JP2536460B2 - Method and device for generating musical sound - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a musical tone generating method and apparatus for generating a musical tone by reading out waveform information corresponding to a desired musical tone stored in a waveform memory.

[0002]

2. Description of the Related Art Conventionally, as a musical tone generating method for generating a desired musical tone, for example, Japanese Patent Laid-Open No. 52-121313.
As shown in the publication, all waveforms from the start of sounding a desired musical sound (such as a musical instrument sound or a human voice sound) to the end of sounding are stored in a waveform memory, and this stored waveform is read out as it is, A method of generating a musical tone corresponding to a desired musical tone, or storing all waveforms of the rising portion of the desired musical tone and a partial waveform after that in a waveform memory, reading all the stored waveforms of the rising portion, and then continuing. There is a method of generating a musical tone corresponding to the desired musical tone by repeatedly reading out a part of the waveform. Further, by reading the tone waveform stored in the waveform memory in the reverse direction, a sound different from the stored sound can be reproduced and generated (for example, "Keyboard Magazine", September 1983 issue, etc.), and its forward and reverse reading directions can be changed. It may be switched in order and repeatedly played (eg, "Keyboard Magazine", February 1985 issue), or the forward / reverse reading direction may be sequentially switched in some sections of the musical tone continuation section to be repeatedly played and the reading section may be slightly shortened. It is also known to perform aperiodic repetitive reproduction by shifting in sequence (for example, US Pat. No. 4,442,74).
No. 5, etc.).

[0003]

The former musical tone generating methods are widely used in musical tone generating devices such as electronic musical instruments in that they can generate high quality musical tones very close to natural musical instrument sounds and human voice sounds. Has been applied. However, since the stored musical tone waveforms are simply read out in the order of their time arrangement, there is a problem that only fixed musical tones can be generated. On the other hand, according to the latter reverse reading, it is possible to generate a strange sound by reading the stored waveform in the reverse direction of the musical tone waveform time array stored in the memory. It is only monotonic and special, because it simply repeats all forward / reverse inversion readings of the stored waveform, or only repeats forward / reverse inversion readings of a relatively monotonous sustain waveform. There was a limit to the generation of musical sounds. The present invention intends to provide a musical tone generating method and device by further devising the musical tone generating method as described above so as to generate a special musical tone based on the musical tone waveform information stored in the waveform memory. is there.

[0004]

SUMMARY OF THE INVENTION A musical tone generating method according to the present invention is a waveform memory, wherein waveforms relating to a plurality of cycles of a musical tone waveform of a rising portion corresponding to a desired musical tone and a part of a plurality of subsequent musical tone waveforms corresponding to a desired musical tone. In a musical tone generating method in which information is stored and the stored musical waveform information is read to generate the desired musical tone, the waveform information is stored in the waveform memory from the waveform in the direction opposite to the time sequence of the musical tone waveform. After the data is read out according to the array, the waveform information relating to at least some of the tone waveforms of the rising portion is repeatedly read according to the time array in the direction opposite to the time array of the tone waveform. The desired musical tone is generated in a time-reversed manner.

Further, the musical tone generating apparatus according to the present invention stores waveform information relating to a plurality of cycles of the musical tone waveform of the rising portion corresponding to a desired musical tone in the waveform memory and a part of a plurality of subsequent periods of the musical tone waveform. A tone generating device for generating the desired musical tone by reading the stored waveform information, wherein the waveform information stored in the waveform memory is read in the same direction as the time sequence at the time of storage or in the opposite direction. In the reading in the same direction, the waveform information is read out from the waveform memory according to the time arrangement of the musical tone waveform, and then, at least a part of the musical tone waveform of a part of the plurality of cycles thereafter. The waveform information about the musical tone waveform including is repeatedly read according to the time sequence of the musical tone waveform, and the reading in the reverse direction is performed. After the waveform information is read out from the waveform memory according to a time sequence in the direction opposite to the time sequence of the tone waveform, the waveform information relating to at least a part of the tone waveform of the rising part is subsequently read out. The desired tone or the desired tone is provided with a reading means for repeatedly reading in accordance with a time arrangement in the direction opposite to the time arrangement in the tone waveform and a selecting means for selectively designating the reading direction in the reading means. It is characterized in that the tone is inverted in time to selectively generate a musical tone.

The tone generating apparatus according to the present invention further provides a predetermined amplitude envelope to the tone waveform corresponding to the waveform information read from the waveform memory, and is selected and designated by the selecting means. It is provided with an envelope applying means for applying an amplitude envelope of an envelope waveform different depending on the read direction, and the musical tone waveform to which the amplitude envelope is applied is generated as a musical sound by the envelope applying means.

[0007]

According to the tone generating method of the present invention, the waveform information stored in the waveform memory is read out according to the time sequence in the opposite direction to the time sequence of the tone waveform, and then the tone waveform of the rising portion is successively read out. Since the waveform information about at least a part of the musical tone waveform is repeatedly read according to the time sequence in the direction opposite to the time sequence in the musical tone waveform, a desired musical tone is simply inverted in time to generate a musical tone. Not only that, a very special musical tone can be generated by repeatedly reproducing at least a part of the musical tone waveform of the rising portion of the desired musical tone in the opposite direction in time.

In general, the rising portion of a musical tone is a good representation of the characteristics of the musical tone, and the waveform changes drastically. Therefore, by repeatedly reading out the musical tone waveform of such a rising portion in the opposite direction in time, a very special musical tone having a great variety and having the individuality unique to the desired musical tone is generated. be able to. Also,
By repeatedly reading out the musical sound waveform of the rising portion, which is a good representation of the characteristics of the musical sound, in the opposite direction in time,
A sound having a unique character / characteristic to a desired musical sound is stretched and continuously generated, so that the generation of a musical sound having a novel characteristic can be realized. Further, by repeatedly reading out the musical tone waveform at the rising portion where the waveform changes drastically in the opposite direction in time, the repeated portion (sustained portion) of the newly generated musical tone becomes rich in waveform change (timbre change), It is possible to generate extremely complex tones.

Further, according to the tone generating device of the present invention,
As described above, the special backward repetitive read control unique to the present invention and the read control along the same direction (forward direction) in the time array can be arbitrarily selected and designated, so that various musical tone generation can be realized. can do. Further, by providing the generated musical tones with amplitude envelopes having different envelope waveforms according to the selected and designated reading direction, it is possible to realize the generation of musical tones with a greater variety.

[0010]

Embodiments of the present invention will now be described in detail with reference to the accompanying drawings. Although FIG. 1 does not directly correspond to the embodiment of the present invention, it shows a basic configuration example of the backward read control of the waveform memory. Therefore, for convenience of description of the embodiment of the present invention in this specification. , First of all. The musical tone generating apparatus shown in FIG. 1 stores waveform information about all musical tone waveforms from the start to the end of sounding percussion instrument sounds such as drum sounds and Tyco sounds in a waveform memory, and stores this waveform information at the time of storage. The reading is performed in the same direction as the time array or in the opposite direction.

In FIG. 1, 1 is a waveform memory, 2 is an inversion circuit, 3 is an address counter, 4 is a final address detection circuit, 5 is a one-shot circuit, 6 is a sound system with a built-in DA converter. As described above, the waveform information on each sample point of all musical sound waveforms (see (a) of FIG. 3) from the start to the end of the percussion sound is directed from the address "0" to the final address according to the arrangement of the time axis. Are sequentially stored.

On the other hand, the address counter 3 is reset by a start pulse SP generated from the one-shot circuit 5 by turning on the tone generation command switch 7. The address counter 3 generates an address signal AD for accessing the waveform memory 1 by sequentially counting clock pulses φ of a predetermined cycle. As will be described later, the value of the address signal AD reaches the maximum value. When it reaches, this is detected by the final address detection circuit 4, the count operation is prohibited by the final address detection signal of the detection circuit 4, and the operation is stopped in the state of the maximum address value until a new tone generation start command is given. .

Therefore, when the start pulse SP is output from the one-shot circuit 5 and the counter 3 is reset, the counter 3 is released from the count-inhibited state by the output of the final address detection circuit 4 and the clock pulse φ can be counted. Thus, the address signal AD that changes sequentially each time the clock pulse φ is generated is output. This address signal AD passes through the inverting circuit 2 and the waveform memory 1
Of the address signal AD, the inverting circuit 2 supplies the address signal AD as it is to the waveform memory 1 when the special effect sound selection switch 8 is off, and each bit of the address signal AD when the switch 8 is on. Is inverted and supplied to the waveform memory 1.

That is, the inverting circuit 2 includes the switch 8
Is ON, the address signal AD changing from "0" to "final address value" is converted into the address signal AD changing from "final address value" to "0" and output. Therefore, if the switch 8 is off, the waveform information stored in the waveform memory 1 is read from "0" toward the "final address value" (that is, in the same direction as the time array at the time of storage), If 8 is on, the data is read from the "final address value" toward "0" (that is, in the direction opposite to the time array at the time of storage).

The waveform information read from the waveform memory 1 in this manner is converted into an analog tone signal in the sound system 6 and then sounded as a tone. In this case, when the switch 8 is off, the percussion instrument sound, which is the original sound of the waveform information, is emitted from the sound system 6 because the waveform information about the sound emission starting portion of the percussion instrument sound is sequentially read from the waveform memory 1. However, when the switch 8 is on, the waveform memory 1 sequentially reads the waveform information relating to the end portion of the percussion instrument sound in the reverse direction. Therefore, the sound system 6 specially displays the percussion instrument sound in the inverted state on the time axis. It is pronounced as a sound effect.

FIG. 2 is a block diagram of a configuration in which an amplitude envelope control function is added to the embodiment of FIG.
The waveform information read from the waveform memory 1 is input to the multiplier 9, where it is multiplied by the envelope waveform signal ED generated from the envelope waveform generator 10 to control the amplitude envelope. Here, the envelope waveform generator 10 starts its operation by the start pulse SP output from the one-shot circuit 5 of FIG. The waveform information stored in the waveform memory 1 is shown in FIG.
In the case where the amplitude envelope is given in advance as shown in (a), if the switch 8 is off, the constant level envelope waveform signal ED is output as shown in (b) of FIG. Further, when the waveform information stored in the waveform memory 1 has its amplitude standardized to a constant level as shown in (c) of FIG. 3, if the switch 8 is off, (d) of FIG. ) Envelope waveform signal E having the same waveform shape as the amplitude envelope of percussion instrument sound as shown in FIG.
Output D.

On the other hand, when the switch 8 is turned on, a properly selected one of the different envelope waveform signals ED as shown in FIGS. 3 (e) to 3 (g) is output. By adding such an envelope control function, it is possible to variously change the special effect sound obtained when the switch 8 is turned on and the waveform information is read out from the waveform memory 1 in the opposite direction. Can be

FIG. 4 is a block diagram showing an embodiment of the present invention. For the waveform memory 1, the entire waveform of the rising portion (attack portion) of the musical tone (percussion instrument sound) and the partial waveform after that (1
Cycle or a plurality of cycles), and after reading all the waveforms at the rising edge once, then repeatedly reading out some waveforms to generate a musical tone,
The reverse read control function according to the present invention is added. In FIG. 4, the same circuits as those in FIGS. 1 and 2 are designated by the same reference numerals.

That is, in the waveform memory 1 in this embodiment, as shown in FIG. 5A, the waveform information about all the waveforms of the rising part of the percussion instrument sound and the subsequent partial waveforms (one cycle or a plurality of cycles). The waveform information is sequentially stored according to the arrangement on the time axis, and when the special effect sound selection switch 8 is off and the start pulse SP is given, the address counter 3 changes from “0” to “final address value”. An address signal AD that changes toward
By this address signal AD, the waveform information of the rising portion and the subsequent waveform information are read out. When the waveform information of this rising part and the subsequent waveform information are read once,
That is, when a detection signal is output from the final address detection circuit 4, the detection signal causes the address counter 3 to be preset with repeated address data indicating the beginning storage address of the waveform information to be repeatedly read from the repeated address data generation circuit 11. It Then, the address counter 3 outputs the address signal AD which changes toward the final address value with the initial storage address as the initial value. When the address signal AD reaches the final address value again, the detection signal is output from the detection circuit 4 and the repetitive address data indicating the leading storage address is reset in the address counter 3, and the same operation is repeated.

As a result, the waveform information constituting the musical tone waveform after the rising portion of the percussion instrument sound is repeatedly read out, the amplitude envelope is set in the multiplier 9, and then the sound is emitted from the sound system 6 as a musical tone. In this case, when the special effect sound selection switch 8 is set to ON, all the bits of the address signal AD output from the address counter 3 are inverted in the inverting circuit 2. Therefore, the waveform information stored in the waveform memory 1 is read out in the direction opposite to the time array at the time of storage and is sounded as a special effect sound. At this time, the repetitive address data is changed to a value slightly different from that when the switch 8 is off.

As a result, the read mode when the switch 8 is off and when it is on becomes as shown in FIGS. 5B and 5C, and the above-described embodiment is performed according to the ON / OFF state of the switch 8. Similar to the example, one waveform memory can be used to selectively generate a normal musical sound and a special effect sound that is a time-reversed version of the normal musical sound. That is, when the special effect sound selection switch 8 is off, the read mode is as shown in FIG. 5B, and the waveform information of the rising portion and the subsequent waveform information are in the same direction (forward direction) as the time array. After being read out once, the subsequent waveform information is repeatedly read out in the same direction (forward direction) as the time array. On the other hand, when the special effect sound selection switch 8 is turned on, the read mode is as shown in FIG. 5C, and the waveform information of the rising portion and the subsequent waveform information are read out in the reverse direction of the time array (that is, Then, the subsequent waveform information is read in the reverse direction, and the waveform information of the rising portion is read in the reverse direction), and then the waveform information of the rising portion is repeatedly read in the direction opposite to the time array. At this time, the starting point of repeated reading of the waveform information of the rising portion to be read in the reverse direction, that is, the repeating address is different from the starting point of repeated reading during forward reading, that is, the repeating address, as shown in the figure. Therefore, it is possible to perform repetitive read control in different modes during forward reading and backward reading.

When the switch 8 is in the ON state, the repeated address data output from the circuit 11 and indicating the leading address for repeated reading is randomly arranged in time as shown in Japanese Patent Laid-Open No. 59-49597. If you change to
It is possible to obtain a special effect sound that changes more complicatedly.

FIG. 6 is a block diagram showing another embodiment of the present invention, in which an external sound such as a natural musical instrument sound is input to the microphone 12.
It is configured to be picked up by the AD converter 13, converted into waveform information of digital data by the AD converter 13 and sequentially written in the waveform memory 1, and thereafter, the waveform information of the written external sound is read out in the direction opposite to the time sequence at the time of writing. It is a thing.

That is, the waveform memory 1 is in the read mode when the selector switch 14 is switched to the "1" side and in the write mode when the selector switch 14 is switched to the "0" side, and the switch 14 is switched to the "0" side. When the switch 7 is turned on in the switched write mode, the address signal AD from the address counter 3 is generated and supplied to the waveform memory 1 via the inverting circuit 2. At this time, the inverting circuit 2 supplies each bit of the address signal AD to the waveform memory 1 without inverting it. Therefore, the waveform information of the external sound picked up by the microphone 12 is sequentially written from "address 0" to "final address value".

After the writing is completed, when the switch 14 is switched to the "1" side to put the waveform memory 1 into the read mode and then the switch 7 is turned on, the address signal AD is generated from the address counter 3. When inverting circuit 2
Is the waveform memory 1 by inverting each bit of the address signal AD.
Supply to. Therefore, the waveform information of the external sound written in the waveform memory 1 is read out in the direction opposite to the time array at the time of writing, and after the amplitude envelope is set in the multiplier 9, the special effect sound is output from the sound system 6. Is pronounced as.

FIG. 7 is a block diagram showing an embodiment in which the same function as that of FIG.
The external sound picked up by
It is D-converted and temporarily stored in the buffer memory 15. The write control at this time is performed by turning on the write switch 14A and generating from the control circuit 16 an address signal that changes from "address 0" to "final address value". Next, the read switch 14B is turned on to generate an address signal that changes from the "final address value" to "address 0" from the control circuit 16, and the waveform information stored in the buffer memory 15 is arranged at the time of writing. The data is read in the opposite direction to the above, and is transferred and written to the waveform memory 1 inside the musical instrument body, for example. In this case, the write address signal for the waveform memory 1 is the address data generation circuit 1
7, the circuit 17 is synchronized with the control circuit 16 by the synchronization signal SYNC given from the control circuit 16, and the address signal for reading the waveform information from the buffer memory 15 and the waveform information in the waveform memory 1 are generated. Is configured to change in synchronization with the address signal for writing. By such an operation, the time information of the external sound waveform information once stored in the buffer memory 15 is converted in the reverse direction, and the waveform information is written in the waveform memory 1.

The waveform information written in the waveform memory 1 is
Start pulse S for the address data generation circuit 17
By giving P, the address signals generated from the circuit 17 sequentially read from "address 0" to "final address value". In this way, the time sequence of the waveform information is inversely converted, the waveform information is written in the waveform memory, and then the waveform information written in the waveform memory is read out as it is. You can get the sound.

In each of the embodiments shown in FIGS. 1, 2, 4, and 6, the address signal AD is inverted by the inverting circuit 2 to read the waveform memory 1 in the backward direction. Alternatively, an up / down counter may be used as the address counter 3 and the counter may be down-counted to read the waveform memory 1 in the reverse direction.

Further, in each of the embodiments shown in FIGS. 1, 2, 4, and 6, when the waveform stored in the waveform memory 1 is reversed (the waveform read in the reverse direction described above is read from the start address to the final address in the reading order). (When sequentially stored toward the address), the inverting circuit 2 operates the special sound effect selection switch 8
The address signal AD may be inverted when turned off and output as it is when turned on.

Further, the waveform stored in the waveform memory may be not only a continuous plurality of cycles but also a plurality of discrete cycles. For example, dividing from the start to the end of musical tone generation into a plurality of frames, storing only waveform information of a typical one cycle or two cycles for each frame, and repeatedly reading the waveform information while sequentially switching the waveform information. Alternatively, if necessary, the waveform information that smoothly changes may be formed by interpolating the previous waveform and the next new waveform during this waveform switching. Further, as disclosed in Japanese Patent Application Laid-Open No. 58-142396, only waveform information for a plurality of cycles may be stored in the waveform memory and this waveform information may be repeatedly read. By doing so, the capacity of the waveform memory can be further reduced.

In each embodiment, when a waveform as a special effect sound is read from the waveform memory, for example, by adding / subtracting time-varying modulation data to / from the lower bit side of the address signal AD, the address signal You may make it modulate AD. Furthermore, in each embodiment, when reading a waveform as a special effect sound from the waveform memory,
A part (of course, a plurality of cycles) of the stored waveform may be designated as appropriate, and only the designated portion may be read.

On the other hand, in each embodiment, the waveform memory does not store the tone waveform amplitude sample data as it is as the waveform information (that is, without the PCM method) but stores the difference data between the adjacent sample amplitude values. The difference data may be cumulatively added / subtracted at the time of reading to obtain the original amplitude sample data (that is, the difference PCM method). Further, waveform data encoded according to another appropriate waveform encoding method such as a delta modulation (DM) method or an adaptive delta modulation (ADM) method may be stored in the waveform memory.

Further, in the above-mentioned embodiment, only one kind of musical sound can be generated, but when it is desired to generate a plurality of musical sounds,
It suffices to provide the same number of circuits (excluding the sound system) shown in the embodiment as the number of kinds of musical tones, add the outputs of the respective waveform memories and supply them to a common sound system. Alternatively, waveform information relating to a plurality of types of musical tone waveforms is stored in the waveform memory in correspondence with the type of musical tone to be generated, and the address counter AD and envelope for each musical tone are generated by operating the address counter and the envelope waveform generator in a time division manner. Waveform data ED is generated in time division,
It suffices that the envelope setting multiplier outputs the tone waveforms to which the envelopes for each tone are added in a time-division manner. However, in this case, the time division data output from the multiplier is accumulated and then supplied to the sound system.

Further, since the musical tone generating device of the above-mentioned embodiment is suitable for generating a musical tone especially corresponding to a percussion instrument tone, it can be used for generating a rhythm tone in an automatic rhythm playing device, for example. In this case, the start pulse output from the one-shot circuit 5 in FIG. 1 may be used to automatically generate a musical tone according to the rhythm pattern.

As a matter of course, the present invention can also be used when generating musical tones other than percussion instrument sounds. For example, when a musical tone of a piano tone is generated, the musical tone waveform of the piano tone may be stored in the waveform memory 1 in advance, and the stored musical tone waveform may be read out in response to a key operation on the keyboard. In this case, even if circuits as shown in FIGS. 1, 2 and 4 are provided corresponding to each key of the keyboard, and a tone waveform corresponding to the pitch of each key is stored in each waveform memory. Alternatively, the circuits shown in FIGS. 1, 2, and 4 may be commonly used for each key, and the frequency of the clock pulse φ input to the address counter may be changed according to the pitch of the pressed key. . In that case, a key-on signal (a signal that becomes “1” when the key is pressed) output from the keyboard circuit is input to the one-shot circuit 5 of FIG. 1 instead of the output signal of the sound generation command switch 7. To do so.

In the present invention, the waveform information relating to the desired musical tone stored in the waveform memory is not limited to the musical instrument sound such as the percussion instrument sound and the piano sound described above, but may be a human voice sound, an animal cry, a whistle, It may be any sound that is present in the natural world, such as the sound of wind.

[0037]

As described above, according to the present invention, the waveform information stored in the waveform memory is read out according to the time sequence in the direction opposite to the time sequence in the tone waveform, and then the tone waveform of the rising portion is continuously read. Since the waveform information about at least a part of the musical tone waveforms is repeatedly read according to the time sequence in the direction opposite to the time sequence of the musical tone waveforms, at least a part of the musical tone waveforms at the rising portion of the desired musical tone is temporally read. It is possible to generate a very special musical tone by repeatedly playing it in the opposite direction repeatedly, and it is quite varied and has a unique character unique to the desired musical tone. Has an excellent effect of being able to generate. In addition, a sound having a unique character / characteristic to a desired musical sound is stretched and continuously generated, so that the generation of the musical sound having a novel characteristic can be realized and the generated musical sound can be realized. Since the repetitive part (continuous part) of (1) becomes rich in waveform change (timbre change), it is possible to produce an extremely complicated musical sound, and so on.

Further, according to the musical tone generating apparatus of the present invention,
As described above, the special backward repetitive read control unique to the present invention and the read control along the same direction (forward direction) in the time array can be arbitrarily selected and designated, so that various musical tone generation can be realized. can do. Further, by providing the generated musical tones with amplitude envelopes having different envelope waveforms according to the selected and designated reading direction, it is possible to realize the generation of musical tones with a greater variety.

[Brief description of drawings]

FIG. 1 is a block diagram showing a basic configuration example of reverse reading control of a waveform memory.

FIG. 2 is a block diagram showing an embodiment in which an amplitude envelope is added to the read output of the waveform memory shown in FIG.

3 is a diagram showing some examples of waveforms stored in the waveform memory in FIG. 2 and envelope waveform data generated from an envelope waveform generator.

FIG. 4 is a block diagram showing an embodiment of the present invention.

5 is an operation explanatory view illustrating a read mode of the waveform memory in FIG.

FIG. 6 is a block diagram showing another embodiment of the present invention.

FIG. 7 is a block diagram showing still another embodiment of the present invention.

[Explanation of symbols]

 1 Waveform Memory 2 Inversion Circuit 3 Address Counter 4 Final Address Detector 5 One Shot Circuit 6 Sound System 7 Sounding Command Switch 8 Special Effect Sound Selection Switch 9 Multiplier 10 Envelope Waveform Generator 11 Repeated Address Data Generation Circuit 12 Microphone 15 Buffer memory

Claims (4)

(57) [Claims] 1. A waveform memory stores waveform information relating to a plurality of cycles of a musical tone waveform of a rising portion corresponding to a desired musical tone and a part of a plurality of subsequent periods of a musical tone waveform, and reads the stored waveform information. In the musical tone generating method for generating the desired musical tone, the waveform information is read out from the waveform memory according to a time sequence in the direction opposite to the time sequence of the tone waveform, and then the rising portion is continuously read. The waveform information about at least a part of the musical tone waveforms is repeatedly read according to a time sequence in the direction opposite to the time sequence of the musical tone waveforms, whereby the desired musical tone is inverted in time. A method of generating musical tones. 2. A waveform memory stores waveform information relating to a plurality of cycles of a musical tone waveform of a rising portion corresponding to a desired musical tone and a part of a plurality of subsequent periods of a musical tone waveform, and reads the stored waveform information. In the musical tone generating device for generating the desired musical tone, the waveform information stored in the waveform memory is read in the same direction or in the opposite direction to the time array at the time of storage, and the reading in the same direction. Reads out the waveform information from the waveform memory in accordance with the time sequence of the musical tone waveform, and then continuously obtains waveform information about the musical tone waveform including at least a part of the musical tone waveforms of a part of the plurality of cycles thereafter. The waveform information is repeatedly read according to the time arrangement in the musical tone waveform. In the reverse reading, the waveform information is read from the waveform memory. After the data is read out according to the time sequence in the direction opposite to the time sequence in the tone waveform, the waveform information about at least some of the tone waveforms in the rising portion is then read in the opposite direction to the time sequence in the tone waveform. And a selection means for selectively designating a reading direction in the reading means, and the desired musical tone or a musical tone in a form in which the desired musical tone is inverted in time. A musical tone generator characterized in that the sound is selectively generated. 3. The musical tone generating apparatus according to claim 2, wherein in the reading means, the starting point of repeated reading differs depending on the reading direction selected and designated by the selecting means. 4. A waveform memory stores waveform information relating to a plurality of cycles of a musical tone waveform of a rising portion corresponding to a desired musical sound and a part of a plurality of subsequent periods of a musical tone waveform, and reads the stored waveform information. In the musical tone generating device for generating the desired musical tone, the waveform information stored in the waveform memory is read in the same direction or in the opposite direction to the time array at the time of storage, and the reading in the same direction. Reads out the waveform information from the waveform memory in accordance with the time sequence of the musical tone waveform, and then continuously obtains waveform information about the musical tone waveform including at least a part of the musical tone waveforms of a part of the plurality of cycles thereafter. The waveform information is repeatedly read according to the time arrangement in the musical tone waveform. In the reverse reading, the waveform information is read from the waveform memory. After the data is read out according to the time sequence in the direction opposite to the time sequence in the tone waveform, the waveform information about at least some of the tone waveforms in the rising portion is then read in the opposite direction to the time sequence in the tone waveform. Read-out means for repeatedly reading out according to the time sequence of, the selection means for selecting and specifying the read-out direction in the read-out means, and the predetermined amplitude of the tone waveform corresponding to the waveform information read out from the waveform memory. An envelope giving means for giving an amplitude envelope having a different envelope waveform according to the reading direction selected and designated by the selecting means, and the musical tone having the amplitude envelope given by the envelope giving means. A musical tone generator that generates waveforms as musical tones.