JPH041698A - Musical sound waveform storing method and waveform reader for electronic musical instrument - Google Patents

Abstract

PURPOSE:To efficiently store a musical sound waveform in a data storage area small in capacity by storing address information and allocation information by attaching on the amplification information of waveform data converted to digital values divided into plural steps corresponding to the peak value of the musical sound waveform and of different number of bits. CONSTITUTION:The musical sound waveform for each tone frequency of several kinds of tone colors of a musical instrument is divided into the plural steps corresponding to the peak value, respectively, and divided musical sound waveform data are digital-converted with different number of bits, and an address is attached on each musical sound waveform data, and they are stored in different waveform data storage areas. In other words, in a piano, it is divided into two steps of a piano waveform 1 of 16 bits and a piano waveform 2 of 12 bits, and one piano tone is comprised by reading out the piano waveform 1 and the piano waveform 2 sequentially. The allocation information and the amplitude information of each waveform data divided respectively are stored in an address storage area. In such a way, it is possible to efficiently store the musical sound waveform by effectively using a memory area.

Description

[Detailed description of the invention]

(Industrial Application Field) The present invention relates to electronic musical instruments, and in particular to a musical sound waveform storage method that can store necessary musical sound waveforms without using a large-capacity memory, and a method using this musical sound waveform storage method. The present invention relates to a waveform reading device for an electronic musical instrument.

[Conventional technology 1] Generally, electronic musical instruments use an oscillator or memory element (circuit) as a sound source and do not have a mechanical vibrating part. Musical sound waveforms for each pitch of multiple types of musical instrument tones are stored in memory in advance, and the information stored in this memory is read out by specifying it by operating the keyboard, etc., and musical tones are produced. ing. The waveform data of this electronic musical instrument is stored using a storage area allocation method as shown in FIG. In the figure, D
o O~D○ 15 is the name of the output signal of the waveform memory. In addition, the address indicates the address of the waveform memory, where 0 represents the start address of the waveform memory, and nEN
D indicates the final address of the waveform memory. This ninth
The waveform memory shown in the figure includes various headers, rhythm accompaniment patterns,
It is classified into six types of data: piano waveform, cymbal waveform, drum waveform, and violin waveform. The various headers contain address data such as where each piece of data starts, and the rhythm accompaniment pattern contains performance pattern data for automatic rhythm accompaniment. Piano waveforms, cymbal waveforms, drum waveforms, and violin waveforms contain the waveform data of the musical tones of each instrument, and one address contains pulse-coded waveform data (one sample data) at one point. There is. nR, np, nc, nd, and nv are the start addresses of the rhythm accompaniment pattern, piano waveform, cymbal waveform, drum waveform, and violin waveform, respectively. (Problems to be Solved by the Invention) Recent electronic musical instruments are required to be miniaturized to the extent that one function is not degraded, and they are also required to have multiple functions than before. The capacity of the built-in memory continues to increase.However, due to the demand for miniaturization, the internal volume of electronic musical instruments has become smaller than before, and as a result, the mounting area of electronic circuits has become smaller than before. Therefore, in the memory embedded in the electronic circuit,
It is not possible to use a memory with a large capacity, and the conventional method of allocating a storage area for storing waveform data of an electronic musical instrument has the problem that the required waveform data cannot be sufficiently stored. The present invention was devised in view of the above-mentioned problems of the conventional technology, and its purpose is to efficiently store necessary musical sound waveforms in a small data storage area without using a large capacity memory. An electronic musical instrument capable of reliably and efficiently reading out and synthesizing waveform data of electronic musical instruments that are efficiently stored while saving waveform memory, and generating accurate musical sound waveforms. The present invention aims to provide a waveform readout device. [Means for Solving the Problems] In order to achieve the above object, in the musical sound waveform storage method for an electronic musical instrument of the present invention, a plurality of musical sound waveforms for each pitch of a plurality of types of musical instrument tones are stored in accordance with the peak value. Divide into multiple steps to create multiple waveform data, digitally convert with a different number of bits for each step, and add an address to each waveform data of the musical sound waveform of each instrument tone for each step to create different waveform data. It is stored in the waveform data storage area, and also indicates allocation information indicating the allocation format of the divided waveform data for each pitch of each musical instrument tone and the amplitude of the divided waveform for each pitch of each musical instrument tone. The amplitude information is stored in an address storage area. In order to achieve the above object, the waveform reading device for an electronic musical instrument of the present invention divides the musical waveform for each pitch of a plurality of types of musical instrument tones into a plurality of steps according to the peak value, and Waveform data converted to digital values with different number of bits. Address information indicating a different storage area for each waveform data of musical sound waveforms of each musical instrument tone that differs for each step, allocation information indicating an allocation format of the waveform data to each storage area, and information stored in each storage area. A waveform memory that stores amplitude information indicating the amplitude of each musical sound waveform, and waveform data corresponding to a waveform of a specific pitch of a specific type of musical instrument tone selected by operating a keyboard etc. from the waveform memory. a waveform selection means for selecting; a waveform data output means for reading address information of waveform data selected by a selection signal from the waveform selection means and outputting waveform data based on the address information; and output by the waveform data output means. waveform data extracting means for extracting only necessary waveform data from the waveform data corresponding to the allocation information;
Shifting means for shifting the waveform data extracted by the waveform data extraction means upward or downward according to amplitude information, converting the converted data into a predetermined number of bits, and outputting the same; and waveform data of the predetermined number of bits in the waveform data extraction means. and a tone waveform synthesizing means for synthesizing the waveform data outputted from the shift means to generate and output one tone waveform.

[Effect]

With the above configuration, musical sound waveforms for each pitch of multiple types of musical instrument tones are divided into a plurality of steps according to the peak value, and the divided musical sound waveform data is digitally converted with a different number of bits, and each musical sound waveform is The data is assigned an address and stored in different waveform data storage areas, and the allocation information and amplitude information of each divided waveform data are stored in the address storage area. Therefore, even if a conventional waveform memory is used, the memory area can be used effectively and musical waveforms can be efficiently stored in a small data storage area, and a storage capacity higher than that of the conventional waveform memory can be provided. Also, with the above configuration, if you specify the musical sound waveform for each pitch of multiple types of musical instrument tones by operating the keyboard, etc.,
From a waveform memory in which amplitude information of waveform data that has been divided into a plurality of steps according to the peak value of a musical sound waveform and converted into digital values of different numbers of bits is stored with address information and allocation information. Waveform data of a specific pitch of the selected musical instrument tone is selected, and based on the address information, assignment information, and amplitude information of the selected waveform data, the waveform data is converted and synthesized into a predetermined number of bits and output. Therefore, it is possible to reliably and efficiently read and synthesize the waveform data of the electronic musical instrument, which is efficiently stored while saving the waveform memory, to generate a musical sound waveform.

【Example】

Examples of the present invention will be described below. (First invention of the present application) FIG. 1 shows an embodiment of a musical sound waveform storage method for an electronic musical instrument according to the first invention of the present application. In the figure, DOO-D○ 15 is 1 as in Figure 9.
This is the output signal name of the 6-bit waveform memory. Also, the address is the address of the waveform memory, as in FIG.
O represents the starting address of the waveform memory, and nEND represents the final address of the waveform memory. The waveform memory shown in FIG. 1 is classified into six types of data: various headers, rhythm accompaniment patterns, piano waveforms, cymbal waveforms, drum waveforms, and violin waveforms. The waveform memory shown in FIG. 1 differs from that shown in FIG. 9 in that the waveform of each musical instrument tone is divided into a plurality of steps. That is, the piano is divided into two steps: a 16-bit piano waveform 1 and a 12-bit piano waveform 2. The piano waveform 1 and the piano waveform 2 are continuous data in one hour, and the piano waveform 1,
By sequentially reading out the piano waveform 2, one piano sound is constructed. Also, for cymbals, 16bi
It is divided into three steps: cymbal waveform 1 of t, cymbal waveform 2 of 12 bits, and cymbal waveform 3 of 8 bits. These three waveforms, cymbal waveform 1, cymbal waveform 2, and cymbal waveform 3, are continuous data in one time period, and by reading cymbal waveform 1, cymbal waveform 2, and cymbal waveform 3 in order, the sound of one cymbal can be heard. It consists of Furthermore, the drum waveform is divided into two steps: a 12-bit drum waveform 1 and an 8-bit drum waveform 2. These drum waveform 1 and drum waveform 2 are
This is temporally continuous data, and one drum sound is constructed by reading out drum waveform 1 and drum waveform 2 in order. Furthermore, for the violin, 12 bits
It is divided into two steps: violin waveform 1 of 8 bits and violin waveform 2 of 8 bits. The violin waveform 1 and the violin waveform 2 are sequential data in terms of time, and by reading out the vinyl waveform 1 and the violin waveform 2 in order, one violin sound is constructed. Therefore, the address 0-npl of the waveform memory in FIG.
-1, various headers corresponding to the stored contents are stored. This various header storage area contains
Contains address data such as where each stored data starts. The detailed contents of these various headers are shown in FIG. 2, and include waveform control data for piano, cymbal, drum, and violin, and headers for rhythm accompaniment patterns (head addresses of each rhythm data, etc.). In addition, 16-bit waveform memory addresses nC1 to n
Piano waveform 1 is in p2-1, and addresses nC1 to n
In p2-1, cymbal waveform 1 is 16 bits each.
Then, the address np2~nc
In the memory area of 3-1, the lower 4 bits and the upper 1
It is divided into 2 bits. Rhythm accompaniment pattern 1 is stored in the lower 4 bits in the memory area of addresses np2 to nc3-1. This rhythm accompaniment pattern contains performance pattern data for automatic rhythm accompaniment. Furthermore, the upper 12 bits in the memory area of addresses np2 to nc3-1 contain piano waveform 2, cymbal waveform 2, drum waveform 1, . . . corresponding to each address. Violin waveform 1 is stored. That is, addresses np2~n
Piano waveform 2 is in c2-1, address nc2~n
Cymbal waveform 2 is in dl-1 at address ndl~
Drum waveform 1 is in nvl-1, address nvl~
Violin waveform 1 is in nc3-1, each 12b
It is stored in it. Further, the memory area of addresses nc3 to nEND is divided into upper 8 bits and lower 8 bits. Top 8 in memory area from address nc3 to nEND
A cymbal waveform 3 and a violin waveform 2 are stored in the bit. Further, a drum waveform 2 and a rhythm accompaniment pattern 2 are stored in the lower 8 bits in the memory area from addresses nc3 to nEND. That is, cymbal waveform 3 is stored in the upper 8 bits of addresses nc3 to nv2-1, and violin waveform 2 is stored in the upper 8 bits of addresses nv2 to nEND. Furthermore, the lower 8 bits of addresses nc3 to nR2-1 contain cymbal waveform 3, and the lower 8 bits of addresses nR2 to nR2-1 contain cymbal waveform 3.
Rhythm accompaniment pattern 2 is stored in the lower 8 bits of nEND. FIG. 3 shows the contents of the waveform memory for each musical instrument tone. FIG. 3(A) shows the contents of the waveform control data for each tone color, and includes first to third waveform control data. FIG. 3(B) shows the contents of the n-th waveform control data. In the figure, STn is the start address and the first address of the nth waveform, and ENDn is the end address and the nth waveform.
Each shows the final address of the waveform. Taking the cymbal of this embodiment shown in FIG. 1 as an example, the cymbal's S
r1 (first address of second waveform) is nc2, END
2 (the final address of the second waveform) is ndl-1. For those other than the n-th waveform, 5Tn=O. For example, if we take the drum shown in FIG. 1 as an example, there is no third waveform, so 5T3=O. DFCn is a data format code that indicates in what format the nth waveform of the musical sound waveform of the musical instrument tone is contained. In addition, SCn is a shift
The code indicates whether to expand or reduce the number of bits of the waveform data, and SBn is the shift bit number, which is the magnification when the number of bits of the waveform data is expanded or reduced. This shows that. FIG. 3(C) shows the specific contents of the data format code DFCn shown in FIG. 3(B). That is, DFCn=O is 16 bits, and DFCn=
1 is 12 bits, DFCn=2 is 8 bits upwards,
DFCn=3 indicates 8-bit lower filling. This will be explained using the cymbal shown in FIG. 1 as an example. That is, for the cymbal, the first waveform is recorded in 16 bits, so DFC1=0. Also, the second
Since the waveform is recorded in 12 bits, DFC2=1
becomes. Furthermore, since the third waveform is recorded with 8 bits shifted upwards, DFC3=2. On the other hand, in the drum shown in FIG. 1, the first waveform is recorded in 12 bits, so DFC1=1. Also, the second waveform is 8 bits
Since the data is recorded in the bottom position, DFC2=3. FIG. 3(D) shows the meaning of the shift code SCn shown in FIG. 3(B). In other words, 5Cn=O means that the waveform data is used as is, and 5Cn=1 means that the waveform data is used as is.
5Cn=2 means that it is used after being enlarged to 2#IIn times. FIG. 4 shows values of waveform control data based on the waveform data shown in FIG. 1 in a tabular form. In the table shown in FIG. 4, the upper column shows the piano waveform control data, cymbal waveform control data, drum waveform control data, and violin waveform control data, and the left column shows the contents of the first to third waveform control data. It shows. The table shown in FIG. 4 means that the start address ST1 of the first waveform control data of the piano is the address npl. One line in the diagram of FIG. 4 indicates that any value is acceptable. In this way, when using memory data as waveform data, unnecessary bits can be removed and the amplitude can be adjusted, so waveform data can be packed tightly. In other words, by effectively using the memory area to efficiently store musical sound waveforms in a small data storage area, it is possible to provide free space (A) in the memory area, as shown in FIG. Therefore, according to this embodiment, even if a conventional memory is used, it is possible to store a capacity greater than that of the conventional memory. (Second invention of the present application) FIG. 5 shows an embodiment of a waveform reading device for an electronic musical instrument according to the second invention of the present application. In the figure, reference numeral 1 denotes a waveform reading device that reads out a waveform memory based on designation of musical sound waveforms for each pitch of a plurality of types of musical instrument tones by operating a keyboard or the like. 2 is a CPU that connects the tone color selection section 3 via a pass line;
A keyboard section 4 and a DSP (digital signal processor) 5 are connected. Further, a memory 6, a timer 7, and an A/D converter 8 are connected to the DSP 5 via a pass line. The A/D converter 8 includes an amplifier 9.
However, a speaker 10 is connected to this amplifier 9. The CPU 2 reads a selection of musical instrument tones input from the outside by operating a switch or the like in the tone selection section 3, and selects the tone specified by operating the keyboard section 4, which specifies ON/OFF of the scale and sound according to the program in the memory. A signal specifying a musical sound waveform for a pitch is output to the DSP 5. The DSP 5 selects from the memory 6 digital musical sound data of a musical waveform corresponding to the pitch specified by the operation of the keyboard section 4 based on a specified signal from the CPU 2, synthesizes it into a desired digital waveform, and outputs it. be. The memory 6 stores waveform data shown in FIG. The A/D converter 8 converts the digital waveform synthesized and outputted by the DSP 5 into an analog waveform. The waveform converted into an analog signal is amplified by an amplifier 9 and emitted from a speaker 10. The timer 7 is a circuit that decrements the value set by the DSP 5, and the current value of the timer 7 can be read by the DSP 5. Further, when the value reaches 0, the timer 7 starts decrementing again from the set value. Next, the operation of this embodiment will be explained using the flowchart shown in FIG. In the processing flowchart shown in FIG.
It starts when a sound generation start command is received from P U 2. The CPU 2 sends commands to the DSP 5, such as asking the DSP 5 to play which tone (musical instrument) and which tone (pitch). For example, if you send a command to play an A4 note using a piano tone, the DS
Send to P5. In the following description, it will be assumed that the CPU 2 outputs a command to the DSP 5 to play an A4 note with a piano tone. When the processing flowchart starts, first, in step 100, a tone color number representing a piano is set in TONEREG (tone register) of the internal register of the DSP 5 shown in FIG. Then, in step 110, the DSP 5 sets a timer value corresponding to the piano pitch A4 in the timer 7. This timer 7 operates and outputs waveform data every time it decrements to O (zero).
The pitch of A4 is played.Next, in step 120, the contents of nREG shown in FIG. 7 are initialized to 1. The content n of this nREG indicates that the n-th waveform control data is currently being handled. When the contents of nREG are initialized to 1 in this step 120, in step 130, the DSP 5 recognizes that the designated instrument is a piano by TONEREG, and registers nREG.
recognizes that the current waveform control data is the nth waveform, reads the piano's nth waveform control data from the memory 6, and writes the start address ST to READADREG.
n to ENDREG, end address ENDn to D
The data format code DFCn is set in FCREG, the shift code SCn is set in 5CREG, and the shift bit number SBn is set in 5BREG, respectively, in register 9. When various registers are set in step 130,
In step 140, it is determined whether the contents of READADREG are O (zero). When the start address STn is O, it means that the sound generation has ended, so it is determined whether or not the sound generation has ended. If it is determined in step 140 that the contents of READADREG are not 0 (zero), then in step 150 the data at the address specified by READADREG is read from the memory 6 and set in WDATAREG. This READADREG is a waveform address pointer, and the memory data at the address specified by this is put into WDATAREG. This WDATAR
EG is a 16-bit register, and the data of the memory 6 is exactly stored therein. In step 160, it is determined what the value of DFCREG is, and if the value of DFCREG is O (zero), the process moves to step 200. Further, if it is determined in step 160 that the value of DFCREG is 1, that is, if it is determined that the value is, for example, piano waveform 2 shown in FIG.
Set the bit to O. In other words, for piano waveform 2,
Since rhythm accompaniment pattern 1 is included in the lower 4 bits, the lower 4 bits are cleared. Further, if it is determined in step 160 that the value of DFCREG is 2, that is, for example. If it is determined that the cymbal waveform is 3 as shown in FIG. 1, the lower 8 bits of WDATAREG are
Make it. In other words, for cymbal waveform 3, the lower 8b
Since drum waveform 2 is included in it, clear the lower 8 bits. Further, if it is determined in step 160 that the value of DFCREG is 3, that is, if it is determined that the drum waveform is 2 as shown in FIG. 1, in step 190,
Insert O from the 9th bit of WDATAREG. That is, in the case of drum waveform 2, since cymbal waveform 3 is included in the upper 8 bits, unnecessary data in the upper 8 bits is deleted. In this manner, in steps 170, 180, and 190, the waveform data read from the memory 6 may contain data other than the waveform data that should be generated once, so that data is deleted. The waveform data processed in this way is used to determine the value of 5CREG in step 200. If the value of 5CREG is 0 (zero), the process moves to step 230. Further, if it is determined in step 200 that the value of 5CREG is 1, in step 210, WD
The value of ATAREG is halved by the number of times of the value of 5BREG.
Multiply. Further, if it is determined in step 200 that the value of 5CREG is 2, in step 220, WDATA
The value of REG is doubled by the number of times of the value of 5BREG. That is, for example, for drum waveform 2, 5C2=2.5B
Since 2=4, the value of WDATAREG is multiplied by 16. Next, in step 230, it is determined whether the value of timer 7 is O (zero). In this step 230, wait until the value of the timer 7 becomes O (zero), and immediately after it becomes 0 (zero), in step 240, the musical tone data of WDATAREG is sent to the D/A converter 8 and the amplifier 9 , the sound is emitted through the speaker 10. Then, in step 250, it is determined whether this series of operations has reached the end of the nth waveform. In this step 250, if it is determined that the end of the nth waveform has not been reached, in step 260, the RE
The value of ADADREG (address pointer) is incremented, the process returns to step 150, and the process is repeated again. Further, if it is determined in step 250 that the end of the n-th waveform has been reached, in step 270, nR
It is determined whether EG is 3, that is, whether the waveform currently being processed is the third waveform. At step 250, if it is determined that the waveform currently being processed is the third waveform, the sound generation ends. Further, in this step 250, if it is determined that the currently processed waveform is not the third waveform, that is, if the currently processed waveform is the first waveform or the second waveform, the second waveform or the third waveform To proceed to the waveform, nREG is incremented in step 280 and the processing from step 130 onward is repeated for the next waveform. In this way, in the region where the amplitude of the musical sound waveform becomes small, only the number of IT necessary for that amplitude is used as waveform data. Taking a piano as an example, the piano exhibits maximum amplitude immediately after a key is pressed, and thereafter the amplitude decreases. Therefore, in the case of a piano, 16 bits are used for the large amplitude area immediately after the key is pressed (piano waveform 1 shown in Figure 1), and 12 bits are used for the area where the amplitude decreases thereafter.
Since t is sufficient, only 12 bitL is used (piano waveform 2 shown in Figure 1).In the case of this 12 bit waveform data, the quantization noise with respect to the maximum amplitude is 1/2
” All piano waveform data can be changed to 1 with almost no effect.
It's the same as when you had it with 6 bits. In addition, for the violin, as can be seen from the data in Figure 4, in the case of violin waveform 1, the 12-bit waveform is
4, and in the case of violin waveform 2, convert the 8-bit waveform to 1.
/2' is used. As a result, even though the sound is initially smaller than the maximum amplitude of the piano, no waveform data is wasted. Furthermore, since the drum waveform 2 shown in FIG. 1 is multiplied by 24, the waveform can be stored in the lower 8 bits of the memory. FIG. 8 shows a memory format different from that in FIG. 1. In this embodiment, the last address of each waveform is the waveform END.
Enter the code. Furthermore, enter the waveform END code for each tone. That is, for example, for cymbals, nc
The waveform END code is contained in the addresses 2-1, nc3-1, nc3-1, and nvl-1. When the waveform END code is entered in two consecutive addresses, it means the end of sound generation, and when there is only one waveform END code, it simply indicates the end of the waveform. In this embodiment, since the waveforms of each tone are included continuously, Sr1 and Sr3 can be omitted from the start address. Of course, ENDn can be omitted entirely. Note that ■ to ■ in the figure are rhythm accompaniment pattern data.

【Effect of the invention】

As explained above, the present invention divides musical waveforms for each pitch of a plurality of types of musical instrument tones into a plurality of steps depending on the peak value, and converts the divided musical waveform data into digital data using different numbers of bits. Each musical waveform data is assigned an address and stored in a different waveform data storage area, and the allocation information and amplitude information of each divided waveform data are stored in the address storage area. Therefore, according to the present invention, when using memory data as waveform data, unnecessary bits of the waveform data in the area where the amplitude of the waveform data is small can be deleted and the amplitude can be adjusted and stored, so that the waveform data can be stored tightly. It is possible to save waveform data without reducing quality for waveforms of tones with different amplitudes or waveforms with rapid changes in amplitude. In other words, by effectively using the memory area and efficiently storing musical sound waveforms in a small data storage area, it is possible to free up space in the memory area, and even when using memory with the same memory capacity as conventional memory. It is possible to store more capacity than before. Further, in the present invention, when a musical sound waveform for each pitch of a plurality of types of musical instrument tones is specified by operating a keyboard or the like, the musical sound waveform is divided into a plurality of steps according to the peak value of the musical instrument tone in advance and converted into a digital value with a different number of bits. Waveform data of a specific pitch of the selected instrument tone is selected from the waveform memory that stores the amplitude information of the converted waveform data with address information and assignment information, and the waveform data of the selected waveform data is Based on the address information, allocation information, and amplitude information, the waveform data is converted and synthesized into a predetermined number of bits and output. Therefore, according to the present invention, waveform data of an electronic musical instrument that is efficiently stored by saving waveform memory can be reliably stored.
Furthermore, it is possible to efficiently read out and synthesize the data to generate a musical sound waveform.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing a waveform memory area of a musical sound waveform storage method for an electronic musical instrument according to the first invention of the present application. FIG. 2 is a diagram showing the contents of various headers shown in FIG. 1. Figures 3 (A), (B), (C), and (D) are diagrams showing the contents of the waveform control data for each tone shown in Figure 1, and Figure 4 is the waveform control data shown in Figure 1. FIG. 5 is a block diagram showing an embodiment of the waveform reading device for an electronic musical instrument according to the second invention of the present application, and FIGS. 6(A) and (B)
, (C) is an overall control flowchart of the waveform reading device of the electronic musical instrument shown in FIG. 5, FIG. 7 is a diagram showing the internal registers of the DSP shown in FIG. 5, and FIG. 8 is different from FIG. 1. A diagram showing a memory format. FIG. 9 is a diagram showing a conventional waveform memory area. Waveform readout device PU Keyboard DSP Memory A/D converter Amplifier Speaker

Claims (2)

[Claims] (1) The musical sound waveforms for each pitch of multiple types of musical instrument tones are divided into multiple steps according to the peak value to create multiple waveform data, and each step is digitally converted with a different number of bits, and the The waveform data of the musical sound waveform of each musical instrument tone for each step is assigned an address and stored in a different waveform data storage area, and the waveform data is divided and allocated for each pitch of each musical instrument tone. 1. A musical sound waveform storage method for an electronic musical instrument, characterized in that allocation information indicating the pitch of each tone of each musical instrument and amplitude information indicating the amplitude of the divided waveform for each pitch of each musical instrument tone are stored in an address storage area. (2) The musical waveform for each pitch of multiple types of musical instrument tones is divided into multiple steps according to the peak value, and the waveform data converted to a digital value with a different number of bits for each step, and the waveform data for each step address information indicating a different storage area for each waveform data of musical sound waveforms of different instrument tones; allocation information indicating an allocation format of the waveform data to each storage area; A waveform memory that stores each piece of amplitude information indicating the amplitude of a musical sound waveform, and a waveform that selects from the waveform memory waveform data corresponding to a waveform of a specific pitch of a specific type of musical instrument tone selected by operating a keyboard or the like. a selection means; a waveform data output means for reading address information of waveform data selected by a selection signal from the waveform selection means and outputting waveform data based on the address information; and a waveform output by the waveform data output means. a waveform data extraction means for extracting only necessary waveform data from the data corresponding to the allocation information; and a waveform data extraction means for shifting the waveform data extracted by the waveform data extraction means to upper or lower positions corresponding to the amplitude information to a predetermined number of bits. Shifting means for converting and outputting the waveform data, and musical sound waveform synthesis for generating and outputting one musical sound waveform by synthesizing the waveform data of a predetermined number of bits in the waveform data extraction means and the waveform data output from the shifting means. A waveform reading device for an electronic musical instrument, comprising: means;