JP2797888B2 - Music synthesizer - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a tone synthesizer for synthesizing a tone signal having a formant characteristic such as a wind instrument sound or a human voice, and more particularly to a subtle temporal variation of the formant characteristic, that is, fluctuation. And other random fluctuation control with a simple configuration.

[0002]

2. Description of the Related Art Various techniques for synthesizing formant sounds in electronic musical instruments and other electronic musical sound synthesizing systems have been considered. For example, in JP-A-2-254497, a sound having a formant characteristic is obtained by AM (amplitude modulation) of a waveform function signal corresponding to a desired formant center frequency by a time window function signal corresponding to a desired tone pitch. At the desired tone pitch. In this case, the desired formant characteristics are:
Basically, it can be created by appropriately setting a parameter for setting the formant center frequency and a parameter for setting the level of the formant. That is, a formant sound is synthesized by executing a predetermined calculation algorithm using these parameters.

[0003]

By the way, in natural musical instrument sounds and natural human voice sounds, the formant characteristics are not invariable, and the formant center frequency and level fluctuate subtly with time. When trying to simulate such subtle variations in formant characteristics in a conventional formant sound synthesizer, it is necessary to store parameter data such as the formant center frequency and level in the memory corresponding to the time-varying characteristics. However, there is a problem that the memory is increased.

[0004] In addition, since the fluctuation of the formant characteristics observed in natural musical instrument sounds and natural human voice sounds has strong randomness,
No matter how many parameters are stored in memory,
There was a limit to getting a natural feeling. For example, considering a chorus based on human voices, not only the differences in formant characteristics of individuals, but also the sounding timing and volume of each person are different for each sounding opportunity, so parameter data covering all of those differences is It is almost impossible to store it in memory. The present invention has been made in view of the above points, and an object of the present invention is to provide a musical sound synthesizer capable of realizing minute temporal fluctuation of formant characteristics, that is, fluctuation and other random control with a simple configuration. .

[0005]

Means for Solving the Problems] tone synthesis apparatus of the present invention, generates a formant sound synthesizing means for synthesizing a tone signal having a plurality of <br/> formants based on the parameter for the formant synthesis, a noise signal a noise generating means, and controlling said parameter by said noise signal,
Put the tone signals synthesized by the formant sound synthesis unit
Control means for performing random shift control of each formant frequency of a plurality of formants independently of each other.
Are not shifted beyond the specified limits for each
It is characterized by doing so.

[0006]

By controlling the parameters for formant synthesis by the noise signal, the formant characteristics of the tone signal synthesized by the formant sound synthesizer can be controlled at random. Therefore, there is no need to store a large number of parameters for formant synthesis in response to subtle variations in formant characteristics, and moreover, temporally subtle variations in formant characteristics, such as fluctuations and other randomness, of a more natural feel. Time variation control can be realized with a simple configuration, and it is excellent in imitating the wind instrument tone and human voice of natural instruments. In addition , not only the random time-varying control but also other random controls are possible. For example,
By controlling the above parameters with a random value for each pronunciation opportunity, it is possible to give a sensation such as a slight shift in each pronunciation opportunity in the human voice chorus. In particular,
According to the invention of the present application, the ease of synthesizing
Each formant cycle of multiple formants in a sound signal
Since the wave number is controlled independently by random shift,
While being able to simulate human voices faithfully,
At that time, the prescribed limit for each formant
As the shift is not shifted further,
-Personality (personal information) is incorrect due to random control
It is possible to prevent a situation that is impaired from hope. example
For example, in a natural human voice, the second formant
Shift to the side to some extent, or change the third formant
After a certain amount of shifting to the low frequency side, the par
It is known that sonarity is lost
In consideration of the above, the random control by noise
The formant is shifted to higher frequencies beyond a certain limit.
Or the third formant is lower than a certain limit
Can not be shifted up
Noh.

[0007]

Embodiments of the present invention will be described below in detail with reference to the accompanying drawings. In the embodiment shown in FIG. 1, the formant sound synthesizing section 10 synthesizes a musical tone having a fixed formant characteristic such as a human voice or a wind instrument sound, and includes various arithmetic means and a waveform data table. In this embodiment, one musical tone is synthesized by four fixed formants.

A formant data supply unit 11 generates various parameters for formant synthesis and supplies them to the formant sound synthesis unit 10 in order to realize the formant characteristics corresponding to the timbre specified by the timbre specification unit 12. Things. As an example, parameters for formant synthesis include a frequency parameter F for setting a formant center frequency, a level parameter L for setting its level (a peak level of a formant envelope), and a parameter for setting a bandwidth. K and
A parameter S for setting the spread shape (skirt characteristic) of the skirt of the formant envelope is composed of these parameters F1 to F for each of the four fixed formants.
4, L1 to L4, K1 to K4, and S1 to S4.
The formant data supply unit 11 has, for example, a memory storing these parameters corresponding to each tone color that can be designated by the tone color designation unit 12, and reads out the parameters corresponding to the designated tone color.

In this embodiment, the frequency parameters F1 to F1
F4 and the level parameters L1 to L4 are multiplied by the multipliers 13 and 14
Are controlled in accordance with the noise signal, and the controlled parameters F1 'to F4' and L1 'to L4' are
0 is supplied. The pitch designation unit 15
It generates a key-on signal KON for instructing generation of a musical tone and pitch data KC for designating a pitch of a musical tone to be generated, and includes performance operation means such as a keyboard or performance data input means.

As the tone synthesis method in the formant sound synthesis unit 10, any method can be used as long as it can synthesize a tone signal having desired formant characteristics based on parameters. For example, Japanese Patent Application Laid-Open No. 2-2
As disclosed in Japanese Patent No. 54497, a method of amplitude-modulating a waveform function signal corresponding to a desired formant center frequency by a time window function signal corresponding to a desired musical tone pitch may be used. FIG. 2 shows a schematic configuration of the formant sound synthesizer 10 employing such an AM system.

Referring briefly to FIG. 2, the window function generator 16 sets the pitch data KC, the parameter K for setting the bandwidth, and the spread shape (skirt characteristic) of the skirt of the formant envelope. And generates a time window function signal corresponding to a desired tone pitch. The repetition period of the time window signal is determined by the pitch data KC, the width of the time window is determined by the parameter K, and the waveform shape of the time window is determined by the parameter S. The sine wave generator 17 generates a sine wave signal having this frequency based on the parameter F for setting the formant center frequency (this is not limited to a sine wave and may be another arbitrary waveform function). Multiplier 1
At 8, the sine wave signal generated from the sine wave generator 17 is amplitude-modulated by the time window function signal generated from the window function generator 16. The envelope generator 19 generates an envelope signal for setting the time change of the formant level in response to the key-on signal, and the overall level of the envelope signal is set by a level parameter L. The level setting envelope signal generated by the envelope generator 19 is supplied to the multiplier 20 and is multiplied by the output signal of the multiplier 18. In the case where the level parameter L is not changed with time in an envelope shape, the level parameter L is directly input to the multiplier 20 without providing the envelope generator 19.

Thus, a signal having a fixed formant characteristic determined by each of the parameters F, L, K, and S is obtained from the multiplier 20. FIG. 2 shows a basic configuration for synthesizing one fixed formant. Such a basic configuration is used in a time-division manner for each formant, or a plurality of such basic configurations are provided in parallel, and a parameter F corresponding to each formant is used.
By performing the above calculations using 1 'to F4', L1 'to L4', K1 to K4, and S1 to S4, partial sound signals corresponding to the respective formants can be obtained from the multiplier 20. become. By adding and summing the partial sound signals corresponding to these formants, one tone signal having a fixed formant characteristic composed of four formants is obtained. FIG. 3 shows such four formants f
1, the spectral envelopes of f1, f2, f3 and f4 are schematically shown.

Returning to FIG. 1, the tone signal synthesized by the formant sound synthesizing section 10 is subjected to appropriate processing such as the provision of a volume amplitude envelope inside the formant sound synthesizing section 10, and then subjected to digital / analog conversion. Then, it is provided to the sound system 21 and is acoustically pronounced. Next, control of parameters for formant synthesis according to the noise signal will be described. The noise data memory 22 stores noise waveform data. In this example,
In order to realize random fluctuations, instead of white noise or a completely random signal as a noise signal, “1/1 /
The noise data memory 22 stores noise waveform data such as "f fluctuation waveform" and "fluctuation waveform" in which data at a certain address is restricted by the value of the data at the address before it.

The read address signal generator 23 includes parameters F1 to F4 to be controlled by the noise signal.
An address signal for reading the noise waveform is generated for each of L1 to L4, and the address is input to the noise data memory 22. For each of the parameters F1 to F4, L1 to L4, the initial value and the address change rate of the noise waveform reading address signal can be made different, whereby different noise signal data can be generated. I can do it. The noise speed setting unit 24
A setting operator group including a slide volume or the like for enabling a player to freely variably set a change rate (speed) of an address signal for reading a noise waveform for each of the parameters F1 to F4 and L1 to L4. It is. As the change rate of the address signal is increased, the noise waveform data in the noise data memory 22 is skipped and read. Since the noise waveform data is a “fluctuation waveform” as described above, the width of change is increased by skipping reading, and the degree of noise is increased. As described above, the “fluctuation waveform” is used as the noise waveform data stored in the noise data memory 22.
, The degree of noise can be controlled by controlling the change rate of the read address.

In the noise data memory 22, each of the parameters F1 to F4, L1
The noise waveform data is read out corresponding to L4.
This reading is performed by, for example, the common noise data memory 22.
Is performed in a time sharing manner. The read noise waveform data is appropriately level-controlled by the multipliers 25 and 26, respectively, and then input to the multipliers 13 and 14, where the frequency parameter F1 supplied from the formant data supplier 11 is supplied.
To F4 and the level parameters L1 to L4. Thus, in the multipliers 13 and 14, the frequency parameters F1 to F4 and the level parameters L1 to L4 are controlled with subtle and random fluctuations according to different noise signal data. Thus, the parameters F1 'to F4' and L1 'to L4' controlled by the noise are supplied to the formant sound synthesizing section 10 and used in the fixed formant synthesis operation as described above. As a result, the characteristics of each formant obtained by the formant sound synthesizer 10 fluctuate randomly and slightly over time (see the arrow in FIG. 3).
A natural formant sound is obtained. In this case, a more natural feeling can be obtained by using the “fluctuation waveform” as described above as the noise waveform.

The noise level control units 27 and 28 generate data for controlling the level of a noise signal used for parameter control. In order to be able to perform different noise level control between the formant frequency and the formant level, two noise level control units 27 and 28 are provided. The noise level setting unit 29 sets the parameters for each of the parameters F1 to F4, L1 to L4,
This is a group of setting operators including a slide volume and the like so that the player can freely and variably set the noise level. Noise level setting data FNL1 to FNL4 set corresponding to frequency setting parameters F1 to F4
Is input to the noise level control unit 27. The noise level setting data LNL1 to LNL4 set corresponding to the level setting parameters L1 to L4 are
8 is input. The key-on synchro setting unit 30 includes a switch for selecting whether or not to control the noise level as the sounding time elapses (this is referred to as a key-on synchro).
"1" is output as the key-on sync signal SYNC,
When not performed, “0” is output.

FIG. 4 shows an example of the configuration of the noise level control unit 27. The envelope waveform generator 31 generates envelope waveform data of an arbitrary shape according to the input of the key-on signal KON. An example of the envelope waveform data is as shown in FIG. 5, where the minimum value is 0,
The maximum value is composed of decimal data of 1 and has a slope such as attack, decay or the like as appropriate. The selector 32 selects the envelope waveform data generated by the envelope waveform generator 31 when the key-on sync signal SYNC is “1”,
When "0", a constant "1" is selected. The multiplying unit 33 calculates, for each of the noise level setting data FNL1 to FNL4,
This is for multiplying envelope waveform data or constant data “1” output from the selector 32, respectively.
Therefore, when the key-on sync signal SYNC is "1", the noise level setting data FNL1 to FNL4 are temporally variably controlled in accordance with the envelope waveform data generated by the envelope waveform generator 31, but the key-on sync signal SYNC Is "0", the noise level setting data FNL1 to FNL4 do not change with time. The other noise level control unit 28 may have the same configuration.

The noise level setting data for formant frequency output from the noise level control section 27 is supplied to the multiplication section 25, and controls the level of each noise waveform data for formant frequency control. Each noise level setting data for formant level output from the noise level control unit 28 is supplied to the multiplication unit 26, and controls the level of each noise signal data for formant level control.

In the above embodiment, the noise speed setting section 24 and the noise level setting section 29 have been described as being constituted by a group of setting operators. However, the present invention is not limited to this. These noise speeds and noise levels may be set by key scaling. Alternatively, these noise speeds and noise levels may be set by key touch, velocity data, or the like. Further, the setting data of the noise speed and the noise level may be variably controlled in accordance with the real-time operation of an operator such as a pedal or a wheel.
The envelope waveforms used by the noise level control units 27 and 28 may be freely created by the player, or may be appropriately selected from a plurality of types of envelope waveforms.

The noise data memory 22 may store different noise waveform data for each formant. By doing so, it is possible to perform preferable control by using noise waveform data having optimal characteristics for each formant. For example, in a natural human voice, if the second formant is shifted to a higher frequency to some extent, or if the third formant is shifted to a lower frequency to some extent, the personality (personal information) of the human voice is lost. Is known to be. Considering this point, the random control by noise may cause the second formant to be shifted to a higher frequency by a certain limit or more, and the third formant to be shifted to a lower frequency by a certain limit or more. Is not preferred. Therefore, if the noise waveform data stored in the noise data memory 22 is made different for each formant, for example, the noise waveform data for the second formant is stored in advance as being offset to the minus side (low frequency side). In advance, the noise waveform data for the third formant is set in advance on the plus side (high frequency side).
The second formant is shifted more than a certain limit to the high frequency side, or the third formant is shifted more than the certain limit to the low frequency side by storing the offset in Can be prevented from occurring.

The means for generating a noise signal is not limited to a memory, but may be any other means such as a means using a random noise generating circuit or a means for passing the output of a white noise generator through a filter. May do it. The type of noise signal is also the "fluctuation waveform" as described above.
The present invention is not limited to this, and may be a completely random noise signal or the like.

In the above-described embodiment, random time-varying control of parameters is continuously performed according to noise data during sounding time. However, the present invention is not limited to this, and other random control is also possible. For example, by controlling the above parameters with random values for each sounding opportunity, it is possible to give a feeling such as a slight shift for each sounding opportunity in a human voice chorus. For this purpose, for example, in FIG. 4, instead of the envelope waveform generator 31 in the noise level control units 27 and 28, a key-on pulse generator that generates one key-on pulse in synchronization with the rise of the key-on signal KON is provided. Use each noise level setting data FNL
1 to FNL4 and LNL1 to LNL4 are input to the multiplying units 25 and 26 only for a moment at the start of sound generation (key-on). Then, the noise level setting data is multiplied by the noise level setting data with respect to the output noise data of the noise data memory 22, which is generated randomly at the start of the sound generation (at the time of key-on). In this case, the multiplication units 13 and 14
As shown in FIG. 6, latch circuits 34 and 35 may be provided on the output side of the device, and the multiplication results of the multipliers 13 and 14 may be latched in synchronization with the start of sound generation (at the time of key-on) by a key-on pulse KONP. . In this way, the parameters F1 'to F4' and L1 'to L4' to be input to the formant sound synthesizer 10 can be controlled at random for each sounding opportunity.

The pitch designation section 15 is not limited to a keyboard, but may be a wind instrument type breath controller or a guitar type controller. Alternatively, MIDI information may be fetched from outside, and the pitch of the generated sound may be specified based on the pitch information included in the MIDI information. Or,
The pitch of the generated sound may be specified based on the pitch information read from the sequencer for automatic performance. In the above embodiment, both the frequency parameters F1 to F4 and the level parameters L1 to L4 are controlled by noise data, but only one of them may be used.
Further, in the above embodiment, the other parameters K1 to K4 and S1 to S4 for formant synthesis are not controlled by noise data, but they may be controlled by noise data. . The formant data supply unit 11 is not limited to the parameter memory, and may be capable of supplying various parameters for formant synthesis by manual setting by a player. Alternatively, the external sound may be analyzed by voice and created based on this. Various parameters for the synthesized formant may be supplied.

The formant sound synthesizing method in the formant sound synthesizing section 10 is based on the AM using the window function as described above.
The system is not limited to the system, and may be an FM (frequency modulation) system or another system. In other words, any system that performs formant sound synthesis based on parameters for setting a formant frequency, a level, and the like may be used. The formant characteristics synthesized by the formant sound synthesizing unit 10 are not limited to fixed formants, but may be moving formants. For example, if the formant frequency parameters F1 to F4 are key-scaled by the pitch data KC, a moving formant can be obtained. The calculation means used for the noise modulation control is not limited to the multiplication means, but may be an addition / subtraction means. In this case, it is needless to say that multiplier data such as noise data should be expressed not as a ratio but as ± increment / difference data. Further, it is needless to say that various arithmetic means and circuits may be appropriately used in a time-sharing manner. Of course, various calculations and processing may be performed by software processing using a microcomputer or the like. Although the above embodiment has been described as a single-tone generation method, it goes without saying that a double-tone simultaneous generation method is also easy.

[0025]

As described above, according to the present invention, since the parameters for formant synthesis are controlled by the noise signal, a large number of parameters for formant synthesis are stored in response to subtle variations in formant characteristics. There is no need to create or pre-create it, and moreover, it is possible to realize a more subtle temporal change in the formant characteristics, that is, fluctuations and other random time-varying controls, with a simple configuration. Excellent at imitating wind instrument sounds and human voice sounds. Further, not only the random time variation control but also the above-mentioned parameters are controlled by a random value for each sounding opportunity, so that a sensation such as a slight shift for each sounding opportunity can be obtained. In particular, the invention
According to the tone signal synthesized by the formant sound synthesizing means,
Each formant frequency of multiple formants at
Independent random shift control allows for human voice sounds
Can be faithfully simulated, and
Each formant exceeds the specified limit
Because it was not shifted, the personality of human voices
(Personal information) is undesirably damaged by random control
That it can prevent things from happening.
It works.

[Brief description of the drawings]

FIG. 1 is an overall configuration block diagram showing one embodiment of the present invention.

FIG. 2 is a block diagram showing an example of a formant sound synthesizer in FIG. 1;

FIG. 3 is a diagram illustrating a spectrum envelope of a plurality of formants.

FIG. 4 is a block diagram showing an example of a noise level control unit in FIG. 1;

FIG. 5 is a view showing an example of envelope waveform data generated in FIG. 4;

FIG. 6 is a diagram extracting and showing a modification of FIG. 1;

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Formant sound synthesis part 11 Formant data supply part 12 Tone designation part 13, 14, 25, 26 Multiplication part 22 Noise data memory 24 Noise speed setting part 29 Noise level setting part

Claims (1)

(57) [Claims] And 1. A formant sound synthesis means for synthesizing a tone signal having a plurality of formants based on the parameter for the formant synthesis, a noise generating means for generating a noise signal, and controls the parameter by said noise signal, wherein Duplicates in the musical sound signal synthesized by the formant sound synthesizer
And control means for controlling the run <br/> dam shifted independently of each formant frequency number of formants, the formants
Not be shifted more than the specified limit
A tone synthesizer characterized in that: