JP4015267B2 - Waveform generator - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a waveform generator used for an electronic musical instrument, for example, and more particularly to a generator for reading and reproducing waveform data stored in a memory.
Such a waveform generator can be used for an electronic musical instrument called a sampler, for example.
[0002]
[Related technologies]
In the previously filed Japanese Patent Application No. 9-300403, the present applicant stores a time series of waveform data representing a waveform in the storage means, compresses and expands the time axis of the waveform, and converts the pitch. A “waveform generator” was proposed as a playback device.
[0003]
The waveform generating apparatus described in this application is provided with a formant shift operator, and the formant shift amount is set in advance to shift and reproduce the formant of the recording waveform, and the modulation signal LFO is used to form a periodic modulation. Or can be granted.
[0004]
[Problems to be solved by the invention]
However, this waveform generator does not change the formant even when the reproduced pitch of the waveform changes. Thus, if the formant does not change at all even if the playback pitch changes, it is still unnatural as a musical sound.
[0005]
The present invention has been made in view of such a problem, and it is an object of the present invention to prevent a musical sound or the like due to a generated waveform from becoming unnatural even when a playback pitch changes.
[0006]
In order to solve the above-mentioned problems, a waveform generator according to the present invention comprises a storage means for storing waveform data representing a series of waveforms arranged in time series, and a storage means for storing the waveform data. Compression / decompression information input means for inputting compression / decompression information for instructing the degree of compression or expansion of the time axis of the waveform, pitch information input means for inputting pitch information for instructing the playback pitch of the waveform, and the waveform A formant shift amount input means for inputting a formant shift amount for shifting the formant, a formant control information generation means for generating formant control information in which the formant shift amount is changed according to the pitch information, and the compression / decompression information. And pitch information and formant control information, and based on the waveform data, a waveform corresponding to a reproduction position that changes at a change speed corresponding to the compression / decompression information, Change the pitch corresponding to the Kion pitch information, and to change the formant corresponding to the formant control information, and a reproduction means for generating a reproduction waveform. In this waveform generator, the formant of the playback waveform changes depending on the formant control information, but the formant control information is changed according to the playback pitch, so that However, the formant can be changed so that the tone of the reproduced waveform does not feel unnatural.
[0007]
The playback means includes time information generating means for generating time information that changes at a change speed unrelated to the playback pitch, and the time information change speed corresponding to the time information and the pitch information. A reading means for reading waveform data from the storage means at a desired reading speed unrelated to the above can be included. In this waveform generator, the waveform reproduction speed is determined by the change speed of the time information of the time information generation means, and the change speed of the time information is independent of the playback pitch. Therefore, the reproducing means reads out the waveform data at a desired reading speed that corresponds to the time information and pitch information and is not related to the change speed of the time information, and the waveform data is instructed by the pitch information. When playback is performed at the playback pitch, the waveform playback speed and playback pitch can be adjusted independently.
[0008]
The waveform generator further includes first control information input means for inputting first control information for changing a change speed of the time information, and the time information generation means is responsive to the input first control information. And a means for changing the rate of change of the time information. The first control information is generated in response to, for example, an operation of a time companding operator for setting a value in advance or a modulation lever operated in real time.
Further, the time information generating means can be constituted by a counter, and the step amount of the counter is changed by the first control information. Thereby, the change speed of the time information can be changed by the first control information input by the first control information input means, and thereby the waveform reproduction speed can be arbitrarily changed. Even in this case, the playback pitch does not change depending on the playback speed.
[0009]
Further, the readout means reads out a waveform section including at least one period of the waveform at a period corresponding to the pitch information at a desired readout speed unrelated to the change speed of the time information. It can be configured to convert data to playback pitch. Further, since the desired reading speed is configured to correspond to the formant control information that is generated by the formant control information generating means and changes in accordance with the reproduction pitch, it is not dependent on the reproduction speed. On the other hand, it is possible to reproduce at a reproduction pitch in which the formant changes corresponding to the reproduction pitch.
[0010]
Further, the reproducing means has two processing systems, and each processing system generates a waveform of at least one cycle of the waveform data of the storage means at a period corresponding to twice the playback pitch. By finally adding the outputs, the waveform data in the waveform section can be converted into a playback pitch.
[0011]
The formant control information generating means further includes formant change information input means for inputting change information for changing the formant of the waveform in the waveform train, and the reading means shifts the formant to the low frequency side. When the change information is to shift the formant to the high frequency side, the reading speed of the waveform data in the waveform section can be controlled to be high. By configuring in this way, by inputting change information by the formant change information input means, the formant of the waveform to be reproduced can be shifted to the high frequency side or shifted to the low frequency side.
[0012]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of the present invention will be described below with reference to the drawings.
FIG. 1 shows a waveform generator for an electronic musical instrument as an embodiment according to the present invention. In FIG. 1, reference numeral 7 denotes a waveform memory comprising a RAM for storing information relating to waveform data to be reproduced. Reference numeral 6 denotes a DSP (digital signal processor) that performs reproduction processing of waveform data in the waveform memory 7 by digital processing. The DSP 6 includes a working memory used for arithmetic processing of the DSP 6 and a memory for storing a program of the DSP 6.
[0013]
8 is an A / D converter that A / D converts the input analog waveform signal into a digital waveform signal and inputs it to the DSP 6, and 9 is a D / A converter that converts the digital waveform signal reproduced and output from the DSP 6 into an analog waveform signal. This is a D / A converter for output. The digital waveform signal input from the A / D converter 8 can be stored as waveform data from the DSP 6 into the waveform memory 7.
[0014]
Reference numeral 1 denotes a CPU (central processing unit) which performs overall control of the apparatus such as control of the DSP 6 and detection and processing of the state of the operator group 2 and the keyboard device 3. A large-capacity hard disk device 4 stores a large amount of waveform data and the like, and the waveform data is transferred to the waveform memory 7 as necessary. Reference numeral 5 denotes a storage unit including a working memory used for arithmetic processing of the CPU 1, a memory for storing the CPU 1 program, a memory for storing the DSP program, and the like.
[0015]
As shown in FIG. 1, the operator group 2 includes a mode selection switch (MSS) 21, a bank selection switch (BSS) 22, a formant shift operation element (FSV) 23, a time companding operation element (TCV) 24, and the like. . Hereinafter, the function of each of these operators will be described.
[0016]
The mode selection switch 21 is a switch for selecting one from a recording mode, an editing mode, and a play (playback) mode. Here, the recording mode (REC mode) is a mode for recording (sampling) an externally input musical tone signal, the editing mode (EDIT mode) is a mode for editing a waveform sampled in the recording mode, and the play mode (PLAY mode) is In this mode, the waveform data stored in the waveform memory 7 is reproduced according to the performance operation of the keyboard.
The bank selection switch 22 is a switch for selecting one of a plurality of waveform data stored in the waveform memory 7.
The formant shift operation element 23 is an operation element for setting a shift amount from the formant of the original waveform data stored in the waveform memory 7, and sets a formant shift amount FSV (also referred to as a formant change coefficient) to be described later.
The time companding operator 24 sets the amount of time compression / expansion (hereinafter referred to as companding) for determining the progress speed of the reproduction position (play position) on the time axis when reproducing the waveform data. It is an operator to do.
[0017]
The keyboard device 3 includes performance operators such as a MIDI keyboard 31 and a modulation lever (hereinafter referred to as a modulation lever) 32. In addition to generating note information (note number NN, velocity V, note on / off NT, etc.) corresponding to the operated key for performance, the keyboard 31 plays back the waveform data stored in the waveform memory 7. It is also used to instruct the playback pitch and the start / end of playback by turning the key on / off.
[0018]
The modulation lever 32 is also referred to as a modulation wheel, and is an operator that controls modulation of a musical tone that is normally generated. In this embodiment, the set time companding amount STCV set by the above-described time companding operator 23 is corrected in accordance with the operation amount of the modulation lever 32 (modulation lever amount MLV), and the pitch of the waveform to be reproduced is changed. The playback speed (play speed) is controlled without any problems. The modulation lever 32 can generate a modulation lever amount MLV in the range of $ 00 to $ 7F. In the following description, “$” added to the head of $ 7F or the like means hexadecimal display. The modulation lever 32 is normally located at the center position of the swinging range, and the modulation lever amount MLV of $ 40 is output at this center position.
[0019]
The waveform memory 7 is a memory for storing data relating to a waveform signal to be reproduced. The waveform memory 7 is roughly divided into a parameter storage unit and a waveform data storage unit. The parameter storage unit further includes a parameter 1 storage unit (status area). ) And parameter 2 storage section (cutout start address area). This parameter storage unit
Parameter 1 storage (status area): $ 20 address / bank
Parameter 2 storage section (cutout start address area): An address area per bank is provided in units of $ 800 addresses / 1 bank to store data. The waveform data storage unit stores all waveform data having a waveform number corresponding to the bank number for each bank in units of $ 8000 addresses / 1 bank. In other words, the parameter 1 storage unit is divided into segments at every $ 20 address, the parameter 2 storage unit is divided into every $ 800 address, and the waveform data storage unit is divided into every $ 8000 address. Bank numbers are assigned in order from the younger, such as bank 0 area, bank 1 area, bank 2 area. For example, if a parameter for a waveform with a certain waveform number is stored in the bank 0 area of the parameter 1 or 2 storage section, the waveform data of that waveform number corresponds to the bank 0 area of the same bank number in the waveform data storage section. To store.
[0020]
Hereinafter, the data configurations of the parameter storage unit and the waveform data storage unit will be described with reference to FIGS. 2 shows the data configuration of the parameter 1 storage unit (status area), FIG. 3 shows the data configuration of the parameter 2 storage unit (cutout start address region), and FIG. 4 shows the data configuration of the waveform data storage unit. .
[0021]
As shown in FIG. 2, the parameter 1 storage unit (status area) stores, for each bank number (that is, each waveform number), a waveform name wn, a sampling frequency sr, an original pitch op, an original pitch shift amount opsv from the original key, The output level ol is stored. The contents of these parameters are as follows.
[0022]
Waveform name wn: This waveform name wn is displayed on a display device and used for easy identification of memory waveforms.
Sampling frequency sr: The sampling frequency of the original waveform (the waveform stored in the memory, the same applies hereinafter) so that the original pitch does not change when playback processing is performed at a sampling frequency different from the sampling frequency of the original waveform Used for correction.
Original pitch op: Original pitch of original waveform data.
Original pitch shift amount opsv: A value for shifting the original pitch op (a magnification value for changing the original pitch OP), and used when changing the original pitch op, for example, when playing with different from the original pitch op. Is done.
Output level ol: for setting the output level of the original waveform.
[0023]
Next, the parameter 2 storage unit (cutout start address area) stores the cutout start address of the waveform data for each bank number as shown in FIG. The cutting start address csa is the head address of a waveform for one pitch (hereinafter referred to as a cutting waveform). Hereinafter, in this specification, the extracted waveforms arranged in time series are referred to as waveform trains. This cut-out start address csa is stored in time series along the time axis over the entire waveform string of the waveform number. The cutting start address csa of the cutting waveform of the last one pitch of the waveform row is stored as the waveform reading end address wea at the end of the area for storing the cutting start address. Further, the waveform readout end address wea of the waveform is stored as a header at the head address of the storage area of the bank number.
[0024]
Note that the cut-out start address csa when the pitch is not determined by a consonant part or the like is an appropriate pitch,
“Cutting start address csa = previous cutting start address csa + pitch”
Are obtained for the entire consonant section and stored in the cut-out address storage area. The pitch here is the address width of the cut waveform for one pitch. An appropriate pitch is, for example, a pitch that is first detected (determined) after shifting from a consonant to a vowel, or a pitch that is determined next.
[0025]
Next, as shown in FIG. 4, the data structure of the waveform data storage unit is such that the waveform data of the waveform sequence with bank number (waveform number) 0 is stored in the bank 0 area of the waveform data storage unit and the waveform sequence with bank number 1 The waveform data (sampling value) wd of the waveform train is stored in each bank area in sequential address order, such as in the bank 1 area of the waveform data storage unit.
[0026]
Various register groups are set in the memory 5 on the CPU 1 side. 5 and 6 show various parameters set in the register group. FIG. 6 shows the MIDI information register group. Hereinafter, these parameters will be described.
[0027]
Bank number BN: Bank number (waveform number) of the waveform data set by the bank selection switch 22 According to the bank number BN, a waveform train (waveform number) to be reproduced is selected.
[0028]
Various parameters stored in the parameter 1 storage unit of the waveform memory 7 That is, the waveform name WN, sampling frequency SR, original pitch OP, original pitch shift amount OPSVPP, and output level OL. Of these, parameters other than the original pitch are parameters that are not directly related to the embodiment of the present invention, and therefore processing other than the original pitch is excluded from the present embodiment.
[0029]
Formant shift amount FSV: Formant shift amount set by the formant shift operator 23. When the formant shift amount FSV is “1”, the waveform row in the waveform memory 7 is reproduced with the same formant as the original waveform, and when the value is larger than “1”, the formant is higher than the original waveform. If the value is smaller than “1”, the formant is shifted to the lower frequency side than the original waveform and reproduced.
[0030]
Set time companding amount STCV: Determines the amount of time compression / expansion (that is, the play speed) when the waveform train stored in the waveform memory 7 is reproduced, and is a numerical value set by the time companding operator 24. . If modulation by the modulation lever 32 to be described later is not applied, if this set time companding amount STCV is “1”, the time changes at the same speed as the time change of the original waveform, and a value larger than “1”. If there is, the time change is faster than the original waveform and the play time is shortened. If the value is smaller than “1”, the time change is slower than the original waveform and the play time is increased.
[0031]
Time companding modulation amount (coefficient) TCMV: a time compression / expansion value set by the modulation lever 32 and a coefficient for correcting the set time companding amount STCV. It is set in the range of “0 ≦ time pressure modulation amount TCMV ≦ 2”.
Pitch ratio PR: A ratio between the pitch corresponding to the pitch (key number) designated by the keyboard 31 and the original pitch OP of the memory waveform, and is stored for each voice module.
Velocity level VL: Velocity V designated by the keyboard 31 and stored for each voice module.
[0032]
The MIDI information register group in FIG. 6 is a register group for storing note information and the like input from the keyboard device 3, and stores the following parameters.
Note number NN: Key number operated on the keyboard 31.
Velocity V: Velocity of keys operated on the keyboard 31.
Note ON / OFF NT: ON / OFF of a key operated by the keyboard 31.
Modulation lever amount MLV: An amount operated by the modulation lever 32.
[0033]
Various register groups shown in FIGS. 7 and 8 are set in the internal memory on the DSP 6 side. Hereinafter, parameters set in these register groups will be described.
[0034]
First, the following parameters are set in the key information register group. In a play process of the CPU 1 described later, an operation of the keyboard 31 is detected, and key information is transferred to the DSP 6 by the assignment process and temporarily stored in this key information register. These parameters are set at regular intervals by the interrupt routine.
-Pitch ratio pr.
Formant shift amount fsv: The actual formant shift amount of the reproduced waveform.
-Velocity level vl.
-Time companding amount tcv: Actual companding amount of the waveform to be reproduced.
Cut-out address pointer cap: pointer indicating the cut-out start address csa in the cut-out start address area of the parameter 2 storage unit (read address of the parameter 2 storage unit).
Bank number bn: Bank number of the waveform memory.
[0035]
Next, various parameters used in the main routine of the DSP 6 include the following.
Waveform read end address wea: Waveform read end address stored in the bank of the parameter 2 storage unit (cutting start address of the last cut out waveform of the waveform train).
Play position pp: Pointer for indicating the current cut position for cutting out the cut waveform sequentially from the waveform row of the waveform memory.
Next cut start address ncsa: Cut start address of the next cut waveform.
First / second extraction start address fcsa / scss: Current extraction start address in processing system 1/2, which will be described later.
Cutout waveform pitch cwp: Pitch value of the cutout waveform currently being processed in the processing system 1/2 (displayed as an address width).
Play pitch (width) ppw: A pitch for reproducing the waveform train stored in the waveform memory.
Reproduction pitch counter ppc: This indicates the current position within one period of the reproduction pitch ppw.
Gate value g: For muting the output waveform.
First / second address counter ac1 / ac2: A counter for calculating the read address of the waveform memory in the processing system 1/2.
First / second envelope window ew1 / ew2: A window for adding a waveform row in the waveform memory to extract a reproduced waveform.
First / second window counter wc1 / mw2: for generating an envelope window.
Identification flag f: a flag for identifying partial 1/2.
Window length wl: The window length of the envelope window.
First / second waveform read address wral / wr2: Read address of waveform data in the processing system 1/2.
Waveform output wo: A composite value of the waveform output extracted by the processing system 1/2.
Output envelope level oel: The current value of the envelope level multiplied by the waveform output wo.
Play waveform output pwo: The output level of the playback waveform that is actually played back.
[0036]
Hereinafter, the operation of the apparatus of this embodiment will be described with reference to a flowchart.
FIG. 9 shows a flowchart of a main routine as processing performed by the CPU 1. When the main routine starts, it is monitored whether the mode selection switch 21 is operated in the recording mode, the editing mode, or the play mode (step A), and when the operation is performed, the operation is performed in the recording mode, the editing mode, and the play mode. It is determined which mode is selected (step B). If it is a recording mode, a recording (REC) mode process is performed (step C), if it is an edit mode, an edit (EDIT) mode process is performed (step D), and if it is a play mode, a play mode process is performed. (Step E).
[0037]
FIG. 10 shows a flowchart of a recording mode processing routine in the recording mode. The recording mode process is a process for recording (sampling) a musical tone signal input from the outside. After setting the recording mode with the mode selection switch 21, the sampling start operator is operated (step C3) to start sampling. Recording (sampling processing) is then performed (step C4). Data of the musical tone signal to be sampled is stored in the waveform memory 7. To exit from this recording mode processing routine and return to the main routine, the EXIT operator is operated (step C2).
[0038]
FIG. 11 shows a flowchart of an edit mode processing routine in the edit mode. The edit mode process changes the waveform sampled in the recording mode, edits the waveform data to be reproducible, transfers the waveform data to the hard disk device 3, and transfers the waveform data from the hard disk device 3 to the waveform memory 7. (Step D3). To exit from this edit mode processing routine, the EXIT operator is operated (step D2).
[0039]
FIG. 12 shows a flowchart of a play mode processing routine in the play mode. In the initial setting of the play mode processing routine (step E1), the register for scanning the state of the operator group 2 is reset to make the operator group operable, and the initial state of each operator is set. Keep it. The initial state is a play mode process, and a process corresponding to the operation of the operation element is performed only when there is a change in each operation element. Therefore, the initial reference state is set.
[0040]
The play mode processing routine is started by operating the play start operator after setting the play mode with the mode selection switch 21. The play process (step E3) is a process for reproducing the waveform data in the waveform memory 7 in accordance with the performance information from the keyboard 31. To exit from this play mode processing routine, the EXIT operator is operated (step E2).
[0041]
13 and 14 show the detailed processing procedure of the play process (step E3) in this play mode. This play process is executed by the CPU 1. In this play process, the setting by the operator group 2 and the performance operation by the keyboard 31 are detected, and the contents of the registers storing the corresponding parameters are updated. Among the contents set in these registers, necessary ones are transferred to corresponding registers on the DSP side. This play process is executed with a longer cycle than the main routine on the DSP side. That is, the main routine on the DSP side is always executed once until this play process is executed once and then executed next.
[0042]
When the play process is started, whether or not there is a change in the operation of the bank selection switch 22 (step F1), and if there is a change, the bank number BN set by the bank selection switch 22 is updated. . Corresponding to the updated bank number BN, the waveform name wn, the sampling frequency sr, the original pitch op, the original pitch shift amount opsv, and the output level ol are read from the parameter 1 storage unit of the waveform memory 7 respectively. The parameter name WN, sampling frequency SR, original pitch OP, original pitch shift amount OPSV, and output level OL are updated (step F3).
[0043]
Next, it is checked whether or not the formant shift operator 23 has changed (step F4). If there is a change, the formant shift amount FSV is updated accordingly (step F5). Similarly, whether or not there is a change in the time companding operator 24 (step F6), and if there is a change, the set time companding amount STCV is updated accordingly (step F7).
[0044]
Next, it is determined whether or not note information has been input from the keyboard 31 (step F8). If note information has been input, assignment processing to the voice module is performed based on the note information (step F8). F9).
The detailed procedure of the voice module assignment process is shown in the voice module assignment process routine shown in FIG. This assignment process to the voice module is a process of assigning note information (note number NN, velocity V, note on / off NT) input from the keyboard 31 to the voice module. Is preferentially assigned to the voice module, or note-off information for turning off already assigned notes is assigned to the voice module (step G1). Since the details of the allocation processing in step G1 are not directly related to the present invention, detailed processing contents are omitted.
[0045]
In the voice module assignment processing routine, after the note information is assigned to the voice module (step G1), it is determined whether the note on / off NT is note on information or note off information (step G2). If it is note-on information, the note number NN in the note information is converted into a note pitch NP and a key follow coefficient KF described later (step G3). Based on this note pitch NP, in the next step G4,
"Pitch ratio PR = note pitch NP / original pitch OP"
That is, the ratio between the note pitch NP and the original pitch OP is obtained and set as the pitch ratio PR,
"Velocity level VL = Velocity V"
That is, the velocity V in the note information is set as the velocity level VL.
If the original pitch shift amount OPSV is also reflected, this step G4 is performed.
“Pitch ratio PR = note pitch NP / (original pitch OP × original pitch shift amount OPSV)
"Velocity level VL = Velocity V"
Like this.
[0046]
Further, as described above, in step G3, the note number NN is also converted into the key follow coefficient KF. In this conversion, the relationship between the note number NN and the key follow coefficient KF as shown in FIG. 16 is stored as a table, and conversion is performed according to the table. Note that a plurality of tables may be prepared and appropriately selected and used. Alternatively, the key follow coefficient KF may be obtained by calculation according to a function.
[0047]
Further, the bank number BN set by the bank selection switch 22 on the CPU side is sent to the DSP side and set as the bank number bn, and the cut-out address pointer cap on the DSP 6 side is set to 0 (step G5).
[0048]
On the other hand, when note-on / off NT is note-off information (step G2),
Velocity level VL = 0
That is, the velocity level VL is set to 0 (step G6).
[0049]
When this voice module assignment process routine (FIG. 15) is completed, the process returns to the original play process routine to determine whether or not the modulation lever 32 has been operated (step F10). The time companding modulation amount TCMV is calculated by the following formula:
Time companding modulation amount TCMV = modulation lever amount MLV / $ 40
(Step F11).
[0050]
This time companding modulation amount TCMV is for correcting the set time companding amount STCV set by the time companding operator 24 to obtain a time companding amount tcv for waveform reproduction in the DSP 6. This is a correction coefficient by which the set time companding amount STCV is multiplied. The above calculation of “modulation lever amount MLV / $ 40” normalizes the modulation lever amount MLV (range of $ 00 to $ 7F) of the modulation lever 32 to a value in the range of 0 to 2. Since $ 40 is output at the center position, the time companding modulation amount TCMV is 1 when the center position is present, and the set time companding amount STCV set by the time companding operator 24 remains as it is. As the output of the modulation lever 32 changes to the $ 00 side, the time companding modulation amount TCMV decreases and the time companding amount tcv decreases (as described later, the play speed (waveform reproduction speed) increases). When the output is $ 00, the time companding amount tcv becomes zero (the play speed stops as will be described later). Further, as the set time companding amount STCV by the modulation lever 32 changes to the $ 7F side, the time companding amount tcv approaches twice the set time companding amount STCV (the play speed increases as will be described later).
[0051]
In the CPU 1, the next interrupt processing is performed at a constant cycle, and the processing result is sent to the DSP 6, and the pitch ratio pr, formant shift amount fsv, velocity level vl, and time companding amount tcv are set in the respective registers. The The interrupt processing routine will be described below with reference to FIG.
[0052]
When the interrupt process is started, the pitch ratio pr, the formant shift amount fsv, the velocity level vl, and the time companding amount tcv are set in the DSP 6 register. That is, first,
“Pitch ratio pr = Pitch ratio PR”
The pitch ratio PR is set to the pitch ratio pr as it is.
[0053]
Next, the formant shift amount fsv is
“Formant shift amount fsv = formant shift amount FSV × key follow coefficient KF”
Ask for. When the formant shift amount fsv is obtained by this calculation formula, the formant shift amount FSV set by the formant shift operator 23 changes according to the key follow coefficient KF, and the formant shift amount changes corresponding to the playback pitch. It will be.
[0054]
next,
“Velocity level vl = velocity level VL”
The velocity level VL is set to the velocity level vl as it is.
Finally, the time companding amount tcv, which is a variable for determining the actual play speed of the waveform to be reproduced, is obtained by the following calculation formula.
“Time companding amount tcv = Set time companding amount STCV × Time companding modulation amount TCMV”
In other words, the time companding amount tcv is determined by multiplying the set time companding amount STCV set by the time companding operator 24 and the time companding modulation amount TCMV which is a correction coefficient inputted by the modulation lever 32. The
[0055]
Next, the processing of the DSP 6 that has received the key information transfer will be described with reference to FIG. FIG. 18 is a flowchart showing the main routine of the DSP 6, which is repeatedly executed at a sampling cycle. Based on the transfer of new key information from the CPU 1, it is monitored whether or not the extraction address pointer cap is set to 0 in the “assignment process to voice module” of the CPU (step J1), and the extraction address pointer cap = 0. When set to, voice sound generation start processing is performed (step J2).
[0056]
FIG. 19 shows the voice generation start process. Hereinafter, these processes will be described sequentially. In the following description, the symbol @n (bank number bn × $...) Reads data from the address of “bank number bn × $...” In the parameter n storage unit (n = 1, 2). Means that.
-Waveform read end address wea = @ 2 (bank number bn x $ 800)
Data is read from the (bank number bn × $ 800) address (= address where the waveform read end address wea is stored) in the parameter 2 storage unit (cutout start address area), and is set in the register as the waveform read end address wea.
Play position pp = @ 2 (bank number bn × $ 800 + cut address pointer cap + 1)
The data (the cut start address csa1 of the first cut waveform in the waveform train) is read from the (bank number bn × $ 800 cut address pointer cap + 1) address in the parameter 2 storage unit and set in the register as the play position pp.
Next cut start address ncsa = @ 2 (bank number bn × $ 800 + cut address pointer cap + 2)
Data (cutout start address csa2 of the second cutout waveform in the waveform train) is read from the address (bank number bn × $ 800 + cutout address pointer cap + 2) in the parameter 2 storage unit and set in the register as the next cutout start address ncsa.
Cutout waveform pitch cwp = Next cutout start address ncsa-Play position pp Cutout waveform pitch cwp (address) by subtracting the set play position pp from the next cutout start address ncsa (= first cutout start address during this sounding start process) Set to the register as the width). This cut-out waveform pitch cwp corresponds to the pitch of the cut-out waveform to be processed (the first cut-out waveform at the time of sound generation start processing).
Reproduction pitch ppw = cutout waveform pitch cwp × pitch ratio pr
The cutout waveform pitch cwp and the pitch ratio pr are multiplied and set as a reproduction pitch ppw in the register.
・ Current cut start address ccsa = play position pp
The play position pp is set in the register as the current cut start address ccsa.
First cut start address fcsa = current cut start address pcsa
The current cut start address ccsa is set in the register as the first cut start address fcsa in the first processing system.
Second cut start address scsa = current cut start address ccsa
The current cut start address ccsa is set in the register as the second cut start address scsa in the second processing system.
First address counter ac1 = 0
As the first address counter ac1, 0 is set in the register.
Second address counter ac2 = reproduction pitch ppw / 2
The reproduction pitch ppw / 2 (a half cycle value of the reproduction pitch ppw) is set in the register as the second address counter ac2.
First window counter wc1 = 0
As the first window counter wc1, 0 is set in the register.
Second window counter wc2 = 1
1 is set in the register as the second window counter wc2.
・ Gate value g = 1
1 is set in the register as the gate value g.
Cut-out address pointer cap = 1
1 is set in the register as the cut-out address pointer cap.
[0057]
When the voice generation start process (step J2) is completed, a read process is performed (step J3). This read processing will be described in detail later. Thereafter, the play waveform output pwo read in the reading process (step J3) is output (step J4).
[0058]
In the above-described “reading process” in step J3, a cut-out waveform is cut out sequentially from the waveform sequence (speech), and the pitch corresponding to the desired reproduction pitch (= reproduction pitch) while maintaining the formant characteristics of the cut-out waveform. ) Is used to convert the pitch while maintaining the formant characteristics of the original waveform sequence. In this reading process, the pitch of the waveform to be reproduced (= reproduction pitch) is changed according to the pitch of the key pressed on the keyboard, but the play time required for waveform reproduction depends on the size of the reproduction pitch (that is, which key is It is not affected by whether it was pressed).
[0059]
Briefly explaining this read processing operation, a cut-out waveform of about 1 to 2 pitches in the vicinity of the position specified by the play position pp is sequentially cut out from the waveform sequence stored in the waveform memory 7 as time passes. The cut out waveform is reproduced with a pitch and formant different from the original waveform. At this time, the cut-out waveform is reproduced in parallel by two processing systems (partial), and each processing system has a period twice as long as the reproduction pitch to be reproduced and a half period (= reproduction) The extracted waveform is reproduced so as to be shifted), and these are combined to obtain the period of the reproduction pitch to be reproduced.
[0060]
20 and 21 are diagrams for explaining the pitch conversion process. If the formant shift amount fsv is 1, there is no change. If it is not 1, the formant is slightly changed. Here, a case where the formant shift amount fsv is 1, that is, without changing the formant and the reproduction pitch ppw is set higher than the original waveform data of the waveform memory 7 will be described with reference to FIG. 20, and the formant shift amount fsv is larger than 1. That is, the case where the formant is changed and the reproduction pitch ppw is made the same as the original waveform row of the waveform memory 7 will be described with reference to FIG.
[0061]
First, referring to FIG. 20, by designating a pitch by pressing a key on the keyboard 31, the pitch is shifted to a higher frequency side than the original waveform sequence, and the formant characteristic is not changed (formant shift amount fsv). = 1) A case will be described.
[0062]
(A) of FIG. 20 shows a waveform sequence in the bank 0 area, and shows cut-out waveforms for five periods starting from the cut-out start address csa1. Each cut waveform of this waveform row has cut waveform pitches cwp1, cwp2, cwp3.
[0063]
Further, the reproduction pitch ppw is based on the pitch ratio pr (= note pitch np / original pitch op) determined according to the note number designated by the pitch of the keyboard 31.
“Reproduction pitch ppw = cutout waveform pitch cwp × pitch ratio pr”
Set by. The period of the reproduction pitch ppw is made by counting with a reproduction pitch counter ppc. Further, in synchronization with the reproduction pitch ppw, a first address counter ac1 and a second address counter ac2 for obtaining a read address of waveform data for waveform reproduction in the two processing systems are created. The first address counter ac1 and the second address counter ac2 are incremented by one every sampling period.
[0064]
The first processing system calculates the first waveform readout address wral performed in step N6 of the “waveform readout processing subroutine” described later.
“First waveform read address wral = first cut start address fcsa + first address counter ac1 × formant shift amount fsv”
The extracted waveform is read at a reading speed determined by the portion of “first address counter ac1 × formant shift amount fsv” in the latter half.
[0065]
Further, the second processing system calculates the waveform read address performed in step N15 of the “waveform read processing subroutine” described later.
“Second waveform readout address wra2 = second extraction start address scsa + second address counter ac2 × formant shift amount fsv”
The extracted waveform is read out at a reading speed of “second address counter ac2 × formant shift amount fsv” in the latter half.
[0066]
Thus, the readout speed of the cut-out waveform is faster or slower than the reference readout speed (= stepping speed of the first / second address counter ac1 / ac2) when the formant shift amount fsv is larger or smaller than 1. . When the waveform reading speed is a reference value, the cut out waveform is read in the order of the addresses according to the stepping speed of the first / second address counter ac1 / ac2, and the original formant of the cut out waveform is stored in the read waveform as it is. Is done. On the other hand, when the waveform reading speed becomes higher than the reference value, the extracted waveform reading speed is read faster than the reference value, and the original formant of the extracted waveform is shifted to the high frequency side in the read waveform. When the waveform reading speed is lower than the reference value, the extracted waveform reading speed is read slower than the reference value, and the original formant of the extracted waveform is shifted to the low frequency side in the read waveform.
In the example of FIG. 20, since the formant shift amount fsv = 1, it is equal to the change of the first address counter ac1 and the second address counter ac2, and as a result, the formant characteristics are not changed.
[0067]
Further, in synchronization with the first address counter ac1 and the second address counter ac2, envelope windows ew1 and ew2 as windows for cutting waveform data for formant processing are created for the first and second processing systems, respectively. The first processing system has the waveform of the envelope window ew1 shown in FIG. 20 (f), and the second processing system has the waveform of the envelope window ew2 shown in FIG. 20 (g). The envelope windows ew1 and ew2 are peak values in the range of 0 to 1, with the window length wl being a half period, and gradually increasing from 0 to 1 in the first half period and gradually decreasing from 1 in the second half period. A triangle that becomes zero. The window length wl which is a half cycle of the envelope windows ew1 and ew2 is determined in step L9 of the “readout subroutine” described later.
“Window length wl = cutout waveform pitch cwp / formant shift amount fsv”
Ask for. However, the window length wl is limited to the value of the reproduction pitch ppw at the maximum so as not to exceed the reproduction pitch ppw as will be described later. FIG. 20 shows an example of such a restriction.
[0068]
In the first processing system, the cutout waveform obtained by extracting about 1 to 2 pitches from the first cutout start address fcsa set in step L14 of the “read processing subroutine” described later is multiplied by the envelope window ew1, and FIG. The waveform shown in (h) is obtained. Similarly, in the second processing system, an envelope window ew2 is multiplied by the extracted waveform obtained by extracting about 1 to 2 pitches from the second extraction start address scsa set in step L15 of the “read processing subroutine” described later. The waveform shown in FIG. 20 (i) is obtained.
[0069]
According to such a processing method, these waveforms retain the formant characteristics of the original cut waveform. The waveforms shown in FIGS. 20 (h) and (i) are twice as long as the period length of the reproduction pitch ppw, but when both waveforms are added, the period length of the reproduction pitch ppw is obtained. Accordingly, the formant characteristic can be maintained as it is while the original sampling data is pitch shifted to the high frequency side by the key designation pitch from the keyboard 31.
[0070]
Further, as will be described later, the play speed of waveform reproduction is adjusted by changing the speed at which the extracted waveforms are sequentially extracted from the waveform train in units of extracted waveforms. This is to extract a cut-out waveform near the address indicated by the play position pp, and the play speed can be increased or decreased by changing the stepping speed at the play position pp.
[0071]
FIG. 21 shows a case where the formant shift amount fsv is made larger than 1 and the reproduced waveform formant is shifted to a higher frequency side than the original waveform formant. Here, for easy understanding, the key designation pitch is shown as being substantially equal to the cut-out waveform pitch cwp.
[0072]
Since the first processing system is “first address counter ac1 × formant shift amount fsv” and the second processing system is “second address counter ac2 × formant shift amount fsv”, the cut-out waveform readout speed is first. As a result, the formant characteristic is shifted to the high frequency side, and the change is given faster than the change of the address counter ac1 and the second address counter ac2.
[0073]
The cut waveform is shortened by increasing the reading speed. Therefore, the window length wl of the envelope windows ew1 and ew3 is also
“Window length wl = cutout waveform pitch cwp / formant shift amount fsv”
As the cutout waveform is shortened, it is shortened.
[0074]
In the examples of FIGS. 20 and 21, the formant shift amount fsv has been described as a constant value for the sake of simplicity. However, in this embodiment, the formant shift amount fsv is determined by the key follow coefficient KF in the interrupt processing routine of FIG. Since they are multiplied, the formant shift amount fsv changes in accordance with the pitch to be reproduced. Therefore, in this embodiment, control is performed so that the formant of the reproduced waveform signal changes in accordance with the pitch to be reproduced.
[0075]
The operation of the reading process outlined above will be described with reference to the flowcharts of FIGS.
First, the overall operation will be described with reference to the flowcharts of FIGS. This flowchart corresponds to the operations of FIGS. Each of the above registers is initialized with parameter values of various registers when the power is turned on. That is,
“Pitch ratio pr = formant shift amount fsv = time companding amount tcv = 1.0”
“Output level ol = Cut address pointer cap = Bank number bn = 0”
“Output envelope level oel = 0”
“Waveform read end address wea = play position pp = next cut start address ncsa = first / second cut start address fcsa / scsa = 0”
“Cutout waveform pitch cwp = reproduction pitch ppw = 0”
“Reproduction pitch counter ppc = gate value g = first address counter ac1 = 0”
“Second address counter ac2 = reproduction pitch ppw / 2”
“First window counter wc1 = 0.0”
“Second window counter wc2 = 1.0”
“Identification flag f = 1”
[0076]
In the following description, it is assumed that a little time has passed after the power is turned on, and appropriate values have already been stored in each register and each counter by the processing of the flowchart. Further, this flowchart is executed at every sampling period in the DSP 6.
[0077]
At the start of sound generation, the play position pp is set to the cut start address csa1 of the first cut waveform of the waveform row indicated by the bank number bn to be processed in the parameter 2 storage unit of the waveform memory 7 by the sound start processing of the voice described above. Is set. The play position pp manages the time progression of waveform reproduction with this as a reference point. Every one sample period, the value of the play position pp is incremented by the time companding amount tcv (step L1). That is,
Play position pp = play position pp + time expansion amount tcv
And As a result of such an update, if the time companding amount tcv is large, the play position pp advances fast and the time required to reproduce the entire waveform train is shortened. Conversely, if the time companding amount tcv is small, the play position pp Since the process proceeds slowly, the time required to reproduce the entire waveform train becomes longer.
[0078]
Next, the play position pp and the waveform read end address wea are compared (step L2). If the play position pp is equal to or higher than the waveform read end address wea, the processing for the waveform string of the bank number is completed. Therefore, the play position pp is fixed to the value of the waveform read end address wea (step L3), so that the waveform processing does not proceed further.
[0079]
Next, the play position pp is compared with the next cut-out start address ncsa (step L4). The next cut start address ncsa is the cut start address (start address) of the next cut waveform, as can be seen from the sound generation start process of FIG. Therefore, if the play position pp is larger than the next cut-out start address ncsa, the cut-out waveform that has been processed so far is completed. Therefore, the cut-out waveform is updated in order to shift the processing from this cut-out waveform to the next cut-out waveform.
[0080]
The cutting waveform is updated as follows (step L5).
Current cut start address ccsa = @ 2 (bank number bn × $ 800 + 1 + cut address pointer cap)
Next cut start address ncsa = @ 2 (bank number bn × $ 800 + 2 + cut address pointer cap)
Cutout waveform pitch cwp = Next cutout start address ncsa−Current cutout start address ccsa
・ Increment cut address pointer cap by one
[0081]
That is, the current cut start address ccsa is read from the address pointed to by “cut out address pointer cap + 1” in the cut start address area of the parameter 2 storage section of the waveform memory 7, and the next point from the address pointed by “cutout address pointer cap + 2”. The cut start address ncsa is read out, and the current cut start address ccsa minus the next cut start address ncsa (address difference) is set as a cut waveform pitch cwp. Thereafter, the cut address pointer cap is incremented by one.
[0082]
In step L4, if the play position pp is smaller than the cut start address ncsa of the next cutout waveform, the cutout waveform is not updated because it is still in the middle of the current cutout waveform. Jump over.
[0083]
As a result of the above steps L1 to L5, when a large amount of time expansion tcv is set, the play position pp advances rapidly, and as a result, the first / second cutout start addresses fcsa / scsa are updated (steps described later). L14, L15) are performed early, so that the play time is shortened. On the other hand, when the time companding amount tcv is small, the play position pp advances slowly and the update of the first / second cutout start addresses fcsa / scsa is performed late, so that the play time becomes long.
[0084]
When the time companding amount tcv is set to a considerably small value, the same clipped waveform is repeatedly reproduced a plurality of times and the waveform reproduction proceeds at a slow speed. This is because the progress of the play position pp is slow, so that the play position pp is not easily larger than the value compared in the judgment of step L4. This is because the reading process is repeated. On the other hand, if the time companding amount tcv is set to a considerably large value, in the update of the cutout waveform, there is a case where the next cutout waveform is skipped and the cutout waveform is not reproduced.
[0085]
Here, when the modulation lever 32 is operated, the time companding amount tcv is changed by changing the time companding modulation amount TCMV in accordance with the operation amount MLV of the modulating lever 32 in Step F11 of the play process described above. Since the play time is set, the play time changes as follows according to the operation amount MLV of the modulation lever 32.
[0086]
(1) When the operation amount MLV of the modulation lever 32 is $ 00 to $ 40
As the operation amount MLV of the modulation lever 32 approaches $ 00, the speed of movement of the play position pp becomes slower. When the operation amount MLV is $ 00, the movement of the play position pp is almost stopped, and the same cutout waveform continues to be reproduced repeatedly. When the operation amount MLV is $ 40 at the center position, the play position pp is moved and reproduced at the speed of the set time expansion amount STCV set by the time expansion operation element 24.
(2) When the operation amount of the modulation lever 32 is $ 41 to $ 7F
As the operation amount MLV approaches $ 7F, the moving speed of the play position increases. When the operation amount MLV is $ 7F, the play position pp is moved and reproduced at a speed twice as long as the set time expansion amount STCV set by the time expansion operation element 24.
[0087]
As described above, when the waveform sequence is reproduced while the cut waveform is updated using the play position pp as a time reference, the time length (play time) required for waveform reproduction is determined by the user regardless of the pitch of the waveform to be reproduced. The time companding amount 24 can be determined by the time companding amount tcv set by operating the time companding operation element 24 and the modulation lever 32.
[0088]
In step L6, the reproduction pitch counter ppc, the first address counter ac1, and the second address counter ac2 are each incremented by one. Next, the reproduction pitch counter ppc and the reproduction pitch ppw are compared (step L7). This reproduction pitch ppw corresponds to the reproduction pitch. If the reproduction pitch counter ppc has not reached the reproduction pitch ppw, the process proceeds to “waveform reading process” in step L16 described later. The reproduction pitch ppw is calculated in step L9 described later.
[0089]
When the reproduction pitch counter ppc has reached the value of the reproduction pitch ppw, steps L8 to L15 are performed. In the processes in steps L8 to L15, the values of various parameters are updated in order to perform the process at the next reproduction pitch period. First, the reproduction pitch counter ppc is set to 0 (step L8). Next, a new reproduction pitch ppw is obtained by multiplying the pitch ratio pr (= note pitch np / original pitch op) based on the key whose pitch is designated by pressing the keyboard 31 and the cut waveform pitch cwp (step L9). ). Next, the window length wl of the envelope window is obtained by dividing the cut waveform pitch cwp by the formant shift amount fsv (step L9). Next, the offset os is set to 0.
[0090]
Next, the window length wl is limited within the reproduction pitch ppw (steps L10 and L11). That is, the window length wl is compared with the reproduction pitch ppw (step L10), and when the window length wl is larger than the reproduction pitch ppw, the window length wl is set as the reproduction pitch ppw (step L11). Further, the offset os is obtained by subtracting (reproduction pitch ppw × formant shift amount fsv) from the cut-out waveform pitch cwp. As will be described later, the offset os is a parameter for enabling the cut-out waveform to be cut out near the center of the envelope window. On the other hand, when the window length wl is less than or equal to the reproduction pitch ppw, the process of step L11 is not performed. As described above, the window length wl is limited so as not to be larger than the reproduction pitch ppw.
[0091]
Next, the reciprocal of the window length wl is obtained and used as the step rate wr (step L12). This step rate wr is used to step the first window counter wc1 and the second window counter wc2. Further, the polarity of the identification flag f is reversed. Since the process of step L11 is performed when the reproduction pitch counter ppc becomes equal to or larger than the reproduction pitch ppw in step L7, the identification flag F is also inverted when the reproduction pitch counter ppc becomes equal to or larger than the reproduction pitch ppw. For example, as shown in FIGS. 20 and 21 (c), a waveform that reverses to 1 and −1 with the period of the reproduction pitch counter ppc is obtained.
[0092]
Next, the value of the identification flag f is compared with 0 to determine whether the identification flag f is 1 or −1 (step L13). The value of the identification flag f being 1 means that the identification flag f has risen from -1 to 1, and in this case, the first address counter ac1 and the first window counter of the first processing system. It is assumed that wc1 is set to “0”, the first cutout start address fcsa is added to the current cutout start address ccsa by the offset os (step L14).
[0093]
Further, the value of the identification flag f being −1 means that the identification flag f has fallen from 1 to −1. In this case, the second address counter ac2 of the second processing system It is assumed that the second window counter wc2 is set to “0”, the second extraction start address scsa is obtained by adding the offset os to the current extraction start address ccsa (step L15).
[0094]
In this case, steps L14 and L15 are alternately executed every time it is determined in step L17 that the reproduction pitch counter ppc has exceeded the reproduction pitch ppw. Therefore, the first address counter ac1 and the second address counter ac2 As shown in (d) and (e) of FIG. 20, the cycle is twice as long as the reproduction pitch ppw, and the phases differ from each other by the reproduction pitch ppw. Also, the first cut-off start address fcsa is the falling part of the first address counter ac1, the second cut-out start address scsa is the falling part of the second address counter ac2, and the timings are different from each other by the reproduction pitch ppw. Will be updated.
[0095]
Subsequent to the processing of step L14 or L15 or when it is determined in step L7 that the reproduction pitch counter ppc has not reached the reproduction pitch ppw, a waveform reading process is performed (step L16).
[0096]
24 and 25 are flowcharts showing the waveform reading process. The waveform reading process will be described in detail below.
[0097]
Waveform readout processing
24 and 25 are flowcharts of the waveform reading process, in which steps N1 to N9 are processes for the first processing system, and steps N10 to N18 are processes for the second processing system. These two processes are performed in time series, but the contents of the processes are substantially the same.
[0098]
As shown in FIG. 24, in the waveform reading process, first, the value of the first window counter wc1 is incremented by the increment rate wr (step N1). Then, it is determined whether the incremented first window counter wc1 is less than 1, 1 or more, less than 2, or 2 or more (step N2). If it is smaller than 1, the value of the first window counter wc1 is set as the first envelope window ew1 (step N3). If it is 1 or more and smaller than 2, the value obtained by subtracting the first window counter wc1 from 2 is As the first envelope window ew1 (step N4), when it is 2 or more, the value of the first envelope window ew1 is set to 0 (step N5).
[0099]
In steps N1 to N5, for example, as shown in FIG. 20 (f), a sawtooth wave whose value increases by the step rate wr is created, and this value is turned back at the peak value “1”. One envelope window ew1 is created. However, if the window counter wc1 exceeds 2, the peak value of the first envelope window ew1 is set to “0” in step N5. That is, a triangular wave that increases to 1 by a step rate wr, which is the reciprocal of the window length wl determined based on the formant shift amount fsv and the cut waveform pitch cwp, and then decreases to 0 is changed to the waveform of the first envelope window ew1. It is created as.
[0100]
Further, following Steps N3 to N5, a value obtained by multiplying the first address counter ac1 (stepping value of the read address) by the formant shift amount fsv is added to the first cutout start address fcsa of the first waveform, The waveform data (sampling value) read address in the first processing system (hereinafter referred to as the first waveform read address) is designated as wra1 (step N6).
[0101]
Further, the first waveform read address wral1 is compared with the waveform read end address wea (step N7), and if it is larger than the waveform read end address wea, the gate value g = 0 is set (step N8). By setting the gate value g = 0, the output level is attenuated and silenced with a certain slope, as will be described later. If it is equal to or lower than the waveform read end address wea, the waveform data wd is read from the waveform data storage unit of the waveform memory 7 using the first waveform read address wral (step N9). As described above, the step width of the first waveform readout address wral1 is changed by the formant shift amount fsv. As a result, the readout speed of the cut-out waveform is changed by the formant shift amount fsv.
In subsequent steps N10 to N18, the same processing as described above is performed for the second processing system.
[0102]
Next, the output envelope level oel to be multiplied by the output o of the read waveform data and the waveform output wo are calculated (step l19). That is, the output envelope level oel is
“Output envelope level oel = output envelope level oel + (gate value g × velocity level vl−output envelope level oel) × K”
Ask for. The output envelope level oe1 is an envelope waveform that gradually rises at the start of waveform waveform reproduction, becomes 1 during waveform reproduction, and gradually decreases when waveform reproduction reaches the waveform read end address wea. K is a coefficient for determining the rising / falling slope.
The waveform output wo is
“Waveform output wo = Waveform data wd1 × first envelope window ew1 + Waveform data wd2 × second envelope window ew2”
Ask for. This is the sum of the waveform output cut using the first envelope window ew1 in the first processing system and the waveform output obtained using the second envelope window ew2 in the second processing system to obtain a reproduced waveform. Is.
[0103]
Finally, the final play waveform output pwo is
Play waveform output pwo = waveform output wo × output envelope level oel
Ask for. Therefore, the output level of the reproduced waveform sequence rises at a slope according to the coefficient K at the start of waveform waveform reproduction, and when the waveform reproduction reaches the waveform read end address wea, the output level falls and is silenced at a slope according to the coefficient K.
[0104]
【The invention's effect】
As described above, according to the present invention, the waveform represented by the waveform data stored in the storage means is reproduced with time axis compression / expansion and pitch change. Can be changed, and a natural reproduction waveform can be generated.
[Brief description of the drawings]
FIG. 1 is a diagram showing an overall block configuration of a waveform generator as an embodiment according to the present invention.
FIG. 2 is a diagram illustrating a data configuration example of a parameter 1 (status area) storage unit of a waveform memory in the embodiment apparatus;
FIG. 3 is a diagram illustrating a data configuration example of a parameter 2 (cutout start address area) storage unit of a waveform memory in the embodiment apparatus;
FIG. 4 is a diagram illustrating a data configuration example of a waveform data storage unit of a waveform memory in the embodiment apparatus;
FIG. 5 is a diagram illustrating various parameters set in a register on a CPU side in the embodiment apparatus.
FIG. 6 is a diagram illustrating various parameters set in a CPU-side MIDI information register in the embodiment apparatus;
FIG. 7 is a diagram illustrating various parameters set in a key information register on the DSP side in the embodiment apparatus.
FIG. 8 is a diagram illustrating various parameters set in a register on the DSP side in the embodiment apparatus.
FIG. 9 is a flowchart showing a main routine in the embodiment apparatus;
FIG. 10 is a flowchart showing a recording mode processing routine in the embodiment apparatus;
FIG. 11 is a flowchart showing an edit mode processing routine in the embodiment apparatus;
FIG. 12 is a flowchart showing a play mode processing routine in the embodiment apparatus;
FIG. 13 is a flowchart showing details of a play processing routine (1/2) in the embodiment apparatus;
FIG. 14 is a flowchart showing details of a play processing routine (2/2) in the embodiment apparatus;
FIG. 15 is a flowchart showing details of an assignment process to a voice module in a play process routine of the embodiment apparatus;
FIG. 16 is a diagram illustrating a relationship for performing conversion between a note number NN and a key follow coefficient KF in the embodiment device;
FIG. 17 is a flowchart showing a CPU interrupt processing routine in the embodiment apparatus;
FIG. 18 is a flowchart showing a main routine in the DSP in the embodiment apparatus.
FIG. 19 is a flowchart showing a “voice generation start process” in the main routine of the DSP in the embodiment apparatus;
FIG. 20 is a time chart for explaining an outline of the operation of “read processing” (no change in formant characteristics, pitch shift in a high frequency range).
FIG. 21 is a time chart for explaining the operation outline of “read processing” (shifting the formant characteristic to a low frequency range).
FIG. 22 is a flowchart showing a read processing routine (1/2) in the main routine of the DSP in the embodiment apparatus.
FIG. 23 is a flowchart showing a read processing routine (2/2) in the main routine of the DSP in the embodiment apparatus;
FIG. 24 is a flowchart showing a waveform read processing routine (1/2) in the read processing routine in the DSP main routine of the embodiment apparatus;
FIG. 25 is a flowchart showing a waveform read processing routine (2/2) in the read processing routine in the DSP main routine of the embodiment apparatus;
[Explanation of symbols]
1 CPU
2 controls
21 Mode selection switch
22 Bank selection switch
23 Formant shift control
24 hours companding operator
3 Keyboard device
31 keyboard
32 Modulation lever
4 Hard disk devices
5 storage unit
6 DSP (digital signal processor)

Claims (1)

Storage means for storing waveform data representing a series of waveforms arranged in time series;
Compression / decompression information input means for inputting compression / decompression information that indicates the degree of compression or expansion of the time axis of the waveform;
Pitch information input means for inputting pitch information indicating the playback pitch of the waveform;
A formant shift amount input means for inputting a formant shift amount for shifting the formant of the waveform;
Formant control information generating means for generating formant control information in which the formant shift amount is changed according to the pitch information;
The compression / decompression information, pitch information, and formant control information are input, and a waveform corresponding to a playback position that changes at a change speed corresponding to the compression / decompression information is associated with the pitch information based on the waveform data. And a reproducing means for generating a reproduction waveform by changing the pitch and changing the formant in response to the formant control information.

Images