JP3404756B2 - Music synthesizer - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [0001] BACKGROUND OF THE INVENTION This invention relates to synthesizing formant sounds.
The present invention relates to a musical sound synthesizer to be formed. [0002] 2. Description of the Related Art The sounding mechanisms of natural musical instruments and human voices are directly
Vibration is excited by "exciter" and excited by exciter
What is considered to be composed of a "resonator" that resonates with vibration
Can be. For example, as shown in FIG.
The string is the “exciter” and the body is the “resonator”. Sack
The lead is an exciter and the tube is a resonator.
You. The vocal cords in the human voice pronunciation mechanism are "exciters"
And the vocal tract is the "resonator". Where the exciter
Of the spectrum that composes the sound by the pitch of the sound to be sounded
The distribution changes. FIG. 13A shows a low pitch sound.
If you utter (upper) and utter a high pitched sound
Of the vibration generated in the vocal cords as the exciter
It is an example of a torr. As shown in these figures,
As the pitch of the generated sound increases, the spectrum distribution becomes
With the envelope shape of each spectrum maintained, the horizontal axis (frequency
The number is uniformly extended in the direction. Such a pic
The behavior of the spectrum distribution with respect to the
For convenience, it is called "moving formant." On the other hand,
As shown in FIG. 13 (b), the vocal tract, which is a sounding device, starts from the vocal cords.
A certain transmission rate regardless of the pitch of the sound vibration
Maintain the obtained frequency characteristics. Thus, the pronunciation or
A constant formman regardless of the pitch of the sound
In the following, we refer to the property of outputting
" As described above, the "exciter" and the "resonator"
Is the variation of the spectrum envelope characteristic with the pitch of the performance.
The way of transformation is different. In addition, to human voice or musical instrument
The contribution of "exciter" and "resonator" to the tone
Are different. For example, in the case of human voice,
The characteristics of the vocal tract, which is the instrument, are dominant and fixed
It becomes a tone like G. On the other hand, in the case of brass instruments,
The sound of a moving formant is not so affected
Becomes Furthermore, in the case of a guitar or the like, the "exciter"
It has a sound with both features of the “instrument”. [0004] Conventionally, as a musical sound synthesizer, FM
Sampler using the PCM method
Etc. are known. The synthesized sound of these is basically
The horizontal axis (frequency)
This is a so-called moving formant sound that expands and contracts in the
You. For this reason, conventionally, the spectrum is not affected by the pitch change.
The so-called fixed formant where the uneven shape is fixed
Various ideas have been devised to express sound. example
For example, in a tone synthesizer using the FM system,
Fixed formman by key scaling exponent
It expressed a tone like a tone. In addition, waveform record of sampler etc.
In a musical sound synthesizer using a memory / readout method,
The keyboard range is finely divided, and each section
Live instrument sounds recorded at different pitches and their waveforms
Take in. Then, play the sample waveform for each range
This creates a fixed formant tone
Was. On the other hand, as a device dedicated to synthesis of fixed formant sounds
Is known to speech synthesizers such as LPC type and PARCO type.
Have been. Other examples are described in JP-A-2-254497.
And JP-A-2-271713. [0005] By the way, the above-mentioned F
In a tone synthesizer using the M system, a specific
Except for the range, it can express a fixed formant tone well
Did not. On the other hand, in the sampler described above, one
If you sample for one key, fixed format
It is thought that the unique tone can be expressed well. However
However, if you do that, the hardware configuration will be complicated
At the same time, the production cost increases significantly, cost / performance
It is expected that it will not match the ratio and is not realistic. this
As in the past, conventional tone synthesizers are not
I can not cope with a fixed formant tone at all times,
Of a dedicated fixed formant sound synthesizer
There was a problem that. The present invention has been made under such a background.
Can express a fixed formant tone well
It is an object of the present invention to provide a musical tone synthesizer. [0007] SUMMARY OF THE INVENTION The present invention has been described in the foregoing.
To solve the problem,Around the original sound waveform consisting of multiple cycles
Waveforms stored as musical tone waveform data divided for each period
A memory for storing the beginning of each period section of the musical tone waveform data;
And the last sample amplitude value is 0,
Section designating means for designating the section of the moly
Delay each other by the time corresponding to the pitch
Musical tone in the section of the waveform memory specified by the means
Because the waveform data is read from the beginning to the end of the section,
Generate multiple N phase data that changes at a constant rate
Phase data generating means, and according to the N pieces of phase data,
The tone waveform data of the section specified by the section specifying means
Reading means for reading N data from the waveform memory
And an N-sequence tone waveform read from the reading means.
And adding means for adding each of the data.
The interval designating means is a waveform memory designated at a predetermined timing.
Change section ofIt is characterized by the following. [0008] According to the above arrangement, each of the pitches to be pronounced is
Delay each other by the corresponding time and store in the above waveform memory
From the value corresponding to the start point of the cycle of the
Multiple that change at a constant rate to a corresponding valueN piecesPhase data
Is generated. And, from the waveform memory, the mark
One cycle of waveform data starting with the waveform data marked with
Read multiple operationsN piecesAccording to the phase data ofN series
MinuteIs executed and after one cycle of waveform data is read
Is supplied with phase data corresponding to the starting point of the cycle.
In response, the waveform of the cycle next to the cycle already read
Data is read. In addition, each read waveform data
Data waveform values are added and output as an integrated waveform value.
Waveform data has not been read from the waveform memory.
For a period of time, an integrated waveform value of amplitude "0" is output.
You. As a result, the fixed form
Sound is synthesized. [0009] BRIEF DESCRIPTION OF THE DRAWINGS FIG.
Will be described. FIG. 1 shows an embodiment of the present invention.
FIG. 2 is a block diagram illustrating a configuration of a musical sound synthesizer. Figure smell
Reference numeral 1 denotes a tone color selection unit. The operator selects the tone selection section
By operating the operation element provided in 1, the sound is generated.
The sound to be played is a moving formant sound or fixed
Sets whether the sound is a formant sound. 2 is the tone setting section
is there. The operator can hear the moving formant sound and the fixed form
When synthesizing a sound including a cloak sound, the timbre setting unit 2
Sounds by operating the controls provided on the
Moving and fixed formants in power sounds
Set the mixing ratio with the sound. Then, the tone color selection unit 1 and
The setting information set by the timbre setting unit 2 includes a timbre parameter.
It is stored in the meter storage unit 3. Reference numeral 4 denotes a key press detection assignment circuit, which is operated by the operator.
When any key on the keyboard 5 is pressed,
Key code signal KC corresponding to the key
KONP is output. Reference numeral 6 denotes a waveform memory which is recorded in advance.
The sample waveform of the sampled original sound will be
It is memorized in a state where it has been worked. Reference numeral 7 denotes a fixed formant synthesis section,
Instructs the parameter storage unit 3 to synthesize a fixed formant sound
Is stored, the key press detection assignment circuit 4
The key code signal KC and the key-on pulse KONP
And synthesizes a fixed formant sound based on this.
Note that the fixed formant synthesis unit 7
The synthesis of the G tone will be described later. Reference numeral 8 denotes a moving formant synthesizing unit,
Instructs parameter storage unit 3 to synthesize moving formant sound
Is stored, the key press detection assignment circuit 4
The key code signal KC and the key-on pulse KONP
And read according to the pitch corresponding to the key code signal KC.
Read sample waveform from waveform memory 6 at readout speed
And synthesize a moving formant sound. The multipliers 9, 10 and the outputs of these multipliers
A fixed formant is added by the adder 11 for adding the force signal.
A mixer that mixes sound and moving formant sounds is configured.
Have been. Here, the fixed formant sound and the moving form
The mixing ratio of the rumant sound is supplied to multipliers 9 and 10.
Is determined by the multiplication coefficient
The arithmetic coefficient is setting information stored in the tone parameter storage unit 3.
Is generated in accordance with. 12 is a sound system
There is a fixed formant sound taken from the adder 11
The sound is generated based on the signal of the mixed sound with the moving formant sound.
U. Next, preprocessing to be performed in this embodiment
A description will be given below. This preprocessing means recording the original sound.
And process the sample waveform and store it in the waveform memory 6
Refers to processing. Hereinafter, in the present embodiment shown in FIG.
A description will be given with reference to a flowchart of preprocessing to be performed.
Make a light. First, in step S1, a musical performance is performed.
Recording of the original sound. Next, the process proceeds to step S2.
And pitch extraction of the original sound collected in step S1
And obtain pitch data. Pitch data is multiple cycles
Can be determined by the average of In general,
Switch extraction, taking into account the enormous amount of calculation,
In the example, pitch extraction is relatively easy with non-real-time analysis.
Do. The processing in steps S1 and S2 is performed by a computer.
Done on Next, proceeding to step S3, step S3
2 based on the pitch data obtained in
Cut out every pitch cycle. At this time,
The closest zero-cross point from the
Int is the boundary of the periodic waveform. Then, from that border,
Cut out at the pitch cycle obtained in step S2, and
Search for the nearest zero crossing point. Repeat this process
You. Processing to find the cutout point of such a periodic waveform is not
This may be performed after performing a low-pass process on the force waveform. here
FIG. 3 is a diagram showing an example of an original sound waveform, and a vertical line L in FIG.
Indicates the boundary of the average pitch period. As shown in P in the figure
In addition, the amplitude of the original sound waveform becomes “0” at the boundary of the average pitch period.
Not. In this state, the waveform is cut out and the waveform
When stored in the memory 6, change the read pitch cycle
This creates a discontinuous point in the waveform, resulting in a noisy sound.
I will. In the zero cross detection process,
The boundary of the first cycle can approach the amplitude “0”,
This is not enough. Therefore, next step S4
Otherwise, the process proceeds to step S4 ', and at the boundary of the pitch cycle
The phase or amplitude is forcibly set to "0". When the process proceeds to step S4, the cut-out waveform
After performing FFT (Fast Fourier Transform) analysis every time,
Perform FT and resynthesize. As a result, starting from phase “0”
A complete waveform is obtained. On the other hand, when the process proceeds to step S4 '
Multiply the window waveform for each clipped waveform,
The amplitude at the boundary of the h cycle is set to “0”. Fig. 4
FIG. 7 is a diagram showing a state in which a window waveform is multiplied by an output waveform.
You. Step S4 or step S4 'is completed.
Then, the process proceeds to step S5. In step S5, the boundary of the pitch cycle is reached.
That is, after the boundary of the cutout waveform can be identified,
It is stored in the shape memory 6. FIG. 5 shows the recording of the waveform memory 6.
It is a figure showing contents. The first field contains the original sound
The sample rate at the time of pull is recorded and the second file
Field contains the pitch of the sampled tone waveform,
The sample key is recorded. And following these
The waveform data cut out to the field
And are sequentially recorded. FIG. 5 shows the waveform data.
An example in which the MSB of the data is a mark bit is shown.
You. At the start point of each pitch cycle, a swing whose MSB is "1" is set.
Waveform data of width “0” is stored, and the pitch cycle
The MSB of the waveform data in the middle is “0”. here
The mark bit is either MSB or LSB.
May be used. Thus, the data stored in the waveform memory 6 is stored.
Sampled waveforms are not suitable for synthesizing fixed formant sounds.
Read at time intervals corresponding to various pitch periods.
You. FIG. 6 shows the case where the sample waveform is reproduced at a low pitch and the case where the sample waveform is reproduced at a high pitch.
FIG. 7 is a waveform diagram showing a case where pitch reproduction is performed. (A) is the original sound
(B) shows a low-pitch reproduction.
(C) shows a waveform read out at a high pitch.
The read waveform is shown. For low pitch playback, individual
The time interval for reading out the waveform data of
The same as the sample period, but the amplitude is
By inserting a section of “0”, the pitch period becomes t ′.
ing. On the other hand, in the case of high pitch playback, individual waveform data
The read cycle of the sample is the same as for low-pitch playback.
Read at the read speed corresponding to the sample period during ringing.
But before the reading of the previous waveform is finished,
Time t ″ has elapsed since the start of reading the previous waveform
At this point, the next waveform is read. And read
High pitch by superimposing each wave and adding amplitude
A readout waveform is obtained. Next, referring to FIG. 7 and FIG.
The configuration of the constant formant synthesis unit 7 will be described in detail. FIG.
Is the configuration of the fixed formant synthesizer 7 (including the waveform memory 6).
It is shown in the figure. FIG. 8 is a block diagram showing
FIG. 7 is a diagram showing a relationship between contents stored in a waveform memory 6 and each address.
is there. In FIG. 7, reference numerals 61a to 61d denote address generators.
Waveforms whose phase is shifted by the pitch period during playback.
The relative addresses AG1 to AG4 of the memory 6 are set to 50 kHz._ZAddress of
Output at the output rate. The relative addresses AG1 to AG4
Means the sample waveform to be read, as shown in FIG.
The address of the start point of the cycle is set to the start address "0".
This is the relative address of the waveform data. Also, in FIG.
Here, an address selection signal source (not shown) is used.
Address generation by the address generators 61a to 61d.
Four times the raw rate, ie 200 kHz_ZPal
Is generated as the address selection signal ADSEL. Reference numerals 62a to 62d denote latch circuits each having an address.
Generators 61a-d correspond to each
The absolute addresses AA1 to AA4 of the starting point of the cycle in the waveform are
Written. These absolute addresses AA1 to AA4 are shown in FIG.
As shown in the figure, starting from the start address of the waveform memory 6
Absolute address to find the start of the period of each sampled waveform
It's about les. 63a to 63d are adders,
Relative addresses output from the generators 61a to 61d
AG1 to AG4 and the gates output from the latch circuits 62a to 62d.
The paired addresses AA1 to AA4 are added for each, and the waveform
Output as effective addresses AD1 to AD4 of memory 6. Numeral 64 denotes a waveform address selector, which is an adder.
The effective addresses AD1 to 4 output from 63a to 63d are
Each time the address selection signal ADSEL is generated, AD1
4 and sequentially selected and output. 65 is a selector
The effective address output from the waveform address selection unit 64.
AD1 to AD4 are input to the input terminal A, and the mark
Period of sample waveform output from address supply unit 69
Signal MA for searching for the starting point of
Is entered. Then, these are applied to the address selection signal ADS.
Select according to EL and sequentially output to waveform memory 6
You. Reference numeral 66 denotes a mark detection unit, and the waveform memory 6
When the mark bit is detected from the read data of
And outputs a peak detection signal MF. 67 is a selector,
A mark detection signal MF is input, and this is used as an address selection signal.
Output from output terminal A or B according to ADSEL
You. That is, the address selection signal ADSEL is High.
In the case of the level, the mark reading pulse MKW is output from the output terminal A.
Is output. On the other hand, when the address selection signal ADSEL is Lo
In the case of the w level, the mark search pulse MK is output from the output terminal B.
Output M. Reference numeral 68 denotes a flip-flop circuit.
The mark search pulse MKM output from the
From the pitch clock generator 81 described later.
It is reset by the output phase pulse φ.
An adder 70 has a start address STA to be described later.
D is added with “1” and output. Mark address supply unit
69 is the timing at which the key-on pulse is captured,
The value output from the arithmetic unit 70 is counted up as an initial value.
Start the loop. And from the flip-flop circuit 68
The counting is stopped and restarted based on the output signal Q1 output.
Opening is controlled, and the count value at each time is searched for the starting point of the cycle.
As an address signal MA. 71 is a latch
And a mark search path output from the selector 67.
Luth MKM is received as a latch signal.
To find the starting point of the cycle output from the
The address signal MA is latched, and this is latched to the address signal MA.
Output as A. Reference numerals 72a to 72d denote selectors. Selector 7
2a, start address STAD is input to input terminal A
Address signal MA output from latch circuit 71
A is input to the input terminal B. Selectors 72b-d
Address “0” is input to input terminal A, and latch circuit 7
1 is input to the input terminal B.
Is forced. When the key-on pulse KONP is at the high level
In this case, the selector 72a sets the start address STAD
Output to the latch circuit 62a and the selectors 72b-d
Outputs the address “0” to the latch circuits 62b to 62d.
Power. On the other hand, the key-on pulse KONP is at the low level.
In this case, the selectors 72a to 72d output the address signal MAA.
It outputs to each of the latch circuits 62a-62d. 73a-
d is an OR circuit, which is a flip-flop circuit to be described later.
Count enable signal C output from paths 80a-d
The OR of E1 to E4 and the key-on pulse KONP
The latch signal is output to the latch circuits 62a to 62d. Reference numeral 74 denotes a latch circuit, which is an address selection
When receiving Lus ADSEL, a different 4
Latch waveform data read by phase address
And output this. 75 is an accumulator,
Waveform data of four different phases output from the latch circuit 74
And outputs the result. The pitch clock generator 81 has a key code
The frequency number corresponding to the key code is stored.
When the signal KC is taken in, the corresponding frequency number is
Accumulate next and generate phase pulse φ based on this result
I do. Therefore, the pulse of the phase signal φ is
Generated in response to the first cycle. 76 is the start address supply
And the key-on pulse KONP
Output the start address STAD stored in the memory
You. This start address STAD is, as shown in FIG.
At the address where the first waveform data of the first wave is stored
is there. Both 77 and 78 are demultiplexers.
is there. The demultiplexer 77 includes the pitch clock generator 8
The phase pulse φ output from 1 is converted into four pulses DP
11 to 14 and output. Demultiplexer 78
Is the mark reading pulse MKW output from the selector 67
Is output while being dispersed into four pulses DP21 to DP24. So
And the four pulses output from the demultiplexer 77
Switches DP11 to DP14 are flip-flop circuits 80a to 80d
Are respectively taken as set signals. Meanwhile, demulti
Four pulses DP21 to DP2 output from the plexer 78
4 is a key-on pulse K by OR circuits 79a to 79d, respectively.
After being ORed with ONP, the flip-flop circuit 80
a to d are fetched as reset signals. Flip flip
Drop circuits 80a-80d provide these set signals and reset signals.
Count enable signals CE1 to CE4 corresponding to the reset signals
, And address generators 61a-d and latches, respectively.
Output to circuits 62a-62d. Next, the wave generated by the fixed formant
The read operation of the shape memory 6 will be described. FIG.
Relative address output from dress generators 61a-d
FIG. 7 is a timing chart of signals related to the signals AG1 to AG4. A
Each of the dress generators 61a to 61d
Address counting according to the bull signals CE1 to CE4.
While 50 kh_ZRelative address according to the address generation rate of
The dresses AG1 to AG4 are output. That is, in FIG.
Thus, for example, the address generator 61a
The enable signal CE1 is changed from a low level to a high level.
Starts counting from the initial value "0"
You. Then, the count enable signal CE1 becomes High.
Keep counting up during the level. Then count
When the enable signal CE1 changes from the High level to the Low level.
When it changes to a count, it stops counting. And county
While the enable signal CE1 is at the low level, the counting is stopped.
Keep a fixed address while stopped. Next, count rice
Cable signal CE1 goes from low level to high level
When it changes, clear the address to the initial value "0" and then again
Start counting up. Address generator 61
b to d, the count enable signal CE2
A similar counting operation is performed according to. Also,
Key-on pulse KON for dress generators 61a-d
When P is captured, each address is initialized to "0"
Cleared. The address generators 61a to 61d
Addresses AG1 to AG4 output from the latch circuit 6
The end of the next periodic waveform output from 2a to d to be read
The absolute addresses AA1 to AA4 indicating the heads are added to the adders 63a to 63d.
Respectively. And the waveform in the waveform memory 6
Effective addresses AD1 to AD4 for reading data
Is output. Next, referring to FIG.
After the data AD1 to AD4 are obtained, the address is stored in the waveform memory 6.
The operation up to the supply will be described. Figure 10 shows the waveform memory
6 of the signal related to the generation of the address signal supplied to
FIG. In this figure, SC is the system of this device.
It is a stem clock. In addition, WA selects the waveform address.
Is an address signal output from the section 64, and SA is
This is an address signal output from the rectifier 65. Waveform add
Address selector 64 takes in the effective addresses AD1 to AD4.
And AD1 to AD1 in synchronization with the address selection signal ADSEL.
The address signals WA arranged in the order of 4 are output. Soshi
Then, the selector 65 determines the start of the cycle with the address signal WA.
Waveform memory based on address signal MA for searching for a point
6 is supplied with an address signal SA. That is,
When the address selection signal ADSEL is at a high level,
Addresses AD1 to AD4 are output and an address selection signal ADS is output.
When EL is at the low level, the search for the starting point of the cycle is performed.
Output addresses M1, M2, M3,... Accordingly
To find the starting point of the cycle
3,... Always precede the effective addresses AD1 to AD4
That is, the starting point of the periodic waveform to be read next in advance
Will be obtained. Thus, the address is stored in the waveform memory 6.
When supplied, read waveform data from waveform memory 6
Alternatively, the start point of the cycle is detected. And
Switch 74 latches the address selection signal ADSEL
Received as a waveform
Latch the waveform data read from memory 6 and
Output. Then, the accumulator 75 operates as a latch circuit.
Four different phases of waveform data output from the path 74 are added.
And output. Next, referring to FIG.
6, after detecting the mark bit, the absolute addresses AA1 to AA1
4 is output from the latch circuits 62a to 62d.
explain. FIG. 11 is a diagram showing signals related to detection of a start point of a cycle.
It is a timing chart. When reading the waveform memory 6,
When a mark bit is detected, the mark detection unit 66
And outputs a lock detection signal MF. Then, the selector 67
In accordance with the address selection signal ADSEL.
Mark reading pulse MKW or mark search pulse
Output as MKM. That is, the mark detection signal MF
High level of the address selection signal ADSEL and Hi of the address selection signal ADSEL
To read out waveform data when gh levels match
Mark bit is detected by the effective addresses AD1 to AD4 of
At this time, the mark reading pulse MKW
Output. On the other hand, the high level of the mark detection signal MF
And the high level of the address selection signal ADSEL match
If not, the addresses M1,
That the mark bit is detected by M2, M3, ...
At this time, a mark search pulse MKM is output. The flip-flop circuit 68 selects
The mark search pulse MKM output from the
Signal from the pitch clock generator 81
Take the output phase pulse φ as a reset signal
Put in. Then, the output signal Q1 corresponding to these signals is
Output. Further, the mark address supply unit 69
When the on-pulse KONP is taken in, it is taken out from the adder 70.
Add "1" to the start address STAD
Address select signal ADSE with the address as the start address
Start counting up according to L. And flip
Based on output signal Q1 output from flop circuit 68
To stop and restart counting. That is,
The output signal output from the flip-flop circuit 68 is High.
During the level, the address is counted up and the cycle starts
Address to search for point ..., M_k-2, M_k-1Output as
Then, the address counting is stopped during the Low level.
M of the start point of the cycle_kIs output. And la
The latch circuit 71 outputs a mark search pulse MKM to a latch signal.
Output from the mark address supply unit 69
Mark address M_kAnd latch this to the address signal.
No. MAA is output. The selector 72a is a key-on pallet.
When receiving the KONP, the star taken into the input terminal A
Select the address STAD and send it to the latch circuit 62a.
Output. The selectors 72b to 72d are key-on switches.
When receiving Luz KONP, the add
Address "0", and for each of the latch circuits 62b to 62d.
Output. At this time, the latch circuits 62a to 62d
Start by receiving latch signals from OR circuits 73a to 73d
Latch address STAD or address "0"
Output as absolute addresses AA1-4. Meanwhile, Selek
Are not receiving the key-on pulse KONP.
The address signal MAA taken into the input terminal B
And outputs it to each of the latch circuits 62a-62d.
Then, of the latch circuits 62a to 62d, the count enable
From the OR circuits 73a to 73d based on the bull signals CE1 to CE4.
The one receiving the latch signal latches the address signal MAA.
And outputs as absolute addresses AA1 to AA4. Examples and
The count enable signal CE3 and the absolute address AA
3 is illustrated. The above is the fixed formant synthesis unit 7
This is an explanation of the read operation of the waveform memory 6 according to the first embodiment. Next, the operation of the musical sound synthesizer according to the present embodiment.
Will be described. The operator is, for example, a fixed formman
8: 2 ratio between the tone like a moving sound and the moving sound like a formant
Operation of the tone selection unit 1 when playing back a sound containing
Manipulate the child to make fixed and moving formants
Select both. Then, the operator of the tone setting section 2 is operated.
Mix fixed formant sound and moving formant sound
Set the ratio to 8: 2. As a result, the tone parameter
Remember 3 has fixed and moving formant sounds
The setting information that the playback ratio is 8: 2 is stored.
You. Next, the operator plays a musical tone using the keyboard section 5.
I do. The key press detection allocation
Key code corresponding to key-on pulse KONP from road 4
The signal KC is output. And a tone parameter storage unit.
In accordance with the setting information of 3, the transfer to the fixed formant synthesis unit 7 is performed.
Musical sound synthesis is performed by the dynamic formant synthesis unit 8. Solid
The constant formant synthesizing unit 7 outputs
The applied key-on pulse KONP and the key code signal KC
Of the key code signal KC.
The waveform memory 6 is read according to the
Synthesizes the rumant sound. On the other hand, moving formant synthesis
The unit 8 includes a key-on pulse KONP and a key code signal KC
And the key code signal K as in the conventional sampler.
Waveform memory with read speed according to pitch corresponding to C
6 is read out and the moving formant sound is synthesized.
U. Output from the fixed formant synthesizing section 7
The signal of the fixed formant sound is converted into a timbre by the multiplier 9.
Multiply by the mixture ratio 0.8 stored in the parameter storage unit 3
The information is output. On the other hand, the moving formant synthesizing unit 8
The moving formant sound signal output from the
0, the mixture stored in the timbre parameter storage unit 3
The output is multiplied by the ratio 0.2. And the multiplier
9 and the signal of the fixed formant sound output from 9 and the multiplier 1
Adds the moving formant sound signal output from 0
Are added in the mixer 11 and output as a mixed sound signal.
You. The sound system 12 outputs from the adder 11
The sound is generated based on the mixed sound signal. As described above, according to this embodiment,
For a fixed formant sound, the pitch at the time of sampling
The original formant is fixed even when playing at different pitches
It will be played back while being dropped. For this reason, conventional samplers
In order to express a fixed formant sound,
It is necessary to prepare a sample waveform for each range by dividing
Disappears. In addition, the operator selects the tone selection unit 1 and the tone setting.
Operate the control of the fixed section 2 to set the fixed folder corresponding to the original sound.
You can set the ratio between the cloak sound and the moving formant sound.
it can. In addition, the operator intentionally has no relation to the characteristics of the original sound.
Set the ratio between fixed formant sound and moving formant sound.
By setting, the tone of nuance different from the original sound
Obtainable. Furthermore, in this embodiment,
The sample rate is fixed at the sample rate, but this is variable
To shift the formant frequency up and down.
You can change the tone
And the range of application expands. Also, as in this embodiment,
Then, the conventional PCM method is used for the moving formant synthesizing unit 8.
If you use the fixed formant synthesis unit 7,
The waveform memory 6 can be shared, and the configuration can be simplified.
In this case, the reading of the waveform memory 6 by both is time-division
Do with. Similarly, address generators 61a to 61d
May be performed in a time-sharing manner. Also add
The number of less generators is not necessarily limited to four.
In the case of four as in this embodiment, the limit of pitch up
The range will be limited to two octaves. The address of the sample waveform to be marked
If the waveform is stored in a separate memory from the waveform memory 6,
It is possible to further simplify the configuration of the constant formant synthesis unit 7.
it can. Also, read the same sample waveform repeatedly.
Control the speed of tone change.
You can control. [0040] As described above, according to the present invention,
IfBy dividing the original sound waveform consisting of multiple periods into each period section
A waveform memory for storing musical tone waveform data;
The first and last sample waveforms of each period section of the musical tone waveform data
Specify a width value of 0 and a section of the waveform memory.
Section designation means and when each corresponds to the pitch to be pronounced
Between each other, and the
The musical tone waveform data in the section of the waveform memory
Since it reads from the beginning to the end, it changes at a predetermined rate.
Data generating means for generating a plurality of N phase data
And according to the N phase data,
Therefore, the tone waveform data of the designated section
Reading means for reading N series from the memory, and said reading means
Each of the N series of tone waveform data read from
And the section designating means,
Change the section of the waveform memory specified at the timing of
soPerform at a pitch different from the pitch at which the original sound was sampled.
Even if played, the characteristics of the original sound such as formants
Be born. For this reason, like a conventional sampler,
To express the formant tone,
There is no need to divide and prepare sample waveforms for each range.
It becomes.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing a configuration of a musical sound synthesizer according to an embodiment of the present invention. FIG. 2 is a flowchart showing preprocessing to be performed in the embodiment. FIG. 3 is a diagram showing an example of an original sound waveform. FIG. 4 is a diagram showing a state in which a cutout waveform is multiplied by a window waveform. FIG. 5 is a diagram showing recorded contents of a waveform memory 6; 6A and 6B are waveform diagrams showing a case where a sample waveform is reproduced at a low pitch and a case where a sample waveform is reproduced at a high pitch, wherein FIG. 6A shows a sample waveform of an original sound, and FIG. , (C) show the readout waveform by the high pitch reproduction. FIG. 7 is a block diagram showing a configuration (shown including a waveform memory 6) of a fixed formant synthesis unit 7; FIG. 8 is a diagram showing a relationship between storage contents of a waveform memory 6 and respective addresses. FIG. 9 is a diagram showing waveforms of signals related to relative addresses AG1 to AG4 output from address generators 61a to 61d. FIG. 10 is a diagram showing signals related to generation of an address signal supplied to the waveform memory 6; FIG. 11 is a diagram showing signals related to detection of a start point of a cycle. FIG. 12: “Exciter” for musical instrument and human voice
And Tables corresponding to "resonators". FIG. 13 is a diagram illustrating a difference in spectrum envelope characteristics between a vocal cord and a vocal tract with respect to a pitch of a pronunciation pitch when a human voice is used as an example. [Description of Signs] 1... Tone color selection unit 2... Tone color setting unit 3... Tone color parameter storage unit 4. ... Fixed formant synthesizer, 8 ... Moveable formant synthesizer, 9,10 ... Multiplier, 11,63a-d, 70 ... Adder, 12 ... Sound system, 61a-d ... Address generator,
62a-d, 71, 74 ... Latch circuit, 64 ... Waveform address selector, 65, 67, 72a-d ... Selector, 66 ... Mark detector, 68, 80a-d ... Flip-flop circuit, 69 ...... Mark address supply unit, 7
3a-d, 79a-d ... OR circuit, 75 ... accumulator, 76 ... head address supply unit, 77,78 ...
Demultiplexer, 81: pitch clock generator

Claims (1)

(57) [Claims] [Claim 1] An original sound waveform composed of a plurality of cycles is generated for each period section.
Waveform memory that stores data as musical tone waveform data
At the beginning and end of each period section of the musical tone waveform data.
A sample amplitude value of 0, a section specifying means for specifying a section of the waveform memory, and a delay corresponding to a pitch corresponding to a pitch to be generated.
And the waveform memo specified by the section specifying means.
From the beginning of the section to the end of the section
N phases that change at a predetermined rate to read
Phase data generating means for generating data; and the section specifying means according to the N pieces of phase data.
The tone waveform data of the section specified by
Reading means for reading N series from the readout means, and N-series musical sound waveform data read from the reading means.
And an adding means for adding each of the data, wherein the section designating means designates a waveform designated at a predetermined timing.
A musical sound synthesizer characterized by changing a memory section .