JP3659121B2 - Music signal analysis / synthesis method, music signal synthesis method, music signal synthesis apparatus and recording medium - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a method for analyzing and synthesizing a musical sound signal used in an apparatus for generating musical sounds, such as an electronic musical instrument and an amusement device, and more particularly to analyzing and synthesizing a musical sound signal suitable for an electronic musical instrument that reproduces a natural musical instrument sound. The present invention relates to a method, a musical tone signal synthesis method, a musical tone signal synthesis apparatus, and a recording medium.
[0002]
[Prior art]
Conventionally, a waveform memory sound source has been used to generate a musical sound signal. As one type, there is known one that records a plurality of periodic waveforms with an envelope and reads them out to generate a musical sound waveform. However, since this method requires a large memory capacity, when there is a cost limitation, a method is adopted in which an envelope is given while recording a loop waveform and repeatedly reading it out.
[0003]
[Problems to be solved by the invention]
By the way, the sound generated by a natural musical instrument includes a noise component due to a physical cause specific to the musical instrument, and this noise component contributes to the expression unique to each musical instrument. However, when a musical instrument sound waveform (loop waveform) is recorded with a relatively short loop size and reproduced repeatedly, there is a problem that this noise component is heard. That is, when the loop waveform is repeatedly reproduced, naturally, the same noise component is repeatedly reproduced, so that there is a problem that a beat sound different from the original noise component can be heard.
[0004]
For example, in the case of a saxophone sound, when the loop size is short, the noise component sounds like a buzzer, and when the loop size is slightly increased, it sounds like jitter. For this reason, in order to reproduce a natural noise component, it is necessary to set the loop size to be considerably large.
The present invention has been made in view of the above-described circumstances, and provides a tone signal analysis / synthesis method, a tone signal synthesis method, a tone signal synthesis apparatus, and a recording medium that can add a natural noise component to a tone signal. It is aimed.
[0005]
[Means for Solving the Problems]
  In order to solve the above problems, the present invention is characterized by having the following configuration. The parentheses are examples.
  In the configuration according to claim 1, the process of analyzing the frequency component of the original waveform (FFT analysis processing unit 34), and the noise component (discontinuous frequency component on the time axis) based on the analysis result ( 50) extracting (subtracting unit 46) and the noise componentBased on noise component waveform dataTheFirstA process of storing in a memory (noise waveform creation unit 48);A process of extracting continuous periodic components on the time axis based on the periodic components of the original waveform (continuous component separating unit 36), and periodic component waveform data is stored in the second memory based on the extracted frequency components. A process (waveform generator 38), a process of generating a read address that changes at a speed corresponding to a desired pitch (address generator 62), and the periodic component waveform data from the second memory is repeated using the read address. Read, based on the read periodic component waveform dataA periodic component signal generation process (periodic waveform generator 60) for generating a periodic component signal;When it is detected that the read address has reached a predetermined trigger value, a trigger signal generation process (synchronization control unit 102) for generating a trigger signal and the noise component from the first memory according to the generation of the trigger signal A noise component signal generation process (noise waveform generator 80) for reading waveform data and generating a noise component signal based on the read noise component waveform data;Generated periodic component signalWhenNoise component signalProcess of synthesizing(Mixing & Reproducing Unit 112).
  Moreover, in the structure of Claim 2,A first memory (noise component memory 84) that stores noise component waveform data of noise components that are discontinuous frequency components on the time axis, and periodic component waveform data of periodic components that are continuous frequency components on the time axis In a musical sound signal synthesis method for reading out a second memory (periodic component memory 64) for storing a musical sound signal and synthesizing a musical sound signal, a process of generating a read address that changes at a speed corresponding to the pitch (address generator 62); A periodic component signal generation process (periodic waveform generator 60) for repeatedly reading the periodic component waveform data from the second memory by the read address and generating a periodic component signal based on the read periodic component waveform data; and the read address Trigger signal generation process (synchronization control unit 102) for generating a trigger signal when it is detected that has reached a predetermined trigger value; A noise component signal generation process (noise waveform generator 80) that reads the noise component waveform data from the first memory in response to generation of a trigger signal and generates a noise component signal based on the read noise component waveform data; A synthesis process (mixing and reproducing unit 112) for synthesizing the generated periodic component signal and noise component signal;It is characterized by having.
  Moreover, in the structure of Claim 3,In a tone signal synthesis method for synthesizing a tone signal by reading out a first memory (noise component memory 84) that stores noise component waveform data of a noise component that is a discontinuous frequency component on the time axis,A periodic component signal generation process (periodic waveform generator 60) for generating a periodic component signal having a continuous frequency component on the time axis, and the periodic component signalWhether or not the phase of the current has reached the specified phaseDetermination process (synchronization control unit 102 / modification (2)),When the determination process determines that the phase of the periodic component signal has reached a predetermined phase,The noise componentWaveform dataRead from the memoryNoise component signal generation that generates a noise component signal based on the read noise component waveform dataProcess (noise waveform generator 80);The generated noise component signal andThe periodic component signalTo generate a musical sound signalAnd a musical tone signal generation process (mixing and reproducing unit 112).
  Further, in the configuration according to claim 4, the claim2 or 3In the described musical sound signal synthesis method,The first memory (noise component memory 84) stores a plurality of types of noise component waveform data, and the noise component signal generation processThe plurality of types of noise componentsWaveform dataAny one of these is randomly selected and read.
  Furthermore, in the configuration according to claim 5,2 or 3In the musical sound signal synthesis method described above,Noise component signal generationNoise component read out in the processWaveform dataOn the other hand, the method further includes a step of performing filter processing based on the filter coefficient (all-pass filter 85) and a step of randomly setting the filter coefficient (coefficient generator 87).
  Furthermore, in the configuration according to claim 6, in the musical sound signal synthesis method according to any one of claims 2 to 5,The first memory (noise component memory 84) stores the noise component waveform data having one cycle length of the cycle component.
  Further, in the configuration according to claim 7, claims 1 to6The method according to any one of the above is executed.
  Moreover, in the structure of Claim 8, it is Claim 1 thru | or.6A program for executing the method described in any one of the above is stored.
[0006]
DETAILED DESCRIPTION OF THE INVENTION
1. First embodiment
1.1. Principle of the embodiment
(1) Noise components included in the musical sound waveform of natural musical instruments
Here, the generation mechanism of noise components in various natural musical instruments will be described.
First, when a saxophone is played, the reed hits the mouthpiece due to the vibration of the resonance tube, so that a noise called “beeps” is generated. This noise occurs once (or several times) in one cycle of lead movement. Thus, it can be seen that this noise is generated according to the period of the movement of the lead, that is, the period of the waveform.
[0007]
Also, when the flute is played, turbulence is generated when the breath blown by the performer hits the singing mouth, and edge noise is generated. Since this is also related to the vibration of the flute itself, there is a relationship between the period of the waveform and the noise. Further, in a bowed instrument such as a violin, noise hardly occurs when a bow sticks to a string, and a large noise occurs when slipping. This noise is also noise corresponding to the slip and stick cycle, that is, the waveform cycle.
[0008]
(2) Details of noise components
When the musical sound waveform of a natural musical instrument is subjected to FFT (Fast Fourier Transform) analysis, the frequency component of the musical sound waveform can be separated into a frequency component that is continuous on the time axis and a frequency component that is intermittent on the time axis. When a waveform is synthesized based on the former frequency component, a “sound component (beat sound component)” of a musical sound waveform is obtained when a waveform is synthesized based on the latter frequency component.
[0009]
An example is shown in FIG. FIG. 4A shows a musical sound waveform (original waveform) of a saxophone. FIG. 4B shows the periodic component, and FIG. 4C shows the noise component. And the figure which expanded Fig.1 (a)-(c) on the time-axis is shown to Fig.2 (a)-(c). When the pitch of the periodic component is obtained in FIG. 2 (b) and the noise component is divided for each pitch period, noise components (beat sound components) N1 to N4 are obtained in FIG. 2 (c).
[0010]
Waveforms obtained by further expanding these noise components N1 to N4 on the amplitude axis and the time axis are shown in FIGS. 3A to 3D, the horizontal axis indicates the number of samples in each period. Comparing these figures, it can be seen that there is a peak of the noise component around 170 to 190 samples, and then the noise component decreases.
[0011]
1.2. Hardware configuration of the embodiment
Next, the hardware configuration of the waveform analysis / synthesis system according to the first embodiment of the present invention will be described with reference to FIG. Note that the hardware of this embodiment is configured by a general-purpose personal computer.
[0012]
In the figure, reference numeral 18 denotes a MIDI_I / O board which exchanges MIDI signals with external MIDI devices. A CPU 20 controls each unit in the waveform analysis / synthesis system via the bus 16 based on a control program described later. A ROM 22 stores an initial program loader and the like. A RAM 24 is used as a work memory for the CPU 20. A hard disk 26 stores an operating system, an application program for a waveform analysis / synthesis system, waveform data, and other various data.
[0013]
A network I / O board 2 exchanges various information with an external network. Reference numeral 4 denotes a display that displays various types of information to the user under the control of the CPU 20. Reference numeral 6 denotes an operator, which includes a keyboard, a mouse, a keyboard, and the like, and inputs various types of information to the CPU 20. Reference numeral 8 denotes an Fs (sampling frequency) generator, which generates a sampling clock having a frequency designated by the CPU 20.
[0014]
Reference numeral 10 denotes a sound I / O board which samples an analog musical sound waveform input from the outside in synchronization with the sampling clock and converts it into a digital musical sound waveform, while converting the supplied digital waveform data into an analog signal. Then output. This analog signal is generated via the sound system 12. A DMA controller 14 transfers digital waveform data between the sound I / O board 10 and the RAM 24 without using the CPU 20.
[0015]
1.3. Operation of the embodiment
1.3.1. Waveform data creation process
When the operating system is started up on the personal computer and the application program of the waveform analysis / synthesis system is started up, and the user performs a predetermined operation, the waveform data creation process is executed. Details of the processing will be described with reference to FIG. FIG. 2 is a functional block diagram showing the contents of processing (program) executed in the CPU 20.
[0016]
In the figure, reference numeral 32 denotes a waveform recording unit, which stores waveform data 30 input from the outside via the sound I / O board 10 in the RAM 24. Reference numeral 34 denotes an FFT analysis processing unit, which performs FFT analysis processing on the waveform data 30. Here, first, a window function having a length of 8 times the pitch period is applied to the original waveform data, and a frequency component within the range of the window function is analyzed.
[0017]
Next, the position of the window function is shifted backward by 1/8 of the pitch period on the time axis, and the frequency component is similarly analyzed. When this process is repeated for the entire original waveform data, a change in frequency component on the time axis is obtained. The series of frequency components obtained in this way is classified into components that are continuous on the time axis (deterministic frequency components) and other components that are not cut off.
[0018]
A continuous component separation unit 36 performs such classification according to the continuity of each analyzed frequency component. A waveform synthesis unit 44 outputs deterministic waveform data (that is, periodic component waveform data in FIGS. 1B and 2B) based only on the deterministic frequency component among these frequency components. Next, 46 is a subtracting unit that subtracts deterministic waveform data from the original waveform data. This subtraction result corresponds to the noise component in FIGS. 1 (c) and 2 (c).
[0019]
Reference numeral 48 denotes a noise waveform generation unit, which extracts a subtraction result of the subtraction unit 46, that is, a noise component, every length that is an integral multiple of the period of deterministic waveform data, and stores the result in the RAM 24 as noise component waveform data 50. . Note that the length and number of the noise component waveform data 50 differ depending on the contents of the waveform synthesis processing described later.
[0020]
Reference numeral 40 denotes a pitch detection unit that detects the pitch of the original waveform data based on the analysis result of the FFT analysis processing unit 34. Next, reference numeral 42 denotes an address setting unit that determines various addresses for periodic component waveform data based on the detected pitch. The address determined here is, for example, an attack start address indicating the beginning of the attack portion read only once at the beginning of waveform reproduction, and a loop start address and a loop end address indicating the beginning and end of the loop portion that are repeatedly read thereafter. Etc.
[0021]
Next, reference numeral 38 denotes a waveform creation unit which creates waveform data of the attack unit and the loop unit based on the deterministic frequency component and each address determined by the address setting unit 42. The created waveform data is stored in the RAM 24 as periodic component waveform data 52. The waveform synthesis unit 44 and the waveform creation unit 38 both create waveform data based on deterministic frequency components, but the waveform creation unit 38 creates waveform data composed of an attack unit and a loop unit. The waveform data is slightly different from the waveform data created by the waveform synthesis unit 44.
[0022]
That is, the waveform data of the loop part can contain only frequency components (referred to as loop harmonic components) that are integral multiples of the loop repetition period. Therefore, the waveform creation unit 38 creates waveform data of the attack part and the loop part by the following method. First, the waveform data of the attack part is processed so that the frequency component not corresponding to the loop harmonic component in the deterministic frequency component is gradually faded out from the middle of the attack part to the end of the attack part. Created by the deterministic frequency component performed. The waveform data of the loop part is created by selectively using only the frequency component corresponding to the loop harmonic component in the deterministic frequency component. The waveform data of the attack section and the loop section created in this way is a tone that is very similar to the deterministic waveform data created by the waveform synthesis section 44, and has a good connection from the loop end to the loop start. It is waveform data.
[0023]
1.3.2. Waveform synthesis processing
After the waveform data is created, when a MIDI event is input via the operator 6 or the MIDI_I / O board 18, a musical sound waveform is synthesized based on the MIDI event. Also, when playing back an SMF (standard MIDI format) file supplied via the network I / O board 2 or the like, a musical sound waveform is synthesized based on the event information. A functional block diagram of this waveform synthesis processing is shown in FIG.
[0024]
In the figure, reference numeral 60 denotes a periodic waveform generating unit, which is read from the periodic component memory 64 storing the periodic component waveform data 52, the address generating unit 62 for generating a read address of the periodic component memory 64, and the address generating unit 62. The interpolation unit 66 performs interpolation processing on the waveform data.
[0025]
The address generation unit 62 starts reading the periodic component memory 64 from the attack start address. When the read address reaches the loop end address, the address generation unit 62 returns the read address to the loop start address, and thereafter, between the loop start address and the loop end address. Is read repeatedly. The reading speed is controlled by the pitch, that is, the F number.
[0026]
The length from the loop start address to the loop end address is about one to several cycles (short loop). Generally, since the musical sound waveform differs for each sound range, the periodic component memory 64 stores different periodic component waveform data 52 for each sound range (for example, a low sound range, a middle sound range, and a high sound range). The waveform data to be reproduced is determined according to the F number.
[0027]
Reference numeral 80 denotes a noise waveform generator, a noise component memory 84 storing the noise component waveform data 50 as in the periodic waveform generator 60, an address generator 82 for generating a read address of the noise component memory 84, and address generation The interpolation unit 86 is configured to perform interpolation processing on the waveform data read from the unit 82.
[0028]
When a predetermined trigger signal is supplied, the address generator 82 reads the waveform data between the start address and the end address only once. Thereafter, the process waits until the trigger signal is input, and the waveform data between the start address and the end address is read once each time the trigger signal is supplied.
[0029]
Reference numeral 102 denotes a synchronization control unit that monitors an address generated by the address generation unit 62, and when this becomes a predetermined trigger value (in other words, when the periodic component waveform has a predetermined phase), the trigger signal is sent to the address generation unit. 82. As a result, the noise component waveform data is generated in synchronization with the periodic component waveform data.
[0030]
Incidentally, examples of waveforms per cycle of the noise component waveform data 50 are as shown in FIGS. 3A to 3D. As described above, the noise component waveform has peaks around 170 to 190 samples. Exists. In this case, it is preferable to shift the phase in advance so that the peak portion is positioned at the head of the noise component waveform data.
[0031]
As a result, a common recognition that “the peak of the noise component waveform data is present” is formed, so that the relationship between the “trigger value” set in the synchronization control unit 102 and the noise component to be given can be intuitively understood. Can be grasped. FIG. 11 shows the relationship between the noise component waveform data, the phase component waveform data, and the trigger signal whose phase is set so that the peak portion is positioned at the head.
[0032]
A random number generator 114 generates a random number used in the address generator 82. Here, the length from the start address to the end address of the noise component waveform data 50 is secured about several tens of times the pitch. This is because the noise component sounds like a buzzer when the loop size is short as described in the prior art, and sounds like jitter when the loop size is slightly increased.
[0033]
In the present embodiment, the noise component waveform data 50 is stored for each one-to-one sound range with respect to the sound range of the periodic component waveform data 52 (for example, a low sound range, a middle sound range, and a high sound range). Then, in accordance with the F number, noise component waveform data 50 corresponding to each sound range is read. Reference numerals 104 and 108 denote filter units, which respectively perform filtering processing on the generated periodic component signal and noise component signal.
[0034]
Reference numerals 106 and 110 denote level control units for adjusting the level of the signal after the filtering process. Reference numeral 112 denotes a mixing and reproducing unit that mixes both level-adjusted signals and outputs them as musical tone signals. As a result, the musical tone signal is generated via the sound I / O board 10 and the sound system 12. In the present embodiment, when the periodic component waveform data has a predetermined phase, by triggering noise waveform data sufficiently longer than the periodic component waveform, the periodic component waveform is synchronized with the phase, and Natural noise with no periodicity like a buzzer can be added.
[0035]
2. Second embodiment
Next, a second embodiment of the present invention will be described. The hardware configuration and software configuration of the second embodiment are the same as those of the first embodiment.
However, as the noise component waveform data 50, a plurality of types (about 4 to 6 types per sound range) are stored for each sound range, and the length of one noise component waveform data 50 is one cycle. Each time the trigger signal is supplied from the synchronization control unit 102, the address generation unit 82 determines the sound range based on the F number at that time.
[0036]
Then, based on the output of the random number generator 114, any one of the noise component waveform data 50 in the sound range is randomly selected and reproduced. In the present embodiment, the required memory capacity of the noise component memory 84 can be significantly reduced as compared with the first embodiment, but the noise component is selected as a buzzer or jitter because a plurality of noise component waveform data 50 are selected at random. Can be prevented, and a natural noise component can be added to the musical sound signal. That is, in the present embodiment, when the periodic component waveform has a predetermined phase, the noise component waveform data selected at random from the plurality of noise component waveform data is read out. Even if the waveform data is as short as the cycle of the cycle component, as in the first embodiment, the cycle component waveform is synchronized with the phase of the cycle component waveform, and there is no periodic sense like a buzzer. Noise can be added.
[0037]
3. Third embodiment
Next, a third embodiment of the present invention will be described. The hardware configuration and software configuration of the third embodiment are the same as those of the second embodiment.
However, the noise component waveform data 50 is further classified into a plurality of groups for each volume (or velocity) range. That is, when the sound range is divided into 3 groups (low range, mid range, high range) and the volume is classified into 3 groups (weak, medium, strong), the noise component waveform data 50 is classified into 9 groups as a whole. Will be. Further, the periodic component waveform data 52 is also stored in the types corresponding to the sound range and volume.
[0038]
Then, about 4 to 6 types of noise component waveform data 50 are stored for each group. Also in this embodiment, the length of one noise component waveform data 50 is one cycle. Each time the trigger signal is supplied from the synchronization control unit 102, the address generation unit 82 selects any group based on the F number and volume at that time.
[0039]
Then, based on the output of the random number generator 114, any one of the noise component waveform data 50 in the group is randomly selected and reproduced. Therefore, in the present embodiment, as in the second embodiment, even if the noise component waveform data is short waveform data that is as short as the period of the periodic component, it is synchronized with the phase of the periodic component waveform, and Natural noise with no periodicity like a buzzer can be added. Furthermore, in the present embodiment, since the noise component waveform data 50 is selected based on the difference in the sound range and volume, a more natural noise component that changes according to the sound range and volume of the periodic component waveform is used as the musical sound signal. It is possible to grant.
[0040]
4). Fourth embodiment
Next, a fourth embodiment of the present invention will be described. The hardware configuration and software configuration of the fourth embodiment are the same as those of the second embodiment.
However, in place of the noise waveform generator 80 in the second embodiment, a noise waveform generator 81 shown in FIG. 7 is provided. In FIG. 7, the noise component memory 84 stores noise component waveform data 50 of one cycle length, one for each sound range.
[0041]
When a trigger signal is supplied to the address generation unit 82, any one noise component waveform data 50 is read based on the F number at each time point. Reference numeral 85 denotes an all-pass filter that passes through the entire band of the read noise component waveform data 50, but varies the phase characteristic (delay amount) for each band based on the filter coefficient. A coefficient generator 87 randomly sets the filter coefficient based on the output of the random number generator 114.
[0042]
Therefore, even when the waveform of the noise component signal output through the all-pass filter 85 is based on the same noise component waveform data 50, it randomly changes based on the phase characteristics at that time. As a result, even in this embodiment, even if the noise component waveform data is waveform data that is as short as the period of the periodic component, it is synchronized with the phase of the periodic component waveform and has a periodic sense like a buzzer. Natural noise without noise can be added. As in the third embodiment, the waveform data may be switched according to the volume in addition to the sound range.
[0043]
5. Fifth embodiment
Next, a fifth embodiment of the present invention will be described. The hardware configuration and software configuration of the fifth embodiment are the same as those of the second embodiment. However, the contents of the waveform data creation process (FIG. 5) and the waveform synthesis process (FIG. 6) in the second embodiment are modified as shown in FIGS. 8 and 9, respectively.
[0044]
In FIG. 8, a pitch component phase extraction unit 54 monitors the pitch component of the deterministic frequency component and detects its phase. Here, the “phase of the pitch component” will be described in more detail. The pitch component detected by the FFT analysis processing unit 34 has a phase (or frequency) “fluctuation”.
[0045]
This “fluctuation” can be observed as a side lobe of the pitch component when the width of the window function used in the FFT analysis processing unit 34 is increased. However, in this embodiment, since the window function width is about 8 times the pitch period, when the FFT analysis result is observed with a spectrum analyzer or the like, for example, the pitch component fluctuates as the position of the window function is shifted. Can be observed.
[0046]
An example of the “fluctuation” of the phase of the pitch component is shown in FIG. The pitch component phase extraction unit 54 cuts out the characteristics shown in FIG. 10 for each pitch period T and creates several (4-6) pieces of pitch component phase data D1, D2,..., Dn.
[0047]
Next, in FIG. 9, reference numeral 83 denotes a noise waveform generation unit. Compared with the noise waveform generation unit 81 of the second embodiment, an address LUT (look-up table) 89 is provided between the address generation unit 82 and the noise component memory 84. Is different in that is inserted. Further, in the noise component memory 84, one cycle length of noise component waveform data 50 is stored, one for each sound range, as in the fourth embodiment.
[0048]
As in the fourth embodiment, when a trigger signal is supplied to the address generator 82, an address signal for reading any one of the noise component waveform data 50 is read from the address generator 82 based on the F number at each time point. Is output. In the address LUT 89, pitch component phase data D1, D2,..., Dn are stored in advance, and any one of them is randomly selected by the output of the random number generator 114.
[0049]
That is, in the address LUT 89, the pitch component phase data is formatted such that both “pitch period T” on the horizontal axis and “phase 2π” on the vertical axis correspond to “number of samples of pitch period”. Then, a fluctuation as shown in FIG. 10 is given to the linear (sawtooth waveform) address signal supplied from the address generator 82, and the result is stored in the noise component memory 84 as a final address signal. Will be supplied.
[0050]
As described above, in this embodiment, even when the same noise component waveform data is read out in the noise component memory 84, the pitch component phase data D1, D2,. Since it changes, the noise component signal changes randomly based on the current phase. As a result, even in this embodiment, even if the noise component waveform data is waveform data that is as short as the period of the periodic component, it is synchronized with the phase of the periodic component waveform and has a periodic sense like a buzzer. Natural noise without noise can be added. As in the third embodiment, the waveform data may be switched according to the volume in addition to the sound range.
[0051]
6). Modified example
The present invention is not limited to the above-described embodiment, and various modifications can be made as follows, for example.
(1) In the first and second embodiments, the noise component waveform data 50 and the periodic component waveform data 52 are provided for each sound region (for example, the low sound region, the middle sound region, and the high sound region). The noise component waveform data 50 may be common to the entire sound range by providing for each sound range. This is based on the fact that the difference in the sound range of the noise component waveform data 50 is relatively difficult to detect.
[0052]
(2) In the second embodiment and the like, the speed at which the noise component waveform data is read is arbitrary, and may be controlled independently of the pitch of the periodic component waveform.
(3) In the second embodiment and the like, the length of the noise component waveform data is one period of the periodic component waveform, but may be any length such as two periods, three periods, or half periods. However, since the noise is synchronized with the periodic component waveform, it is preferable that the length is related to the periodic length of the periodic component waveform.
[0053]
(4) In the second embodiment, a plurality of types of noise component waveform data are stored for each sound range, but some of the plurality of types of noise component waveform data for each sound range are shared between different sound ranges. It may be used. Similarly, in the third embodiment, plural types of noise component waveform data are stored for each group. However, some of the plural types of noise component waveform data for each group are shared between different groups. You may make it use for.
[0054]
(5) In the synchronization control unit 102 according to the second embodiment or the like, the trigger signal is generated at the timing when the address (phase of the periodic component waveform) of the address generation unit 62 becomes a predetermined trigger value. The timing may be slightly varied by adding a small random number. That is, in the waveform period of the periodic component waveform, the phase at which the trigger is generated may be shifted by, for example, about 0 to 20%. This is because there are fluctuations in the generation timing of noise in natural musical instruments.
[0055]
(6) In each of the above embodiments, the periodic component signal is constituted by the waveform memory sound source. However, it goes without saying that the periodic component signal may use various sound source methods such as an FM sound source and a physical model sound source. In each embodiment, the synchronization control unit 102 monitors the address generated by the periodic waveform generating address generation unit 62, and generates a trigger signal when this reaches a predetermined trigger value.
[0056]
However, the trigger for generating the trigger signal is not limited to this address value. For example, the trigger signal may be generated when the level or inclination of the periodic component signal is in a predetermined state. This is particularly suitable when the sound source method other than the waveform memory sound source is adopted as the periodic waveform generation unit 60.
[0057]
(7) In each of the above embodiments, the waveform analysis / synthesis system is realized by software operating on hardware. However, each component shown in FIGS. 4 to 9 is realized by hardware, and various electronic musical instruments and others. It may be used for amusement equipment. Further, the software used in the above embodiment can be stored in a recording medium such as a CD-ROM or a floppy disk and distributed, or can be distributed through a transmission path.
[0058]
【The invention's effect】
As described above, according to the present invention, a periodic component signal having a continuous frequency component on the time axis is generated, and a noise component is added to the periodic component signal. Therefore, a natural noise component can be added to the musical sound signal. Is possible.
[Brief description of the drawings]
FIG. 1 is a waveform diagram showing an analysis result of a saxophone musical sound waveform.
FIG. 2 is a waveform diagram obtained by enlarging FIG. 1 on a time axis.
FIG. 3 is a waveform diagram obtained by dividing the noise component of FIG. 2 (c) for each pitch period of the periodic component.
FIG. 4 is a hardware block diagram of a waveform analysis / synthesis system in each embodiment of the present invention.
FIG. 5 is a block diagram of waveform data creation processing in the first to fourth embodiments of the present invention.
FIG. 6 is a block diagram of waveform synthesis processing in the first embodiment.
FIG. 7 is a block diagram of a noise waveform generator 81 in the second embodiment.
FIG. 8 is a block diagram of waveform data creation processing in the fifth embodiment.
FIG. 9 is a block diagram of waveform synthesis processing in the fifth embodiment.
FIG. 10 is an operation explanatory diagram of a noise waveform generator 83 in the fifth embodiment.
FIG. 11 is a diagram illustrating a relationship among noise component waveform data, periodic component waveform data, and a trigger signal.
[Explanation of symbols]
20 ... CPU, 22 ... ROM, 24 ... RAM, 26 ... hard disk, 30 ... waveform data, 32 ... waveform recording unit, 34 ... FFT analysis processing unit, 36 ... continuous component separation unit, 38 ...... Waveform creation unit, 40 ... Pitch detection unit, 42 ... Address setting unit, 44 ... Waveform synthesis unit, 46 ... Subtraction unit, 48 ... Noise waveform creation unit, 50 ... Noise component waveform data, 52 ... Period component waveform data 54... Pitch component phase extractor 60. Period waveform generator 62 62 Address generator 64. Period component memory 66. Interpolator 80, 81, 83. ... Noise waveform generator, 82 ... Address generator, 84 ... Noise component memory, 85 ... All-pass filter, 86 ... Interpolator, 87 ... Coefficient generator, 102 ... Synchronization controller, 112 ... Mixing & Playback part, 1 4 ...... random number generator.

Claims (8)

Analyzing the frequency components of the original waveform;
A process of extracting a noise component which is a discontinuous frequency component on the time axis based on the analysis result;
Storing noise component waveform data in the first memory based on the noise component ;
A process of extracting continuous periodic components on the time axis based on the periodic components of the original waveform;
Storing periodic component waveform data in the second memory based on the extracted frequency component;
Generating a read address that changes at a rate depending on the desired pitch;
Periodically reading the periodic component waveform data from the second memory by the read address, and generating a periodic component signal based on the read periodic component waveform data ; and
A trigger signal generating process for generating a trigger signal upon detecting that the read address has reached a predetermined trigger value;
A noise component signal generation step of reading the noise component waveform data from the first memory in response to the generation of the trigger signal and generating a noise component signal based on the read noise component waveform data;
A method for analyzing and synthesizing a musical sound signal, comprising: a synthesis process for synthesizing a generated periodic component signal and a noise component signal. A first memory that stores noise component waveform data of noise components that are discontinuous frequency components on the time axis, and a second memory that stores periodic component waveform data of periodic components that are continuous frequency components on the time axis In a musical sound signal synthesis method for synthesizing a musical sound signal by reading
The process of generating a read address that changes at a speed according to the pitch,
Periodically reading the periodic component waveform data from the second memory by the read address, and generating a periodic component signal based on the read periodic component waveform data; and
A trigger signal generating process for generating a trigger signal upon detecting that the read address has reached a predetermined trigger value;
A noise component signal generation step of reading the noise component waveform data from the first memory in response to the generation of the trigger signal and generating a noise component signal based on the read noise component waveform data;
A method for synthesizing a musical sound signal, comprising: a synthesis process for synthesizing the generated periodic component signal and noise component signal . In a musical sound signal synthesis method for synthesizing a musical sound signal by reading a first memory that stores noise component waveform data of a noise component that is a discontinuous frequency component on the time axis,
A periodic component signal generation process for generating a periodic component signal having a continuous frequency component on the time axis;
A determination process for determining whether or not the phase of the periodic component signal has reached a predetermined phase ;
When the determination process determines that the phase of the periodic component signal becomes a predetermined phase, the noise component waveform data read out from said memory, the read noise to generate a noise component signal based on the noise component waveform data Component signal generation process;
A musical sound signal synthesizing method comprising: generating a musical tone signal by synthesizing the generated noise component signal and the periodic component signal. The first memory stores a plurality of types of noise component waveform data,
4. The musical tone signal synthesis method according to claim 2, wherein, in the noise component signal generation process , any one of the plurality of types of noise component waveform data is randomly selected and read out. With respect to the noise component signal generated noise component waveform data read in step, the steps of performing a filtering processing based on the filter coefficients,
The method for synthesizing a musical sound signal according to claim 2 or 3 , further comprising the step of randomly setting the filter coefficient. 6. The musical sound signal synthesis method according to claim 2, wherein the first memory stores the noise component waveform data having one cycle length of the cycle component. Tone signal synthesizing apparatus characterized by performing a method according to any one of claims 1 to 6. Recording medium characterized by storing a program for executing the method according to any of claims 1 to 6.

Images