JPH0519758A - Electronic musical instrument - Google Patents

Abstract

PURPOSE:To enable a player to set a control function matching a player's taste or style by selecting a fuzzy rule for generating musical tone control parameters by fuzzy inference by the player. CONSTITUTION:A table ROM 13 connected to a CPU 10 is stored with a membership function which uses the condition part arithmetic of fuzzy inference, etc., and a fuzzy inference device 15 is equipped with plural function generators and performs fuzzy inference arithmetic according to an inputted variable. In this case, when the fuzzy rule which is made to function in performance is selected with a rule selection switch 18, the fuzzy inference device 15 uses the selected fuzzy rule to infer a musical tone control parameter according to inputted performance information. Consequently, the player can select the fuzzy rule matching the player's taste or performance style.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an electronic musical instrument for controlling a musical tone signal to be formed on the basis of fuzzy inference.

[0002]

2. Description of the Related Art The present applicant has already proposed a method of controlling musical tone control parameters of an electronic musical instrument by fuzzy reasoning, or detecting a musical technique to control musical tones (Japanese Patent Laid-Open No. 2-146094). , JP-A-2-146596). By generating musical tone control parameters by fuzzy inference and determining performance methods by fuzzy inference, it is possible to perform delicate musical tone control with a simple configuration that considers various performance information in a complex manner. It was

[0003]

However, in the above invention, a plurality of fuzzy rules and membership functions are set in advance, and these cannot be selected or changed later. For this reason, there is a drawback that the player cannot adjust the characteristics of the electronic musical instrument according to his or her own taste and playing style.

It is an object of the present invention to provide an electronic musical instrument which allows a player to freely select a plurality of fuzzy rules set for tone control.

[0005]

According to the present invention, performance information input means for inputting performance information, rule storage means for storing a plurality of fuzzy rules, and a plurality of fuzzy rules stored in the rule storage means are provided. And a fuzzy inference means for deducing a tone control parameter based on the performance information input from the performance information input means by using a fuzzy rule selected by the rule selection means. , Is provided.

[0006]

In the electronic musical instrument of the present invention, the fuzzy rules used in the fuzzy inference means can be selected from the fuzzy rules stored in the rule storage means (rule selecting means). The fuzzy inference means infers a musical tone control parameter using the selected fuzzy rule. This allows the player to freely select a fuzzy rule that suits his or her taste and playing style.

[0007]

1 is a block diagram of an electronic musical instrument which is an embodiment of the present invention. This electronic musical instrument is a digital electronic musical instrument controlled by the CPU 10. The program RO is sent to the CPU 10 via the address / data bus 11.
An M12, a table ROM 13, a RAM 14, a fuzzy inference device 15, a keyboard 16, a membership function editing device 17, a rule selection switch 18, a display device 19 and a sound source 20 are connected. The program ROM 12 stores a program shown in a flowchart described later. The table ROM 13 stores the membership function used for the calculation of the conditional part of the fuzzy inference and the data of the amplitude fluctuation waveform and the pitch fluctuation waveform shown in FIG. RAM
A register 14 is set to temporarily store data generated during performance. As shown in FIG. 2, the fuzzy inference device 15 includes a plurality of function generators and performs fuzzy inference operation based on the input variables. The keyboard 16 has five octave keys (61 keys),
ON / OFF data of each key, velocity data, and aftertouch data can be output. The membership function editing device 17 is a tablet type input device shown in FIG. 5, and the shape of the membership function can be freely set by operating this. The rule selection switch 18 is a switch for selecting a fuzzy rule to be operated at the time of performance or a rule to be edited by the membership function editing device 17, and + key, -key for selecting a rule, ON / OFF of the rule.
It contains an on / off key to specify off and a cursor key to select membership functions. The display device 19 is an LCD matrix display, and displays membership data to be edited as well as setting data during performance (see FIG. 15). The sound source 20 has a tone generator 40 as shown in FIG.
A tone signal is formed by giving various parameters to the tone generator 40.
The tone generator 40 may be of any type such as an FM sound source.

FIG. 2 is a detailed block diagram of the fuzzy inference device 15. The fuzzy inference device 15 has 11 rule calculators 30 (30-1 to 30-11) and 1 m.
An ax calculator 32 and a center of gravity calculator 33 are provided. This fuzzy reasoning device 15 uses the amplitude fluctuation control amount AFL,
Three types of independent fuzzy inference operations, that is, a pitch fluctuation control amount PFL and a noise control amount (noise level NL, noise number NN) are executed in time division. 11
The individual rule calculators 30 calculate different fuzzy rules. Each rule calculator 30 has an internal RAM (FnA, FnP, FnN (n = 1 to 11)) that stores output membership functions for performing the above three types of fuzzy inference.
Is set. When this membership function is rewritten by the edit operation, the function writing device 3
4 writes the new membership function. Further, the condition part measure corresponding to the membership function is input to each rule calculator 30. The condition section measure is
It is calculated by the CPU 10 (see FIGS. 6 and 7) and set in the register 29 (29-1 to 29-11).

The calculation result of the rule calculator 30 is input to the max calculator 32 via the gate 31 (31-1 to 31). The max operator 32 is an operator that performs an operation (max operation) in which a plurality of input output functions are combined. The calculated max function is input to the centroid calculator 33. The centroid calculator 33 finds the centroid of the input function and outputs it as a fuzzy inference value. The inferred value that has been output is temporarily stored in the output register 38, and is sent to the CPU 1 via the bus.
It is taken into 0. Rule calculator 30, max calculator 3
The operations of 2 and the center of gravity calculator 33 are synchronized by a timing signal generated by a timing signal generator 36. Here, the 11 gates 31 inserted between the rule calculator 30 and the max calculator 32 open and close the connections between the rule calculators 30-1 to 30-11 and the max calculator 32, respectively. Opening / closing of each of the gates 31-1 to 31-11 is controlled by each bit of 11-bit data input to the register 35. The data input to the register 35 is set by the rule selection switch 18. That is, the rule selection switch 18 can be used to select which of the rule calculators (fuzzy rules) 30-1 to 30-11 is used to perform fuzzy inference.

The fuzzy inference apparatus 15 having the above-mentioned configuration executes the following operations. First, the rule calculators 30-1 to 30-
11 is a membership function and condition section measure register 29 which are simultaneously set in the respective units in synchronization with the timing signal.
The functions resulting from the min operation with the outputs 1 to 29-11 are sequentially output one by one in the horizontal axis direction. The output function value is input to the gates 31-1 to 31-11, but only the open gates are input to the max calculator 32. The max calculator 32 operates the gate 31 at each timing.
The maximum value of the function output of the fuzzy rule selected and input by is selected and output to the centroid calculator 33. The center-of-gravity calculator 33 accumulates the output of the max calculator 32 in synchronization with the timing signal, and writes the accumulation result at each timing in the memory 37. When the accumulation is completed, ½ of the final accumulated value is calculated (1 bit is shifted down in the actual calculation), and the area where the accumulated value equal to that value is stored is searched from the memory 37. .. The horizontal axis value corresponding to the timing when the accumulated value is obtained is the center of gravity. This barycentric value is written in the output register 38.

FIG. 3 is a block diagram of the sound source 20.
The sound source 20 has a tone generator 40. The tone generator 40 is composed of an FM sound source LSI, and the formed musical tone signal is added to the noise waveform signal generated by the noise waveform generator 52 in the adder 55, and D
The signal is converted into an analog signal by the / A converter 56 and then output to the sound system. Tone generator 40
The cent data, decibel data, waveform number and note-on signal are input to. Further, the noise number, noise level and note-on signal are input to the noise waveform generator 52. The cent data is formed by the key code register 41, the pitch generator 49, the pitch change register 47 and the adder 53. In the key code register 41, the key code of the key that has been keyed on is the CPU 1
Input from 0. The pitch generator 49 converts the key code input to the key code register 41 into numerical data corresponding to the frequency. The pitch fluctuation data generated by the fuzzy inference device 15 is input to the pitch change register 47. This data is numerical data regarding frequency similar to the data generated by the pitch generator 49. These pieces of data are added in the adder 53 to be cent data input to the tone generator 40. The decibel data is generated in the initial touch register 40 by the after touch register 43, the amplitude generator 50, the amplitude change register 48, and the adder 54. The amplitude generator 50 uses the initial touch register 42.
Also, an amplitude value is generated based on the initial touch data and the after touch data input from the after touch register 43. The generated amplitude value is added to the amplitude change data in the adder 54 to become decibel data. Furthermore, the waveform number is the initial touch register 4
2, generated by the after-touch register 43 and the waveform selection signal generator 51. The waveform number is a number indicating the waveform that the tone generator 40 should use. The noise number and noise level data input to the noise waveform generator 52 is input to the noise number register 45 and the noise level register 46 from the CPU 10.

FIG. 4 is a diagram showing the amplitude fluctuation waveform AFW (CNT) and the pitch fluctuation waveform PFW (CNT) stored in the table memory 13. This change waveform is obtained by sampling the rising of a musical tone of an actual brass instrument, and the data for each sampling timing is stored in the table memory 13. When the formation of a musical tone signal is started, the CPU 10 sequentially reads this data and sets it in the pitch change section register 47 and the amplitude change register 48.

FIG. 5 is a view showing a tablet input device used as the membership function editing device. The tablet input device includes a tablet body 50 and a pen 51. Pen 5 in edit mode
By drawing the shape of the membership function on the tablet body 50 using 1, the shape is set as the function shape of the designated membership function.

FIG. 6 shows a fuzzy inference rule for inferring the amplitude fluctuation control amount AFL. In addition, FIG.
Indicates a fuzzy inference rule for inferring the noise control amounts NL and NN. Each rule is inferred based on the initial touch data VEL, the time ΔT from the immediately preceding key-off to the key-on, the time CNT from the start of sound generation, the after-touch data AFT, the key code KCD, the pitch (pitch difference from the previous sound) ΔKCD, etc. It's a rule.

The first rule to the tenth rule for inferring the amplitude fluctuation control amount AFL shown in FIG. 6 are in groups of two rules, each of which is the initial touch data VEL.
Size / smallness and shortness / length of time from the start of pronunciation,
Aftertouch size / Small size, Key code size /
Make inferences based on smallness, pitch size / smallness, and interval time length / shortness. In addition, the eleventh rule makes an inference based on the legato degree (determined by another inference). Register 29 of the fuzzy inference device 15
Condition part measures set to -1 to 29-11 are CPU
Fuzzy measure (affiliation degree) calculated by the antecedent part by 10.
Is. Fuzzy inference of the pitch fluctuation control amount PFL is performed using the same rule as the amplitude fluctuation.

Further, similarly to the rule for inferring the noise control amount in FIG. 7, the size / smallness of the initial touch data VEL, the size / smallness of the key code, the size / smallness of the aftertouch, The inference is performed based on the first rule to the sixth rule and the eleventh rule, and the inference is performed to determine the noise level NL.
Inference is performed to determine the noise stability (noise number NN) according to one rule.

Here, the rule based on the magnitude of the after-touch AFT (fifth rule, sixth rule) and the rule based on the legato degree SL (11th rule) are commonly used for both inferences. Therefore, inference for noise control is executed in the fuzzy inference device 15 for two cycles.

FIG. 8 shows a part of the membership function (first, second, first of the amplitude fluctuation control) for obtaining the condition part measure.
One rule) is illustrated in FIG.

9 to 14 are flowcharts showing the operation of the electronic musical instrument. FIG. 9 shows the main routine. First, immediately after the start, the initial setting operation (n1) is executed.
The initial setting operation includes resetting the register and transmitting preset tone color data. After this, the key processing operation (n2),
The panel switch processing operation (n3) and the other processing operation (n4) are repeatedly executed.

FIG. 10 is a flowchart showing the key-on event routine. First, the data on the key that has been turned on is loaded into the register (n10). The data to be taken in is the key code KCD, the velocity (initial touch) data VEL, the after touch data AFT, and the time ΔT from the last key-off. Next, the time counter CNT is reset (n11), and interruption during the sound generation processing is prohibited (n12). Next, the legato degree of the performance is inferred by fuzzy inference, and the legato degree register SL
(N13). The determination of the legato degree may be performed by the method disclosed in Japanese Patent Application Laid-Open No. 2-146596 filed by the present applicant. Next, the pitch difference between the immediately preceding musical sound and the keyed-on musical sound this time, that is, the pitch is calculated, and ΔKC is calculated.
Set to D (n14). The key code KCD, velocity data VEL, and after-touch data AFT are set in the tone generator registers 41, 42, and 43 (n15). Next, the condition part measure is calculated based on the data and sent to the fuzzy inference device 15 to infer the parameters of the amplitude fluctuation control amount AFL, the pitch fluctuation control amount PFL, and the noise control amounts NL, NN (n16, n17, n18). ). Data of these inference results is taken out (n19), and the amplitude fluctuation control amount AFL and the amplitude fluctuation waveform data A in FIG.
FW (CNT) is multiplied to calculate the amplitude change data AFR, and the pitch fluctuation control amount PFL is multiplied by the pitch fluctuation waveform data PFW (CNT) of FIG. 4B to calculate pitch change data PFR. To do. These data and N
L and NN are sent to the sound source 20 (n20). After these data are set in the sound source 20, a note-on signal is sent (n21). That is, the note-on register (ON
R) 44 is set to 1. After the above operation is completed, the interruption prohibition is released (n22), the key code of the current key-on is set in KOLD (n23), and the operation is completed.

FIG. 11 is a flowchart showing the interrupt processing operation. First, it is determined in n30 whether or not there is an interrupt. The presence / absence of an interrupt (a timer interrupt executed at regular time intervals) is determined by setting / resetting a flag set by an interrupt operation. If there is no interrupt, this operation is not performed and the process directly returns. When there is an interrupt, the value of CNT is incremented by 1 (n31), and it is determined whether or not it is the count end (n32). If it is the count end, the interrupt is prohibited and the operation ends (n33). That is, the interrupt is prohibited when the time for performing the fluctuation control by the interrupt operation has passed. If it is not the count end, capture the after touch data of the key that is turned on and AFT
The data is set in the register (n34), this data is copied to the after-touch register (AR) 43 (n35), and C
A condition part measure (see FIGS. 6 and 7) calculated using NT and AFT is calculated and sent to the fuzzy inference apparatus 15 (n36). The inference output is fetched from the fuzzy inference device (n37), and when the key is on, AFW and PFW are calculated by the same calculation as n20 and sent to the sound source 20 (n3).
9), the amplitude fluctuation waveform AFRW (CNT) at the time of release when the key is off, and the pitch fluctuation waveform PF at the time of release
AFR and PFR are calculated using RW (CNT) and sent to the sound source 20 (n40).

FIG. 12 is a flowchart showing the key-off event routine. The key code of the turned off key is fetched into the key code register KCD (n45), and it is judged whether or not the tone of this key code is being generated (n).
46). If the sound is being generated, the amplitude fluctuation amount AFW at that time
The same value as (CNT) is released (attenuation) amplitude fluctuation amount A
The FRW is searched, and the position of the same value is newly set in CNT (n47).

After that, the note-on signal ONR is reset (n48) and the process returns.

FIG. 13 is a flow chart showing the switch event processing operation. First, the operation mode is set according to the turned-on switch (n51), and the mode screen is displayed on the display (n50). If the operation mode is edit 1, that is, the amplitude fluctuation rule edit, the operation proceeds to n52, if it is edit 2 (pitch fluctuation rule edit), the operation proceeds to n53, and if it is edit 3 (noise control rule edit), n.
Proceed to operation 54. N5 for other operation modes
When the judgment is 1, the process proceeds to n55 and the corresponding process is executed.

FIG. 14 is a flowchart showing the rule editing operation. This operation is n52, n53, n54
Executed in. First, a rule to edit (see FIGS. 6 and 7) is designated (n60). This designation is made with the + and-keys. The on / off selection process of the designated rule is accepted (n61). This operation depends on the on / off switch. Next, the membership function to be edited is designated from the designated rules (n62). This designation is made by moving the cursor using the cursor keys or the like. At this time, the designated function is displayed on the display device 19 as shown in FIG. The membership function editing device 17 is operated to input the shape of the function (n63). This input operation may be performed by drawing the shape of the function by freehand, or by plotting a plurality of points as shown in FIG. n
The on / off data of the rule set in 61 is sent to the register (RX) 35 of the fuzzy inference device 17 (n
64). Further, when the function edited in n63 is the output membership function of the conditional part, the function is sent to the fuzzy inference device 15 and written in the corresponding internal RAM (rule operation unit) F1A to F11N (n65).

By the above operation, the player can freely turn on / off (select) the fuzzy rule and edit the membership function. The membership function editing device 17 is not limited to the tablet input device, and a mouse, a digitizer or the like can be used.

In this embodiment, fuzzy inference is large /
Although it is performed in two small steps, it may be divided into three or more steps. Also, when a fuzzy rule is edited, the new rule is overwritten on the old rule to erase it. However, even if the preset rule is stored in the ROM and the original data is not erased. Good.
In addition, if you make a sound source that can produce several tones at the same time,
A control rule or membership function may be provided for each tone color. Further, in this case, the shared rule and the rule to be separately provided may be divided.

Further, the data used for the fuzzy inference is not limited to the data used in this embodiment, and the operation amount data of the joystick, the pitch bend wheel, etc. may be used. Further, instead of performing fuzzy inference in real time during performance as in the present embodiment, fuzzy inference for all cases is performed and written in the memory during free time, and this memory is accessed during actual performance. The inference result may be read.

[0029]

As described above, according to the electronic musical instrument of the present invention, the player can select the fuzzy rule or the membership function, so that the musical tone can be controlled in accordance with his or her own taste and playing style. ..

[Brief description of drawings]

FIG. 1 is a block diagram of an electronic musical instrument that is an embodiment of the present invention,

FIG. 2 is a diagram showing a configuration of a fuzzy inference device of the electronic musical instrument,

FIG. 3 is a diagram showing a configuration of a sound source of the electronic musical instrument,

FIG. 4 is a diagram showing a pitch fluctuation waveform and an amplitude fluctuation waveform stored in the electronic musical instrument.

FIG. 5 is a view showing a tablet device used in the electronic musical instrument,

FIG. 6 is a diagram showing a fuzzy rule of fuzzy inference performed by the electronic musical instrument;

FIG. 7 is a diagram showing the same fuzzy rule,

FIG. 8 is a diagram showing a membership function used for fuzzy reasoning;

FIG. 9 is a flowchart showing the operation of the electronic musical instrument,

FIG. 10 is a flowchart showing the operation of the electronic musical instrument,

FIG. 11 is a flowchart showing the operation of the electronic musical instrument,

FIG. 12 is a flowchart showing the operation of the electronic musical instrument,

FIG. 13 is a flowchart showing the operation of the electronic musical instrument,

FIG. 14 is a flowchart showing the operation of the electronic musical instrument,

FIG. 15 is a diagram showing an example of a display screen in the edit mode,

FIG. 16 is a diagram showing an input example of a membership function,

[Explanation of symbols]

15-fuzzy inference device, 17-membership function editing device, 18-rule selection switch, 30-rule calculator, 34-function writing device.

Claims (1)

Claims: 1. Performance information input means for inputting performance information; rule storage means for storing a plurality of fuzzy rules; and a plurality of fuzzy rules stored in the rule storage means. A rule selecting means for selecting a rule to be operated, a fuzzy inference means for inferring a musical tone control parameter based on the performance information input from the performance information input means using the fuzzy rule selected by the rule selecting means, An electronic musical instrument characterized by having.