically, the modulated wave generator (modulation signal supply means) 2 and the modulated wave generator (carrier wave signal generation means) according to the performance operation information from the performer, performance information from an automatic performance device, a performance computer, or the like. ) 3, the command signals KONC and KONM for instructing the generation of the waveform are output, and the modulated wave amplitude control EG parameters EGPAR and EGPAR are output to other circuits.
2. Output offset data OFSET that gives a reference point for frequency modulation. The modulated wave generator 2 generates a modulated signal for frequency-modulating the carrier signal, and outputs this as an input modulated wave MW to the interpolation filter 4. Although the modulated signal may be generated by the modulated wave generator 2 itself, the modulated signal may be generated / supplied by a feedback process such as feedback FM. on the other hand,
The modulated wave generator 3 generates a carrier wave signal for FM modulation and outputs it to the interpolation filter 5 as an input carrier wave CW.

The interpolating filters 4 and 5 input the modulated wave of the low speed sampling frequency and the carrier wave of the high speed sampling frequency while limiting the carrier wave to a predetermined frequency band, for example, 20 KHz or less, and the high speed sampling modulated wave MW.
H and high-speed sampling carrier CWH. In this embodiment, oversampling is performed four times at a sampling frequency of 50 KHz to 200 KHz. Since the oversampling technique itself is a well-known technique, its details are omitted. As a filter in this technique, a known structure such as FIR can be applied.
Here, the modulated wave generator 2 and the modulated wave generator 3 operate at the low speed sampling clock φs (50 KHz), and the interpolation filters 4 and 5 operate at the high speed sampling clock φH (2
00 KHz). The internal basic clock φH
H is 800 KHz.

The high-speed sampling modulation wave MWH which is the output of the interpolation filter 4 is supplied to a multiplier 6, which multiplies the high-speed sampling modulation wave MWH by a signal from an envelope generator (EG) 7 and, as a result, Value IN of the integer part of the high-speed sampling modulated wave MEGWH
The T is output to the FM processing unit 8. The numbers on the signal lines in the figure represent the number of bits for digital processing.
The envelope generator 7 supplies the information ENV for controlling the amplitude of the modulated wave to the modulated wave amplitude control EG parameter EGPAR.
Output according to. In this case, as the amplitude of the modulated wave changes, the depth of frequency modulation with respect to the carrier changes, and the output FM wave undergoes a complicated vector change, that is, a tone color change. The FM processor 8 also receives the high-speed sampling carrier CWH, which is the output of the interpolation filter 5. The FM processing unit 8 includes (n + 1) -stage shift register 9
-0 to 9-n, a selector 10, and a read control unit 11. The FM processing unit 8 sequentially loads the high speed sampling carrier CWH into the shift registers 9-0 to 9-n. This is the value IN of the integer part of the modulated wave and the high-speed sampling modulated wave MEGWH after amplitude control in the read controller 11.
According to T (upper 11 bits of MEGWH) and offset data OFSET, [INT-2] -OFSET, [I
NT-1] -OFSET, [INT] -OFSET,
A command is issued to the selector 10 to select the output of the [INT + 1] -OFSET stage shift register, and the selector 10 selects the output of the corresponding shift register to switch the adjacent four modulated wave data to SW (shift register). Output waveform from the output interpolation calculation unit 12
Output to.

On the other hand, of the output from the multiplier 6, the fractional value FRAC (ME) of the high-speed sampling modulated wave MEGWH
The lower 12 bits of GWH) are output to the coefficient generation unit 13, and the coefficient generation unit 13 is configured by a memory such as a ROM. The coefficient generation unit 13 outputs the interpolation coefficient COEF to the output interpolation calculation unit 12 according to the fractional value FRAC of the high-speed sampling modulated wave MEGWH. Specifically, the high-speed sampling modulated wave MEGWH (200KH
The four interpolation coefficients COEF are sequentially supplied to the output interpolation calculation unit 12 for the fractional value FRAC given for each z). The output interpolation calculation unit 12 interpolates the four adjacent waveform data output from the FM processing unit 8 and the corresponding coefficient output from the coefficient generation unit 13 by performing convolution calculation, and outputs the FM output waveform of the high-speed sampling frequency. Data WH
(Interpolated FM output) is obtained and output to the thinning unit 14. A specific technique and configuration for performing convolution calculation of the same kind of musical tone waveform is disclosed in, for example, Japanese Patent Laid-Open No. 6190514, but four waveform data are time-divisionally multiplied for each coefficient. The accumulated result may be output every time. The output interpolation calculation unit 12 is, for example, as shown in FIG.
As shown in (a), the multiplier 21 and the accumulator 22
Composed by. Note that FIG. 2C shows a clock. The thinning unit 14 is composed of a latch, and outputs the FM wave output WH output at the high sampling frequency (200 KHz) to the low sampling frequency (50 K).
Hz) and re-sampling is performed to output the final FM output waveform WS. An envelope generator (EG) 24 as shown in FIG. 2B is provided on the upstream side of the thinning unit 14, and the signal from the envelope generator (EG) 24 is output by the multiplier 23 to the signal from the output interpolation calculation unit 12. It is also possible to multiply by and output the result to the thinning unit 14. The interpolation filters 4, 5, the multiplier 6, the FM
The processing unit 8, the coefficient generation unit 13, and the output interpolation calculation unit 12 constitute the modulation calculation means 15 as a whole.

Next, the operation will be described with reference to the timing chart shown in FIG. First, according to the performance operation information of the keyboard or the like by the performer, the control unit 1 sends an instruction signal KONC to the modulated wave generator 2 and the modulated wave generator 3.
KONM is output, and modulated wave amplitude control EG parameters EGPAR and EGPAR2 and offset data OFSET are output to other circuits. As a result, the input modulated wave MW and the input carrier wave CW from the modulated wave generator 2 and the modulated wave generator 3, respectively, are interpolated by the interpolation filters 4 and 5.
Is output to. In this case, the modulated wave generator 2 and the modulated wave generator 3 use the low-speed sampling clock φs (50 KH
z), and the interpolation filters 4 and 5 operate with the high-speed sampling clock φH (200 KHz).
In the interpolation filters 4 and 5, the modulated wave of the low-speed sampling frequency and the input carrier wave are respectively limited to a predetermined frequency band, for example, 20 KHz or less, while the waveform data of the high-speed sampling frequency, the high-speed sampling modulated wave MWH, and the high-speed sampling carrier wave CWH, respectively. It will be converted.
That is, four times oversampling is performed at a sampling frequency of 50 KHz to 200 KHz.

Next, the high-speed sampling modulated wave MWH is multiplied by the signal ENV (information for controlling the amplitude of the modulated wave) from the envelope generator 7 in the multiplier 6,
Among the multiplication results, high-speed sampling modulated wave MEGWH
The value INT (upper 11 bits of MEGWH) of the integer part is supplied to the FM processing unit 8, and the high-speed sampling modulated wave ME
The fractional value FRAC (lower 12 bits of MEGWH) of GWH is supplied to the coefficient generator 13. On the other hand, the high speed sampling carrier wave CWH is sequentially written in the shift registers 9-0 to 9-n in the FM processing unit 8. Here, the modulation process is preferably performed after the carrier wave has been written to some extent. To realize this, it is easy to give the modulated wave generation instruction signal KONM later than the carrier wave generation instruction signal KONC in time. In other words, the carrier wave should be generated earlier for the modulated wave and written in the buffer.

In this embodiment, a sampling frequency of 200
A waveform buffer capable of writing a carrier wave of 2048 points at KHz is provided, and a waveform for one cycle can be written completely up to a frequency of 100 Hz as follows. 20000/2048 ≈ 97.66 Hz ≈ 100 Hz Conversely, a waveform of 10.24 ms can be written in time. Assuming that there is no problem in FM processing as long as a waveform of at least π / 2, that is, 1/4 cycle is stored in the buffer, the lower limit of the frequency of the carrier wave is about 25 Hz in the buffer amount of this embodiment. 1 for a 100 Hz carrier
The waveform for / 4 cycle can be captured in the buffer if it is 2.5 ms. Considering from the above, it is ideal that the time to advance the carrier wave with respect to the modulated wave at key-on should be determined according to the pitch of the carrier wave. For example, 100Hz
(A3 = around 440 Hz and around G1) Above 1.
If it is less than 5 ms and 100 Hz (G1), the time required to write a 1/4 cycle waveform in the buffer according to the pitch or KC, the generation start time of the carrier wave and the modulated wave (KON
The time difference between C and KONM) may be shifted.

FIG. 4 shows a signal KONC in the control unit 1,
A configuration example of the generation unit of the KONM is shown. In this example, the control unit 1 includes a delay information generating unit 31 and a variable delay unit 32. The key-depression information KON given as the performance information directly instructs the generation of a carrier wave as KONC, and delays the KON by the delay time DLY by the delay information generator 31 according to the pitch information KC.
Further, KONM is output via the variable delay unit 32.

Next, in the processing in the FM processing unit 8, the read control unit 11 [INT-2] according to the value INT of the integer part of the modulated wave and the high-speed sampling modulated wave MEGWH after amplitude control and the offset data OFSET. -OFSET,
[INT-1] -OFSET, [INT] -OFSE
T, [INT + 1] -the output of the shift register of the OFSET stage is selected, and the adjacent four modulated wave data are S
It is output as W to the output interpolation calculation unit 12. On the other hand, the output interpolation calculation unit 12 outputs the high-speed sampling modulated wave MEG.
Fractional value FRA given from WH (200 KHz)
The four interpolation coefficients COEF for C are sequentially supplied to the output interpolation calculation unit 12. Then, in the output interpolation calculation unit 12, the adjacent waveform data from SW is converted into SW (−2), SW
If (-1), SW (0), and SW (1) and the corresponding coefficients are C1, C2, C3, and C4, the FM waveform data WH is WH = C1.SW (-2) + C2.SW ( The convolution operation of −1) + C3 · SW (0) + C4 · SW (1) is performed, and the result is output to the thinning unit 14. In the thinning unit 14, the FM wave output WH output at the high sampling frequency (200 KHz) is
Resampling is performed at the low speed sampling frequency (50 KHz), and the final FM output waveform WS is output.

As described above, in this embodiment, both the modulated wave and the carrier wave have a higher sampling frequency (200 KHz).
Is converted into a sound synthesis signal by FM modulation processing, and the FM output of a higher sampling frequency is resampled at a lower sampling frequency (50 KHz) to output a sound synthesis signal. This will be specifically described. The upper part of the result of the modulated wave Wm (t) × modulation index I is used as the integer part of the data, and which data of the carrier wave is selected is determined. 4 points in the vicinity and modulated wave Wm
(T) × interpolation value Y using the lower part of the result of modulation index I
Is output. This method is shown in FIG. 5, and the convolution operation is expressed by the following equation.

[0017]

[Equation 1]

[0018] In Figure 5 the carrier is converted to 200 KHz, the sample sequence is stored in the shift register, the modulation wave Wm (t) × modulation index I is _{Y n - 2, Y n -} 1, Y n, Y n
+ ₁ state to produce an interpolated value Y using depicted.

In other words, the above processing raises the sampling frequency (Fs) of the input signal to n times (n = 4 in this embodiment) by using the interpolation filters 4 and 5, where 1 /
This means that tone synthesis is performed in a cycle of n times, the synthesis result is returned to the original Fs using the thinning unit 14, and D / A conversion is performed. .. Further, when the modulation is performed, a high precision synthesis operation is realized by the same method as the pitch shift interpolation operation using the decimal part of the modulation data. Where 200KH
Consider why there are no aliases when operating on z. First, assume that there is a carrier wave at 50 KHz as shown in FIG. When modulated, it becomes as shown in (b) of the same figure,
Alias occurs. However, if the method of increasing the sampling frequency (Fs) as in the present embodiment is adopted, FIG.
The spectrum is as shown in (c), and alias does not occur even when modulated (see FIG. 6D). Further, the output obtained through the decimation unit 14 with the cut-off frequency of 20 KHz is as shown in FIG. 6E, and although there are many high frequency components,
This is a musical sound with very few aliases. As described above, in the present embodiment, the FM modulation processing is performed with high accuracy by converting both the modulated wave and the modulated wave into the high sampling frequency, so that alias (folding noise) can be reduced, The quality as a musical sound can be improved.

The FM processing unit 8 in the above embodiment is a system of sequentially storing carrier wave data by using a shift register, but as a modification thereof, for example, one using a RAM as shown in FIG. 7 is also considered. Be done. F shown in FIG.
The M processing unit (modulation wave calculation means) 40 is a timing control unit 4.
1, address counter 42, read address generator 4
3, an address selector 44, a RAM 45, and a latch circuit 46. In the figure, each signal is as follows. WRT: Write timing signal for RAM 45 WAD: Write address RAD: Read address ADDR: Address signal for RAM 45

As shown in the timing chart of the FM processing unit 40 in FIG. 8, the timing control unit 41 controls the timing of the data writing operation based on the clock signal, and the address counter 42 performs the counting operation to perform RA.
The write address signal WAD for M45 is sequentially generated. This is a high sampling frequency (200KH
The update operation is performed in z). The RAM 45 has a structure of 2048 words and the address has a structure of 11 bits.
The input carrier wave data are sequentially stored in the RAM 45 from 0 to
It is repeatedly written at address 2047. On the other hand, the read address generator 43 generates the addresses of the waveform data at the four adjacent points read from the RAM 45 according to the integer part INT of the modulated wave, the write address WAD from the address counter 42 and the address offset OFSET, and the address is used as the read address RAD. Output to the selector 44. Basically, the address output at each high-speed sampling cycle is
WAD + [INT-2] -OFSET, WAD + [IN
T-1] -OFSET, WAD + [INT] -OFSE
There are 4 points of T, WAD + [INT + 1] -OFSET.

A non-linear table 51 as shown in FIG. 9 (a) is installed at the position indicated by a black circle on the signal line in FIG. 1, and its non-linear characteristics are shown in FIGS. 9 (b) and 9 (c). As shown, the waveform may be modified. As a result, it is possible to obtain further variations of musical tones (timbres). Further, although the above-mentioned embodiment is an example of a case of a single tone, the present invention is not limited to this, and for example, parallelization and compounding by time division multiplexing can be easily performed.

Further, in the above embodiment, the final FM output is obtained by four-point interpolation, but the interpolation order is not limited.
Similarly, there is no limitation on the multiple of oversampling (four times in the embodiment). In addition to the EG, the modulated wave may be modulated by operation information of various operators, output waveforms of other further waveform generators, signals extracted from voice, and the like.

[0024]

As described above, according to the present invention, the FM modulation processing can be performed with high precision by converting both the modulated wave and the modulated wave into high sampling frequencies, and the aliasing noise can be reduced. The quality as a musical sound can be improved.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an exemplary embodiment of the present invention.

FIG. 2 is a diagram showing a modified example of the circuit of the embodiment.

FIG. 3 is a timing chart of the operation of the embodiment.

FIG. 4 is a diagram showing a configuration of a control unit of the embodiment.

FIG. 5 is a diagram illustrating a manner of obtaining an interpolation value according to the same embodiment.

FIG. 6 is a diagram for explaining the effect of the same embodiment.

FIG. 7 is a diagram showing a modified example of the FM processing unit of the same embodiment.

8 is a timing chart of the operation of the FM processing unit shown in FIG.

FIG. 9 is a diagram showing a circuit example of a modification of the embodiment.

FIG. 10 is a diagram showing how sampling is performed by a conventional tone signal generator.

[Explanation of symbols]

1: controller, 2: modulated wave generator (modulated signal supply means),
3: Modulated wave generator (carrier signal generating means) 4, 5:
Interpolation filter, 8, 50: FM processing unit, 12: Output interpolation calculation unit, 14: Decimation unit (interpolation unit), 15: Modulation calculation unit

Claims (1)

[Claims] 1. A modulation signal supply means for supplying a modulation signal according to performance information, a carrier signal generation means for generating a carrier signal as a modulated signal, and the modulation signal and the carrier signal at higher sampling frequencies. Modulation calculation means for converting and performing tone synthesis by FM modulation processing, and FM of high-order sampling frequency in the modulation calculation means
A musical tone signal generator comprising: an interpolating unit that resamples an output at a low-order sampling frequency and outputs a musical tone synthetic signal.