JPH067336B2 - Music signal generator - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a musical tone signal generator used in electronic musical instruments, gaming equipment, and the like, and in particular, gradually changes the waveform of a musical tone signal generated over time. The present invention relates to a musical tone signal generator. .

(Prior Art) Conventionally, in this type of apparatus, as shown in Japanese Patent Laid-Open No. 61-107298, after repeatedly outputting the waveform data regarding one waveform a predetermined number of times, the waveform data regarding the next waveform is repeated a predetermined number of times. Waveform data generating means for sequentially switching and outputting waveform data representing a plurality of different waveforms by outputting at predetermined time intervals, and storage means for storing waveform data representing the waveform of a tone signal to be generated are provided. The waveform data from the means and the waveform data from the storage means are input, the difference between the two data is calculated, the difference is multiplied by an appropriate coefficient, and the multiplication result is converted into the waveform data stored in the storage means. By adding repeatedly,
The waveform data stored in the storage means is gradually brought closer to the waveform data from the waveform data generating means, and the gradually changing waveform data is output as a tone signal.

(Problems to be Solved by the Invention) However, in the above-mentioned conventional device, the changing speed of the waveform data, that is, the interpolation speed between the waveform data is determined by the above coefficient, and is included in the generated tone signal. It was not possible to change the interpolation speed for each overtone component. Therefore, for example, a musical sound in which a high-order overtone rises faster than a low-order harmonic overtone, such as a brass tone, or conversely, a low-order overtone rises faster than a high-order overtone, such as a string tone, that is, a certain waveform. When it reaches, there is a problem that it is not possible to better simulate a musical tone in which the changing speed of each harmonic component included in the waveform is different.

The present invention has been devised in view of the above problems,
The object is to provide a musical tone signal generator capable of better simulating musical tones such as brass tones and string tones in which the changing speed of each harmonic component included in the musical tone signal is different. Is to provide.

(Means for Solving Problems) In order to solve the above problems and achieve the object of the present invention,
A structural feature of the present invention is that waveform data representing a plurality of different waveforms are sequentially output over time, and waveform data regarding one waveform is repeatedly output, and then waveform data regarding the next waveform is repeatedly output. Data generation means, storage means for storing waveform data representing the waveform of a tone signal to be generated, waveform data output from the waveform data generation means and waveform data stored in the storage means are input. The waveform data stored in the storage means is output from the waveform data generation means by repeatedly adding the difference data corresponding to the difference between the two input data to the waveform data stored in the storage means. Update means for updating the waveform data so as to gradually approach the waveform data, and repeatedly output the waveform data stored in the storage means. In the musical tone signal generating apparatus adapted to generate a musical tone signal having the waveform represented by the waveform data is to provided a filter in the transmission path of said difference data.

(Operation of the Invention) In the present invention configured as described above, when the waveform data generating means repeatedly generates waveform data different from the previously generated waveform data, the updating means outputs the waveform data generating means. The waveform data stored in the storage means is generated by repeatedly adding the difference data corresponding to the difference between the waveform data and the waveform data stored in the storage means to the waveform data stored in the storage means. Since the updated waveform data is output as a musical tone signal, the waveform of the musical tone signal is gradually changed, and eventually the waveform data generating means is updated. The waveform corresponds to the waveform data generated from. When the waveform data generating means again generates different waveform data, the waveform of the output musical tone signal gradually changes again to the waveform corresponding to the different waveform data as described above. In this way, the waveform of the output musical tone signal gradually changes over time.

Further, when updating the waveform data, the filter provided in the transmission path of the difference data makes the transfer characteristic of the difference data different depending on the frequency component included in the signal represented by the same difference data. The increase / decrease of the generated waveform data differs depending on the frequency component included in the signal represented by the same waveform data. As a result, the output musical tone signal gradually changes while changing its waveform at different speeds according to the frequency components included in the signal.

(Effect of the Invention) As can be understood from the above description of the operation, according to the present invention, since it is possible to change the musical tone signal by changing the changing speed for each frequency component included in the signal,
It becomes possible to better simulate a musical sound having a different changing speed for each overtone component. That is, if the filter is set to the low-pass filter characteristic, the low-order harmonics change faster than the high-order harmonics to better simulate a brass tone, and if the filter is set to the high-pass filter characteristic, the high-order harmonics are set. Can be changed faster than low-order harmonics to better simulate string-type musical tones.

(Example) Hereinafter, an example of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram showing an electronic musical instrument to which the musical tone signal generating apparatus according to the present invention is applied.

This electronic musical instrument has a key switch circuit 11. The key switch circuit 11 is composed of a plurality of key switches corresponding to the respective keys of the keyboard, and the switches are opened and closed according to the pressing and releasing of the keys. A key depression detection circuit 12 is connected to the key switch circuit 11, and the detection circuit 12 detects the key depression / release of each key on the keyboard by detecting the open / closed state of each key switch in the key switch circuit 11. A key code KC representing a key pressed on the keyboard, and a high level "1" (hereinafter simply referred to as "1") when the key is pressed, and a low level "0" when the key is released. The address generator 1 sends a key-on signal KON (hereinafter simply referred to as “0”)
Output to 3.

The address generator 13 outputs the first and second address signals AD1 and AD2 for controlling the reading of the waveform data stored in the waveform memory 14, and as shown in detail in FIG. It has a container 13a.
The note clock generator 13a outputs a note clock signal φn having a frequency m times (m is equal to the number of sampling data of a musical tone waveform for one cycle) of the pitch frequency of the pressed key, based on the key code KC. To do. The note clock signal φn is input to the counter 13b, and the counter 13b counts the note clock signal φn and outputs a count value that repeatedly changes from "0" to "m-1" as the first address signal AD1. . A key-on pulse signal KONP obtained by differentiating the key-on signal KON by a differentiating circuit 13C is supplied to the reset terminal R of the counter 13b.
b is reset when a key is pressed in response to the arrival of the key-on pulse signal KONP.

The carry signal CA generated from the counter 13b is input to the counter 13d every time the counter value of the counter 13b changes from "m-1" to "0".
The counter 13d outputs a counter value that is incremented by 1 every cycle of the first address signal AD1 by counting the carry signal CA. The reset terminal R of the counter 13d is supplied with a key-on pulse signal KONP and a match signal EQ from a comparator 13e described later via an OR circuit OR, and the counter 13d receives the key-on pulse signal KONP or the match signal EQ. In response to the key depression, the reset signal is reset when the coincidence signal EQ is generated from the comparator 13e. The count value output from the counter 13d is supplied to one input of the comparator 13e, and the repeat count value is supplied from the repeat count memory 13f to the other input of the comparator 13e. The repetition number memory 13f stores the number of times the same waveform stored in the waveform memory 14 is repeatedly output for each waveform. The repetition number memory 13f outputs a tone color selection signal output from the tone color selection switch circuit 15 and representing a tone color selected by the performer. Count value from TSW and counter 13g (second address signal AD2)
According to the above, the value of the number of repetitions is output. As a result, the comparator 13 outputs the coincidence signal EQ and resets the counter 13d via the OR circuit OR at the time when the count value from the counter 13d coincides with the repeat count value from the repeat count memory 13f. , The same signal EQ
To the counter 13 via one input of the AND circuit AND
It is designed to supply g. The counter 13g counts the supplied coincidence signal EQ and outputs a count value that is incremented by 1 as the second address signal AD2.
The other input of the AND circuit AND is supplied with the output of the NAND circuit NAND which detects that all the bits of the second address signal AD2 have become "1", and all the bits of the second address signal AD2 are " When it reaches 1 "(in this embodiment, the count value of the counter 13g is 7), the coincidence signal E
Q is not supplied to the counter 13g via the AND circuit AND. This makes the counter 13g
Outputs a second address signal AD2 that increases its count value by 1 in response to the arrival of the coincidence signal EQ and changes from "0" to "7", for example. A key-on pulse signal KONP is supplied to the reset terminal R of the counter 13g, and the counter 13g is reset when a key is pressed.

Further, the address generator 13 has a gate control circuit 13h, and the control circuit 13h has a predetermined count value of the counter 13d, for example, "0, 3, 6 ...", "0,
The gate signal GON which becomes "1" is output at 5, 10, ... "The tone color selection signal TSW from the tone color selection switch circuit 15 is supplied to the gate control circuit 13h, and the gate signal GON is set to" 1 ". The frequency of "1" is changed by the tone color selection signal TSW.

The waveform memory 14 corresponds to n timbres and is designated by the timbre selection signal TSW from the timbre selection switch circuit 15 as waveform data memories 14-1, 14-2 ... 14-n.
Have. Each waveform data memory 14-1, 14-2 ...
14-n are a plurality of areas each designated by the second address signal AD2 (areas E ₀ , E _1, ...
E ₇ ). Areas E ₀ , E ₁ ... E ₇ )
Stores m pieces of sampling data having different waveforms each corresponding to one cycle of a musical tone, and the data is read by the first address signal AD1.

An interpolation circuit 20 is connected to the waveform memory 14, and the circuit 20 includes a subtracter 21, a filter 22, and a gate circuit 2.
3, a multiplier 24, an adder 25, and a shift register 26. The subtractor 21 uses the sampling data supplied from the waveform memory 14 to convert the shift register 2
The sampling data supplied from the final stage of 6 is subtracted, and the difference data resulting from the subtraction is output to the filter 22. The filter 22 has a transfer characteristic determined by the transfer function H (Z), converts the difference data according to this transfer characteristic, and outputs it to the gate circuit 23. This filter 2
The reference numeral 2 is, for example, as shown in FIG. 3, a delay circuit 2 which delays input data by one bit of the note clock signal φn and is reset by the key-on pulse signal KONP.
2a, a well-known FIR type filter 22-1 composed of a multiplier 22b and an adder 22c, or, as shown in FIG. 4, a delay circuit 22d equivalent to the respective circuit elements 22a, 22b and 22c, and a multiplication circuit. Unit 22e and adder 22f
It is possible to use a known IIR type filter 22-2 configured by the above. The frequency characteristics of the filters 22-1 and 22-2 are determined by the filter coefficient a supplied to the multipliers 22b and 22e. This filter coefficient is supplied from the filter coefficient memory 31.
Has coefficient data memories 31-1, 31-2 ... 31-n corresponding to n timbres and designated by the timbre selection signal TSW from the timbre selection switch circuit 15. Each filter coefficient data memory 31-1, 31-2 ... 31
-N stores a plurality of filter coefficient data (8 data a ₀ , a ₁ ... A _{7 in} this embodiment) which are each addressed by the second address signal AD2.

The gate circuit 23 is controlled to be conductive or non-conductive by the gate signal GON from the address generator 13. When the signal GON is "1", the input data from the filter 22 is supplied to the multiplier 24 and the same signal is supplied. When GON is "0", the supply is prohibited. The multiplier 24 multiplies the input data from the gate circuit 23 by the gain coefficient g and supplies it to one input of the adder 25. This gain coefficient g is the gain coefficient memory 3
2, the memory 31 corresponds to n tone colors and the tone color selection signal T from the tone color selection switch 15 is supplied.
A gain coefficient data memory 32-1 designated by SW,
32-2 ... 32-n. Each of the gain coefficient data memories 32-1, 32-2 ... 32-n has a plurality of gain coefficient data (8 data g ₀ , g ₁ ... In this embodiment) addressed by the second address signal AD2.
g ₇ ) is memorized.

The other input of the adder 25 is supplied with the sampling data from the final stage of the shift register 26,
The adder 25 adds the sampling data and the input data from the multiplier 24 and supplies the result to the first stage of the shift register 26. The shift register 26 has m stages that store m pieces of sambling data corresponding to one cycle of the tone color waveform. The sampling data stored in each stage is sequentially shifted by the note clock signal φn and is turned on. It is adapted to be reset by the pulse signal KONP.

A multiplier 33 is connected to the cyst register 26,
The multiplier 33 multiplies the sampling data from the shift register 26 and the envelope waveform data and outputs the result.
This envelope waveform data is stored in the envelope generator 34.
The key generator 34 is supplied from
In response to the key-on signal KON from 2, the envelope waveform data representing the envelope waveform of the musical sound is formed and output. The envelope waveform is controlled by the tone color selection signal TSW from the tone color selection switch circuit 15, and is formed into a shape suitable for the tone color of each tone.

A digital-to-analog converter 35 is converted to the multiplier 33, and the converter 35 converts the digital signal from the multiplier 33 into an analog signal and outputs it to the sound system 36. The sound system 36 is composed of an amplifier, a speaker, etc., and the digital analog converter 3
A musical tone corresponding to the analog signal supplied from 5 is generated.

The operation of the embodiment configured as above will be described. When any key is pressed on the keyboard and the key switch corresponding to the pressed key in the key switch circuit 11 is closed,
The key-depression detection circuit 12 detects this key-depression, and supplies a key code KC representing the depressed key and a key-on signal KON to the address generator 13. In the address generator 13, the note clock generator 13a starts outputting the note clock signal φn based on the supplied key code KC, and the differentiating circuit 13c differentiates the key on signal KON to generate the key on pulse signal KONP. ,
This key-on pulse signal KONP causes the counter 13
b, 13d and 13g are reset. The key-on pulse signal KONP is also supplied to the filter 22 and the shift register 26 of the interpolation circuit 20, and the delay circuit 22a (or 22d) and the shift register 26 in the filter 22 are supplied.
Are reset respectively.

After these resets, the counter 13b in the address generator 13 counts the note clock signal φn and outputs the first address signal AD1 that is synchronized with the note clock signal φn and repeatedly changes from "0" to "m-1". Waveform memory 1
Supply to 4. A tone color selection signal TSW representing a tone color selected by the player is supplied from the tone color selection switch 15 to the waveform memory 14, and a second address signal AD2 set to "0" by the reset is supplied from the counter 13g. The waveform data memory 14-i (i is 1 to n corresponding to the selected tone color is supplied.
Sampling data stored in the first area E ₀ within any of the integers) is sequentially output. This sampling data is supplied to the subtractor 21, which subtracts the sampling data from the shift register 26 from the sampling data and outputs the subtracted difference data to the filter circuit 22. In this case, since the shift register 26 is reset by the key-on pulse signal KONP, the sampling data supplied from the register 26 to the subtractor 21 is “0”, and the difference data supplied to the filter 22 is It is equal to the sampling data from the waveform memory 14. The filter 22 has a filter coefficient data memory 31-i (i is 1 to 1) in the filter coefficient memory 31 specified by the tone color selection signal TSW.
from any integer n), the filter coefficient data a ₀ designated by the second address signal AD2 set to “0” is supplied, and the filter 22 converts the difference data into the filter coefficient data a _0. It is converted according to the transfer characteristic determined by and output to the gate circuit 23. Since the gate circuit 23 is controlled to be conductive by the gate signal GON which initially becomes “1”, the data from the filter 22 is supplied to the multiplier 24.

The multiplier 24 includes a filter coefficient data memory 32 in the gain coefficient memory 32 designated by the tone color selection signal TSW.
The gain coefficient data g ₀ read by the second address signal AD2 set to “0” is supplied from −i (i is an integer of 1 to n), and the multiplier 24 operates as described above. The supplied data is multiplied by the gain coefficient data g ₀ and supplied to one input of the adder 25. In this case, since the sampling data set to “0” as described above is supplied to the other input of the adder 25 from the final stage of the shift register 26, the data from the multiplier 24 is supplied. Is input to the first stage, and the register 26 stores the same data while sequentially shifting in synchronization with the note clock signal φn. In this way, when the counter 13b finishes counting "m-1", the waveform memory 14 outputs the sampling data for one cycle of the waveform to the interpolation circuit 20, and each stage of the shift register 26 has The sampling data for one cycle is stored.

In this state, when the next note clock signal φn is input to the counter 13b, the count value of the counter 13b changes from "m-1" to "0", and thereafter the counter 13b
Similarly to the above case, b repeatedly outputs the first address signal AD1 which changes from "0" to "m-1". When the counter value changes from "m-1" to "0",
The counter 13b outputs the carry signal CA to the counter 13d.
, The count value of the counter 13d becomes "1". In this case, the gate control circuit 13h outputs the gate signal GON which becomes “0” based on the count value set to “1”, so that the gate circuit 23 is non-conductive controlled and one of the adders 25 is controlled. Data representing "0" is supplied to the input.

As a result, the adder 25 supplies the sampling data supplied from the final stage of the shift register 26 to the first stage of the shift register 26 as it is, and the register 26 cyclically stores the stored sampling data. On the other hand, the sampling data output from the final stage of the shift register 26 is also supplied to the multiplier 33, is multiplied by the envelope waveform data supplied from the envelope generator 34 in the multiplier 33, and is supplied to the digital analog converter 35. To be done. The sampling data multiplied by the envelope waveform data is converted into an analog signal by the digital analog converter 35, and this analog signal is supplied to the sound system 36, which generates a musical sound corresponding to this analog signal. start. As described above, while the gate signal GON is “0”, the shift register 26 circulates and stores the sampling data representing the same waveform and the multiplier 3
3 and the digital-to-analog converter 35 continue to output to the sound system 36.
The waveform of the musical sound generated from 6 does not change.

In this state, when the count value of the counter 13d increases and reaches a predetermined value (for example, "3" or "5"), the gate signal GON becomes "1". In this way, the gate signal GO
When N becomes "1", the gate circuit 23 is controlled to be in the conductive state, and similarly to the above, the sampling data read from the first area E ₀ in the waveform data memory 14-i and the final stage of the shift register 26 are read. Difference data from the output sampling data is supplied to one input of the adder 25 via the filter 22, the gate circuit 23, and the multiplier 24. Since the sampling data output from the final stage of the shift register 26 is supplied to the other input of the adder 25, the adder 25 processes the sampling data stored in the shift register 26 by the filter 22. Corrected according to the difference data
The shift register 26 updates the previous sampling data by storing the modified sampling data again. Then, when the count value of the counter 13d again increases and the gate signal GON becomes "0", the gate circuit 23 is controlled to the non-conduction state, and the shift register 26 again circulates and stores the sampling data. After that, the sampling data in the shift register 26 is updated in accordance with the conduction control of the gate circuit 23, and the sampling data in the shift register 26 is stored in the shift register 26 in accordance with the non-conduction control of the gate circuit 23. The sampling data gradually approaches the sampling data stored in the first area E ₀ of the waveform data memory 14-i while being controlled by the characteristics of the filter 22. The sampling data that gradually changes in this way is supplied to the sound system 36 via the multiplier 33 and the digital-analog converter 35 as in the case described above, and the system 36 generates a musical sound corresponding to this sampling data. Therefore, the generated tone is the waveform data memory 1
The waveform represented by the sampling data stored in the first area E ₀ of 4-i gradually approaches according to the characteristics of the filter 22.

Further, after the lapse of time from key depression, the count value of the counter 13d is addressed by the tone color selection signal TSW and the second address signal AD2, and the repeat count memory 1
When the number of repetitions read from 3f becomes equal, the comparator 13e outputs the coincidence signal EQ. This coincidence signal EQ
Thereby, the count 13d is reset and starts counting from "0" again, and the count 13g raises the count value by "1" to set the second address signal AD2 to "1".
Change to. By changing the second address signal AD2, the waveform data memory 14-i repeatedly outputs the sampling data stored in the second area E ₁ . At the same time, the filter coefficient data memory 3
The 1-i and the gain coefficient data memory 32-i output the filter coefficient data a ₁ and the gain coefficient data g ₁ , respectively. In this case as well, the interpolation circuit 20 operates the gate circuit 2 as in the case where the second address signal AD2 is "0".
The sampling data stored in the shift register 26 is updated by updating the sampling data in the shift register 26 according to the conduction control of 3 and storing and holding the sampling data in the shift register 26 according to the non-conduction control of the gate circuit 23. The sampling data stored in the second area E ₁ of the waveform data memory 14-i is gradually brought closer to the sampling data while controlling the characteristics of the filter 22. As a result, the musical sound generated from the sound system 36 is generated from the waveform represented by the sampling data in the first area E ₀ of the waveform data memory 14-i,
According to the characteristics of the filter 22, it gradually approaches the waveform represented by the sampling data in the second area E ₁ . In this way, as the second address signal AD2 increases, the waveform of the musical sound generated by the sound system 36 is changed to the areas E ₀ , E _{0 of} the waveform data memory 14-i.
It changes gradually according to the waveform represented by the sampling data in E ₁ , ..., E ₇ and according to the characteristics of the filter 22. Then, when the second address signal AD2 reaches "7", the NAND circuit NAND outputs "0", so that the updating of the second address signal AD2 by the counter 13g is stopped.

As can be understood from the above description of the operation, according to the above-described embodiment, the waveform of the generated musical tone is gradually changed by the interpolation circuit 20 including the filter 22, and the transfer characteristic of the filter 22 is changed to the selected tone color. Since the update control is performed by using the filter coefficient a from the filter coefficient memory 31 determined by the waveform to be performed, the changing speed can be controlled for each overtone component included in the musical tone when the waveform changes, and the musical tone can be controlled. Better synthesis of is possible. For example,
If the filter 22 is made to have a low-pass filter function when the musical tone rises, the low-order harmonics rise faster than the high-order harmonics when the musical tone rises, so that a brass-type musical tone having this type of characteristic can be better simulated. You can
If the filter 22 has a high-pass function at the start of the musical sound, the higher harmonics rise faster than the lower harmonics when the musical sound rises, so that a string-type musical sound having this type of characteristic can be better simulated. You can

Next, another example of the filter 22 used in the interpolation circuit 20 of the above embodiment will be described. FIG. 5 shows an interpolation circuit 20 using the filter 22 according to this other example. The filter 22 includes delay circuits 22g and 22h, a multiplier 22i and an adder 22 similar to those in the above embodiment.
j, 22k, and the transfer characteristic of the filter 22 is the transfer function H (z) = Z ⁻¹ + a · (1 + Z
^Represented by ^-2 . Z ⁻¹ in this case represents a 1-bit delay, and if the sampling data delayed by 1 bit is used as a reference, the transfer function H (z) is
It becomes equivalent to (Z) = 1 + a · (Z + Z ⁻¹ ), and it can be seen that this filter 22 is a filter having a linear phase characteristic in which the phase does not change depending on the frequency. As described above, since the sampling data delayed by 1 bit is used as a reference, that is, the output of the delay circuit 22g is used as a reference, in this modification, the shift register 26 is provided in the "m-1" stage shift register 26a and the "1" stage And the shift register 26b, the other input of the adder 25 is supplied with the sampling data from the shift register 26b, and the subtraction input of the subtractor 21 is the shift data which is one bit before the sampling data. The output of the register 26a is supplied. The rest of the circuit is the same as in the above embodiment. By configuring the filter 22 in this way, it is possible to change the waveform of the generated musical sound by changing the time for each harmonic component, without changing the phase for each harmonic component of the musical sound signal.

Although the outputs of the shift registers 26 and 26b are guided to the multiplier 33 in the above-described embodiments and modifications, the inputs to the shift registers 26 and 26b are indicated by the broken lines in FIGS. 1 and 5. May be guided to the multiplier 33. By doing so, the musical tone can be generated earlier by one wave of the musical tone waveform than in the above embodiment. In the above embodiment, the filter coefficient a and the gain coefficient g are discretely changed and controlled by the filter coefficient memory 31 and the gain coefficient memory 32, but both the coefficients a and g may be continuously changed. You may In this case, instead of both the memories 31 and 32, the waveform shape is determined by the tone color selection signal TWS and the key-on signal KON is generated.
It suffices to use a function generator that generates a waveform signal that is activated by the above and continuously changes with time. Further, in the above embodiment, the waveform data is generated by reading the waveform data stored in the waveform memory 14, but the waveform data may be generated by a method such as calculation and oscillation. Good. Further, in the above embodiment, the shift register 26 is used as a means for storing the waveform data representing the waveform of the musical tone signal to be generated. However, as a means for storing the waveform data, the data is stored by the address signal. A writable memory (R that controls writing and reading of data by specifying a position
AM) may be used.

[Brief description of drawings]

FIG. 1 is a block diagram of an electronic musical instrument to which a tone signal generator according to an embodiment of the present invention is applied, and FIG. 2 is a diagram showing a detailed example of the address generator of FIG. 1, FIG. 3, and FIG. FIG. 5 is a diagram showing an example of the filter of FIG. 1, and FIG. 5 is a diagram showing a modification of the interpolation circuit of FIG. Explanation of symbols 13 ... Address generator, 14 ... Waveform memory, 20 ...
... Interpolator, 21 ... Subtractor, 22, 22-1, 22-
2 ... Filter, 23 ... Gate circuit, 24 ... Multiplier, 25 ... Adder, 26 ... Shift register, 31 ...
... filter coefficient memory, 32 ... gain coefficient memory.

Claims (1)

[Claims] 1. Waveform data generating means for sequentially outputting waveform data representing a plurality of different waveforms over time, and repeatedly outputting waveform data for one waveform and then repeatedly outputting waveform data for the next waveform. Storage means for storing waveform data representing the waveform of the tone signal to be generated, and waveform data output from the waveform data generation means and waveform data stored in the storage means are input and both input By repeatedly adding the difference data corresponding to the data difference to the waveform data stored in the storage means, the waveform data stored in the storage means is gradually changed to the waveform data output from the waveform data generating means. Update means for updating the waveform data stored in the storage means to repeatedly output the waveform data. In the musical tone signal generating apparatus adapted to generate a musical tone signal having the waveform represented by data, the musical tone signal generating apparatus characterized in that a filter in the transmission path of said difference data.