JP2001100756A - Method for waveform editing - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for waveform editing which automatically matches the timing of an object waveform and that of a reference waveform with each other so that correlation between the object waveform and the reference waveform may be high. SOLUTION: Original waveform data is prepared in a step S21. In a step S22, original waveform data is analyzed and processed and sound source waveform data is synthesized and is written in a waveform memory again. A range of high correlation is roughly set on the basis of levels of envelopes of the object waveform and the reference waveform, and a degree of correlation, for example, a correlation function is calculated in this range to perform accurate timing matching with high correlation. The object waveform is shifted relatively to the reference waveform by a time difference which gives a maximum value of the degree of correlation. In a step S23, sound source waveform data is used to give a performance.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a waveform editing method for adjusting timings of a plurality of waveforms so as to increase the correlation between the waveforms. Especially in electronic musical instruments,
This is suitable for waveform editing when a plurality of related waveforms are mixed to form a sound source waveform.

[0002]

2. Description of the Related Art In a waveform memory sound source of an electronic musical instrument,
In some cases, two waveforms having the same generation source are mixed to form a sound source waveform. However, if the timing (position or phase on the time axis) of the two waveforms is shifted, noise may occur in the mixed waveform or the waveform may be distorted. Therefore, it is necessary to adjust the timing of the two waveforms so that the correlation between the waveforms is high. Conventionally, there has been no method for automatically performing timing adjustment (phase adjustment) for increasing the correlation between two waveforms having the same generation source. The user operates the operator while watching the two waveforms displayed on the display, and moves the waveforms little by little to adjust the timing of the two waveforms.

On the other hand, there is a time axis compression / expansion technique for compressing / expanding the time axis of a waveform by partially thinning out or repeating the waveform. In this case, in order to smooth the connection of the waveforms before and after thinning or the connection of the repeated parts of the waveform, the cross-correlation (Cross C
orrelation) function, and the connection was made at the part where the cross-correlation function value was high. In this case, the cross-correlation value could be calculated because the position to be connected was determined.

On the other hand, in the timing adjustment of two waveforms to be mixed as described above, a comparison range in which a cross-correlation function is to be calculated is not specified in advance. For example, it is inefficient to calculate the cross-correlation function of the entire waveform from the rising edge to the falling edge of the musical tone to match the timing of the two waveforms at the rising portion (attack portion) of the musical tone. Further, it is not always possible to perform effective timing adjustment for enhancing the correlation of the waveform of the rising portion of the musical sound.

[0005]

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and is intended to adjust the timing of a target waveform and a reference waveform so as to increase the correlation between the target waveform and the reference waveform. It is an object of the present invention to provide a waveform editing method for automatically performing the above.

[0006]

According to the present invention, the timing of the target waveform and the reference waveform are adjusted so that the correlation between the target waveform and the reference waveform is increased. A waveform editing method, comprising: detecting an envelope of a certain waveform parameter for the target waveform and the reference waveform; and detecting the target waveform and the reference waveform based on the envelopes of the target waveform and the reference waveform. Setting a comparison range, calculating a correlation between the target waveform and the reference waveform in the comparison range, and calculating a time difference between the target waveform and the reference waveform when the correlation is maximized. And shifting the target waveform relative to the reference waveform by the time difference. Therefore, the timing of both waveforms can be automatically adjusted so that the correlation between the target waveform and the reference waveform is high. At that time, the comparison range is set based on the envelope of the waveform parameter, which reflects the characteristics of the waveform well, so that the comparison range estimated to have high correlation can be set in advance, so that it is accurate and The degree of correlation can be calculated with a small amount of processing data.

According to a second aspect of the present invention, there is provided a waveform editing method for adjusting the timing of the target waveform and the reference waveform so that the correlation between the target waveform and the reference waveform is high. Regarding the reference waveform,
Detecting an envelope of a certain waveform parameter; setting a comparison range for the target waveform and the reference waveform with reference to the envelopes of the target waveform and the reference waveform; The waveform and the reference waveform, and the polarity inversion waveform and the reference waveform, the polarity of the target waveform and the reference waveform, calculate the degree of correlation, when the correlation degree is maximized,
Calculating a time difference between the target waveform or the polarity-reversed waveform and the reference waveform, and when the correlation between the target waveform and the reference waveform has the maximum correlation, the target waveform is While relatively shifting the reference waveform by the time difference, and when the degree of correlation between the polarity-reversed waveform and the reference waveform maximizes the degree of correlation, the target waveform is replaced with the polarity-reversed waveform. , A step of performing a process of relatively shifting the reference waveform by the time difference. Therefore, in addition to the same operation and effect as the first aspect, even if the polarity of the target waveform and the reference waveform is opposite, the timing can be adjusted after detecting the polarity inversion.

[0008]

FIG. 1 is a block diagram of a waveform editing apparatus for executing a waveform editing method according to the present invention. In the figure, reference numeral 1 denotes a CPU (Central Processes) for controlling the entire waveform editing apparatus.
sing unit), 2 is a ROM for storing various programs such as control programs and various control information, 3 is a RAM used as a work area or a buffer area or an area for storing various programs, 4 is a clocking operation or C
A timer for interrupting the PU 1 with a timer, 5 is a panel switch provided with various operation switches, and 6 is a panel display for displaying various types of processing target waveforms and the like. 7 is an external MIDI (Musical Instrument Digital I
nterface) MIDI interface for exchanging MIDI events with devices, 8 is a CD-ROM
(Compact Disk-Read Only Memory), a drive device for accessing a recording medium 9 such as an HD (hard magnetic disk) and an FD (flexible magnetic disk).

A waveform memory 10 stores input waveform data and generated waveform data, and is capable of writing and reading a plurality of waveform data. An access management unit 11 manages access time slots of the waveform memory 10 so that accesses from the writing circuit 13, the buffer 14, or the tone generator 15 to the waveform memory 10 do not collide with each other. 12 is an external waveform input terminal, 13
Is a writing circuit which samples a target waveform inputted from the external waveform input terminal 12 and writes it in the waveform memory 10. Reference numeral 14 denotes a buffer, which is a recording medium 9
3 is transferred to the waveform memory 10 or the waveform data read from the waveform memory 10 is transferred. Reference numeral 15 denotes a tone generator, which generates a tone signal using the waveform data read from the waveform memory 10. 1
Reference numeral 6 denotes a sound system, which outputs a tone signal output from the sound source unit 15. 17 is a bus line,
Used for exchanging information between the above-described elements.

Here, ROM2 or RAM3 contains:
In addition to a program that performs general control, it analyzes a target waveform that is input externally and edits the waveform to create a sound source waveform. Performance processing program. CPU1 is ROM2 or R
In accordance with various control programs stored in the AM 3, various controls are performed according to inputs from the panel switch 5 and the MIDI interface 7. In addition, during the waveform editing processing and the sound source waveform creation processing, the data in the waveform memory 10 is read / written via the buffer 14, the waveform data is read out, analyzed and processed, and written again in the waveform memory 10 or the recording medium. 9 and the waveform data supplied from a communication network (not shown) is written into the waveform memory 10, and conversely, the waveform data is supplied from the waveform memory 10 to the recording medium 9 and the communication network.

Further, the CPU 1 controls the tone generation state of the tone generation channel of the tone generator 15 according to the performance information supplied from the MIDI interface 7, the recording medium 9 or the RAM 3 when performing the performance processing. For example, when a note-on signal indicating the start of sound generation is input from the MIDI interface 7, the generation of the musical tone is assigned to one of the sound channels of the sound source unit 15, and the tone parameters (pitch information, pitch information,
Waveform selection information, volume envelope control information, effect information, etc.) and an instruction to start sounding. In response to this, the tone generator 15 uses the assigned sounding channel to generate a tone signal corresponding to the tone parameter described above using the tone waveform data read from the waveform memory 10 in accordance with the waveform selection information. Generate.

Although not shown, a communication interface circuit for connecting to a communication network such as a LAN or the Internet may be provided, and waveform data and various programs may be downloaded from a server via the communication network. Furthermore, a keyboard operator can be provided, and the user can perform using this keyboard operator. In the above description, the waveform editing method and the performance processing method
1 may be supplied from a recording medium 9 such as a CD-ROM, and may be installed and executed on the recording medium 9 such as a hard magnetic disk. Further, the program may be downloaded from a server on the network to the recording medium 9 and executed.

FIG. 2 shows a waveform analysis using the waveform editing apparatus shown in FIG. 1, processing of the waveform using the analysis result to create a sound source waveform, and a process until a person (operator) plays. It is a flowchart showing an outline of the processing. In S21,
The original original waveform data is prepared. In response to the recording switch operation of the panel switch 5 shown in FIG.
And is sampled and written to the waveform memory 10 via the access management unit 11. Alternatively, the original waveform data stored in the recording medium 9 is read by the driving device 8 and recorded in the waveform memory 10 via the buffer 14 and the access management unit 11.

In the next step S22, sound source waveform data is created from the original waveform data in the waveform memory 10. That is,
According to the operation of the edit switch of the panel switch 5, the original waveform data is analyzed and processed, and the sound source waveform data is synthesized.
The data is written into the waveform memory 10 again. In S23,
Perform using the sound source waveform data. That is, the sound source waveform data is selected in accordance with the performance switch operation of the panel switch 5. Next, using sound source waveform data, MIDI
The performance is performed based on performance data such as MIDI data input from the interface 7, the recording medium 9, or a communication network (not shown).

According to the embodiment of the waveform editing method of the present invention, in the above-mentioned S22, based on the envelope levels of the two waveforms, a roughly correlated range is set in a desired waveform portion, and this range is set. By calculating the degree of correlation in (1), timing adjustment with high correlation is accurately performed. One of the two waveforms will be described as a target waveform (target), and the other will be described as a reference waveform (reference).

The outline of the processing steps is as follows as an example. First, a comparison range in which it is estimated that the correlation is high based on the envelope levels of the reference waveform and the target waveform in the partial waveform of the musical sound waveform whose timing is to be adjusted, for example, in the rising portion (attack portion) of the musical sound waveform. To determine. Second, in the comparison range, the degree of correlation between the target waveform and the reference waveform is calculated using, for example, a cross-correlation function. Third, the degree of correlation between the reference waveform and a waveform obtained by inverting the polarity of the target waveform is calculated. Fourth, second and third
The maximum value of both correlation degrees calculated in the step is searched. Fifth, the maximum values of the correlations are compared, and the waveform having the highest correlation is adopted from the target waveform and the inverted polarity waveform.

Sixth, a process of shifting the target waveform relative to the reference waveform by a time difference (phase difference) when the adopted target waveform or polarity-inverted waveform gives the maximum value of the correlation degree. I do. Here, the process of relatively shifting is, for example, adding the above-described time difference to the start position of the comparison range of the target waveform as a new start position, setting this as a cutout start position, and storing the target waveform in a waveform memory. It is to do again. Alternatively, when the target waveform and the reference waveform are simultaneously read and the waveform data is used, read control information for delaying the read start timing of the target waveform by the above-described time difference (advancing when the time difference is negative) is stored in the memory. Means to memorize it.

By the above-described relative shift processing, the timing can be automatically adjusted so that the correlation between the two waveforms is increased. At this time, since a comparison range in which the degree of correlation is estimated to be high is set based on the envelope level, effective timing adjustment can be performed. Hereinafter, the operation of this embodiment will be specifically described with reference to flowcharts and waveform diagrams.

FIG. 3 is a first flowchart for explaining the operation of the embodiment of the waveform editing method according to the present invention. A comparison range for calculating a cross-correlation function value is set. FIG. 4 is a second flowchart for explaining the operation of one embodiment of the waveform editing method of the present invention. The time difference at which the cross-correlation function value becomes the maximum is calculated, and the timing of the target waveform and the reference waveform is adjusted so that the cross-correlation function value of the target waveform and the reference waveform becomes the highest. 5 to 10
FIG. 7 is an explanatory diagram showing a target waveform, a reference waveform, an envelope, and a cross-correlation function value at each stage during a waveform editing process. These are displayed on the screen of the panel display 6 in FIG. 1 during the editing process.

FIG. 5 is a waveform diagram of the target waveform A and the reference waveform B. FIG. 5A is a waveform diagram of a target waveform, and FIG. 5B is a waveform diagram of a reference waveform. The horizontal axis corresponds to the time axis and is represented by a sample number. The vertical axis is the amplitude. Description will be made from the flowchart of FIG. In S31, the respective volume envelopes of the target waveform A and the reference waveform B whose timing is to be matched are detected. In the next step S32, based on the level of each volume envelope, a partial waveform whose timing is to be matched, in this specific example, an attack portion is detected.

FIG. 6 is a diagram showing the volume envelopes of the target waveform A and the reference waveform B. FIG. 6A is a diagram illustrating the volume envelope (part) of the target waveform, and FIG. 6B is a diagram illustrating the volume envelope (portion) of the reference waveform. The horizontal axis is represented by the sample number. The vertical axis is the volume envelope. The volume envelope is detected by, for example, performing peak hold with a relatively short time constant on the input waveform. Here, the maximum value of the volume envelope from the rise to the end of the musical tone waveform is 100.
Is normalized so that

The attack portion is usually considered to be a period from the head position of the musical tone waveform (which is also the head position of the volume envelope) to the time when the musical tone waveform takes the maximum value (the volume envelope also takes the maximum value). . However, according to this definition, the time length of the attack portion of the target waveform A does not always match the time length of the attack portion of the reference waveform B. However, the calculation of the degree of correlation is performed in the comparison range having the same time length. In addition, since the vicinity of the leading position of the waveform has uncertainty due to the influence of noise or the like, if the timings of the leading position of the target waveform A and the leading position of the reference waveform B are aligned, the correlation is not necessarily high. .

In the next step S33, a comparison range is set for each of the target waveform A and the reference waveform B. Since the tone waveform is simply reflected in the volume envelope, a comparison range in which the correlation between the target waveform A and the reference waveform B is estimated to be high is set based on the volume envelope. The start position of the comparison range on the reference waveform B side is the head position itself. The start position of the comparison range on the target waveform side A is the target waveform A
And the level of the volume envelope of the reference waveform B.

First, a reference position at which the level of the volume envelope of the reference waveform B shown in FIG. 6B exceeds a certain threshold value (for example, 25% of the maximum value) for the maximum value is detected. Time length from the start position to the reference position of the reference waveform B to the t _0. On the other hand, a reference position where the level of the volume envelope of the target waveform A shown in FIG. 6A exceeds the same threshold value (25% of the maximum value) for the maximum value is similarly detected. The timing which is traced back from the reference position to the beginning of the waveform by the above-described time length t ₀ is set as a temporary start position, and this is set as the start position of the comparison range on the target waveform A side.

The above-mentioned time length t ₀ depends on the target waveform A
May be defined as the time length from the reference position where the volume envelope of the target waveform exceeds the above-described threshold value (25% of the maximum value) to the actual start position of the target waveform A. In this case, a temporary head position is set on the reference waveform B side using the time length t ₀ . Alternatively, the time length t described above
₀ may be set to a constant value regardless of the actual start positions of the reference waveform B and the target waveform A.

By the way, the time length (T _B ) of the attack portion from the top position of the reference waveform B until the volume envelope becomes maximum, and the case where the volume envelope of the target waveform from the above-mentioned provisional top position of the target waveform A is maximum. The time length of the attack portion (T _A ) until the attack portion does not always match. Therefore, the shorter of the time length (T _A ) and the time length (T _B ),
Alternatively, a predetermined time shorter than this is set as the time length of the comparison range.

Since the comparison range of the two waveforms is determined using the time point of the predetermined level determined by the maximum value of the envelope as the reference position, the reference waveform B and the target waveform A have a rough but highly correlated comparison range. Can be set. further,
Because the time length from the reference position to the beginning is fixed,
A highly correlated comparison range in which the influence of uncertainty near the head is reduced can be set. Therefore, in the strict calculation of the degree of correlation performed later, it is possible to reduce the calculation error of the time difference at which the degree of correlation between the reference waveform B and the target waveform A is maximized.

The setting of the comparison range is not limited to the above example. A range in which the volume envelopes of the reference waveform B and the target waveform A go back from the positions where the volume envelopes are maximum and have a predetermined time length can be set as a comparison range of the reference waveform B and the target waveform A. This predetermined time length is more preferably set near each of the head positions of the reference waveform B and the target waveform A. Also, a reference position where the volume envelope of the reference waveform B exceeds a certain threshold value (for example, 25% of the maximum value) with respect to the maximum value is set as the start position of the comparison range on the reference waveform side, and the volume envelope of the target waveform is similarly set. Then, a reference position exceeding the same threshold value (25% of the maximum value) with respect to the maximum value is set as the start position of the comparison range on the target waveform side, and in each case, the time length of the comparison range is fixed. It is more preferable that the predetermined time length be close to the maximum value of each volume envelope of the reference waveform B and the target waveform C.

FIG. 7 is an enlarged waveform diagram of the target waveform A and the reference waveform B. FIG. 6A is a waveform diagram showing a target waveform A (part), and FIG. 6B is a waveform diagram showing a reference waveform B (part). In the figure, the horizontal axis is the sample number, and the vertical axis is the amplitude. The timing of the reference waveform B and the target waveform A is adjusted by calculating the degree of correlation. In this specific example, the calculation of the degree of correlation between the reference waveform B and the target waveform A is performed by FFT (Fast Fourier Tra
nsform) to calculate a cross-correlation function (strictly, a discrete cross-correlation function, but abbreviated to a cross-correlation function).

Here, a method of calculating a cross-correlation function using FFT will be described. If there is a waveform x (t) and y (t), the cross-correlation function c (Δt) is

(Equation 1) The waveforms x (t) and y (t) are assumed to be periodic functions having the same period, and the cross-correlation function c (Δt) is Fourier-transformed for each period. C (ω) = X (ω) ^* × Y (ω) where C (ω) is the Fourier transform of c (Δt) F (c (ω)), X
(ω) ^* is the complex conjugate of X (ω), X (ω) is the Fourier transform F (x (t)) of x (t), and Y (ω) is the Fourier transform F (y (t )). That is, the Fourier transform of the cross-correlation function c (Δt) is a value obtained by multiplying the complex conjugate of the Fourier transform of x (t) by the Fourier transform of y (t). By inversely transforming it, F ⁻¹ (C (ω)) = c (Δt) = F ⁻¹ (X (ω) ^* × Y (ω)) Therefore, a cross-correlation function can be obtained.

The above-mentioned waveform x and (t) and y (t), but sampling waveform x (n) and y (n) (t = n × t s: t s is the sampling period) in the case of, Delta] t = as [Delta] n × t _s, the cross-correlation function c ([Delta] n) is

(Equation 2) This, when FFT ^{_{processing, C m = X m * ×}} Y m where, C _m is the FFT c ([Delta] n), the complex conjugate of X _m ^* is X _m, X _m
The FFT of x (n), Y _m is the FFT of y (n). By inverting C _m ,

(Equation 3) Therefore, the value R _AB of the cross-correlation function can be calculated using the FFT process, where x (n) is the reference waveform and y (n) is the target waveform. However, the cross-correlation function is calculated assuming that the reference waveform and the target waveform are aperiodic waveforms, but are periodic functions having a comparison range of one cycle.

The timing adjustment will be described with reference to the flowchart of FIG. In S41, each comparison range between the target waveform A and the reference waveform B is subjected to FFT processing (FF
TA, FFTB). Next, in S42, a value R _AB of a cross-correlation function between the target waveform A and the reference waveform B is calculated based on the result of the FFT processing (FFTA, FFTB). In S43, the value at which the value R _AB of the cross-correlation function becomes maximum and the value of Δt at that time are detected. FIG. 8 is a diagram showing a cross-correlation function R _AB between the target waveform A and the reference waveform B. In the figure,
The horizontal axis is the time difference Δt expressed by the number of samples, and the vertical axis is the value (%) of the cross-correlation function R _AB . In the illustrated example, Δt
Takes a slightly positive value, the value of the cross-correlation function R _AB takes the maximum value.

In the next step S44, FFT processing (FFTC) is performed on the polarity-reversed waveform C obtained by inverting the polarity of the target waveform A in the same comparison range as the target waveform A. In S45,
FFT processing of reference waveform B (FFTB) and F of polarity inverted waveform C
Based on the result of the FT processing (FFTC), the cross-correlation function R _CB
Is calculated. FIG. 9 is a diagram showing a cross-correlation function R _CB between a polarity inverted waveform C obtained by inverting the polarity of the target waveform A and a reference waveform B. In the figure, the horizontal axis is the time difference represented by the number of samples, and the vertical axis is the value (%) of the cross-correlation function _RCB . Note that the horizontal axis is compressed by half compared to FIG. S46
In, the value at which the cross-correlation function value R _CB is the maximum and the value of Δt at that time are detected.

In S47, the maximum value of the cross-correlation function value R _AB obtained in S43 is compared with the maximum value of the cross-correlation function value R _CB obtained in S46, and the maximum value of the cross-correlation function is obtained. Is larger, the moving amount Δt of the target waveform A is determined according to the time difference Δt when the maximum value is obtained. In the example of FIG. 9, the maximum value of the cross-correlation function R _CB is smaller than the maximum value of the cross-correlation function R _AB shown in FIG. Therefore, the moving amount Δt of the target waveform A is determined according to the time difference Δt at which the value of the cross-correlation function R _AB becomes maximum.

Here, the reason for calculating the discrete cross-correlation function with the reference waveform B for the polarity inverted waveform C of the target waveform A will be described. The polarity of the target waveform A and the reference waveform B may be reversed (opposite phase) due to the output characteristics of the microphone, the setting of the cable connection when recording the target waveform A and the reference waveform B, and the connection characteristics of the connector. is there. Therefore, as described above, since the cross-correlation function is calculated for both the target waveform A and its polarity-inverted waveform, the timing can be automatically adjusted without any problem even when the phases are opposite.

In the next S48, in S47,
The larger the maximum value of the cross-correlation function is the cross-correlation function value R _CB
, The polarity determination waveform C obtained by inverting the polarity of the target waveform A is replaced with the target waveform A, and is set as the target waveform A again. In the next S49, the start address of the target waveform A is shifted from the tentative head position in accordance with the movement amount Δt determined in S47. Since the relative time difference between the target waveform A and the reference waveform B is significant, the start address of the target waveform A and / or the reference waveform B may be shifted while keeping the relative time difference at Δt.

FIG. 10 is a comparison diagram of the target waveform A and the reference waveform B after the start address of the target waveform A is shifted by the time difference Δt and the timing is adjusted. The upper part is a waveform diagram of the target waveform A after the timing adjustment, and the lower part is a waveform diagram of the reference waveform B. The horizontal axis represents time represented by the number of samples, and the vertical axis represents amplitude. In this specific example, by shifting the start address (cutout position) of the target waveform A, the target waveform A shifts all data from the start position of the waveform to the end position of the waveform. The shifted target waveform A and reference waveform B have the highest correlation in the comparison range of the attack portion.
If both waveforms are read from their respective start addresses (cutout positions) after the timing of the target waveform A and the reference waveform B are adjusted, both waveforms are reproduced simultaneously at the timing with the highest correlation in the comparison range in the attack section. Will be done. If the start address of the target waveform A is not moved, the above-mentioned time difference Δt is stored, and when both waveforms are read, the read start timing of both waveforms may be shifted according to the time difference Δt.

In the specific example described above, the cross-correlation function of the target waveform A and the reference waveform B is calculated by the FFT processing, but the cross-correlation function is multiplied by the waveform amplitude on the real time axis. May be calculated as originally defined. In the above description, the start timing has been shifted based on the value of the cross-correlation function so that the correlation between the target waveform A and the reference waveform B is high. Instead, the correlation between the target waveform A and the reference waveform B may be calculated by another method. For example, when the target waveform A is shifted on the time axis with respect to the reference waveform B, a time difference when the sum of the absolute values of the amplitude differences between the two waveforms in the comparison period becomes minimum is detected. Further, it is also possible to detect a time difference when the sum of the squares of the amplitude difference between the two waveforms in the comparison period becomes minimum. In these examples, the correlation degree is highest when the sum is minimum.

In this embodiment, a portion where timing is adjusted so as to increase the correlation is set in the attack portion of the waveform. However, depending on the purpose of waveform editing, the part that adjusts the timing is the continuation part, falling part,
It is also possible to set these mutually transient connection parts including the attack part. In any case, by setting a comparison range in which both waveforms are estimated to have generally high correlation, and then calculating the degree of correlation, the accuracy of timing adjustment is improved.

In the above description, the target waveform and the reference waveform are the entire period from the rise to the fall of the musical tone waveform. However, according to the purpose of the waveform editing, the timing can be adjusted only for a part of the musical tone waveform, for example, the waveform portion of the attack portion as the target waveform A and the reference waveform B. When storing a waveform, there is a case where it is desired to store a silent period before the rising of the waveform. In such a case, the time length of the silence period can be set in units of time, and after the timing of the reference waveform B and the target waveform A is adjusted, a silence period may be added to the leading portions of both waveforms. Alternatively, before the timing adjustment, it is also possible to add the timing to the target waveform A after adding to the head portion of the reference waveform B.

In the above description, the timing of the two waveforms is adjusted so as to increase the correlation between the two waveforms. In order to adjust the timing so as to increase the correlation for three or more waveforms, one is determined as a reference waveform B, and this reference waveform B is compared with the other target waveforms A by two.
The timing of the other target waveform A may be adjusted based on the reference waveform B. In the above description, a monaural waveform has been described. In the case of a stereo waveform, the target waveform A
And the reference waveform B, calculating the degree of correlation between the target waveform A and the reference waveform B with respect to one of the left and right channels, detecting the time difference at which the degree of correlation is the highest, and dividing the left and right channels of the target waveform A into the respective channels. It is only necessary to shift together by the time difference.

Next, a first application example of the above-described timing adjustment of the target waveform A and the reference waveform B will be described. FIG.
FIG. 1 is a flowchart illustrating a first example of generating a sound source waveform for touch control. In the first application example, the timing of the target waveform A and the reference waveform B is adjusted by focusing on the attack portion, and then the phases (phase angles) of the Fourier components (frequency components including non-overtones) of the target waveform A are adjusted. ),
The composite waveform A is created by matching the phase (phase angle) of the Fourier component of the same frequency of the reference waveform B. Then, the combined waveform A and the reference waveform B are mixed to form a sound source waveform in the entire period of the musical tone waveform.

In S51 of FIG. 11, the comparison range of the target waveform A and the reference waveform B is set, and the process proceeds to S52.
When the comparison range is the attack portion, the processing shown in FIG. 3 is executed. Next, in S52, the processing shown in FIG. 4 is executed in order to match the timings of the target waveform A and the reference waveform B so that the correlation between the two waveforms is high in the comparison range. In the next step S53, a Fourier component A obtained by subjecting the target waveform A to FFT processing over the entire period of the waveform is detected. For the main Fourier components,
Waveforms are synthesized by adding (sine wave addition synthesis) corresponding sine waves of a plurality of frequencies. This composite waveform is subtracted from the target waveform A, and the residual component A is detected. S
In the case where the cross-correlation function is calculated using the FFT in 52, the Fourier component A may be calculated in the entire period of the waveform in step S52, and this may be used.

In the next S54, as in S53,
For the entire period of the reference waveform B, the Fourier component B obtained by performing the FFT processing and the waveform synthesized from the Fourier component B are subtracted from the reference waveform B, and the residual component B is detected. In the step of S52, FF
When calculating the cross-correlation function using T, this S5
In the second step, the Fourier component B for the entire period of the waveform may be calculated and used. next,
In S55, a Fourier component A 'is created by replacing the phase of each frequency of the Fourier component A with the phase of the Fourier component B of the same frequency. Next, in S56, a composite waveform A is obtained from the Fourier component A ′ and the residual component A.
Are synthesized. This composite waveform A is different from the target waveform A. The reference waveform B is used as it is.

The composite waveform A and the reference waveform B are mixed according to a key touch to form a sound source waveform. The reproduction of the composite waveform A and the reference waveform B stored in the waveform memory is simultaneously started according to the note-on (sound generation instruction signal). At this time, the mixing ratio (weight coefficient) of the composite waveform A and the reference waveform B is controlled according to the value of the velocity (key touch data) included in the note-on information. In the case of a piano tone, it is preferable to use a target waveform A that is played with fortessimo (ff) and a reference waveform B that is played with mesoforte (mf).

When the synthesized waveform A is used, the timing is adjusted so that the correlation between the synthesized waveform A and the reference waveform B becomes high in the target attack portion. Occasionally, noise or distortion of the waveform is rare. In addition, since the phase spectrum of the composite waveform A is aligned with the phase spectrum of the reference waveform B, the possibility of disappearance or attenuation of a specific spectrum due to mixing is reduced. As a result, the timbre change due to the velocity is less uncomfortable.

In the above description, the mesoforte (mf)
Thus, the phase spectrum of the other synthesized waveform A is replaced so as to match the phase spectrum of the reference waveform B flipped. However, it is also possible to replace the phase spectrum of the other reference waveform B so as to match the phase spectrum of the target waveform A played as fortissimo (ff). However, in a normal performance, mesoforte (m
Since the mixing ratio of the reference waveform B played as f) becomes larger, the use of the reference waveform B as it is is less uncomfortable.

In the first application example described above, the mixing ratio is controlled according to the velocity of the key touch. Instead of this, one of the composite waveform A and the reference waveform B may be selectively reproduced according to the note-on (sound generation instruction signal). Which waveform is selected depends on the above-mentioned velocity. That is, when the velocity is larger than the predetermined value, the synthesized waveform A is selected, and when the velocity is smaller, the reference waveform B is selected. Since the timings of the composite waveform A and the reference waveform B are aligned so that the correlation between the composite waveform A and the reference waveform B is high, the waveform can be switched with less discomfort.

In the above description, two-stage waveforms of strong touch (ff) and middle touch (mf) are prepared, but three or more waveform data separated by three or more touches are mixed to generate a sound source waveform. May be created. According to the velocity included in the note-on information, the mixing ratio of three or more waveform data can be changed or one of the three or more waveform data can be selected.

Next, a second applied example of the timing adjustment of the target waveform A and the reference waveform B will be described. FIG.
2 is a flowchart illustrating a second example of creating a touch control sound source waveform. In this example, from the attack part to the introduction part of the sustain part, the partial waveforms A,
The partial waveform B of the reference waveform B is mixed and used according to the velocity value. Then, in the sustain part, using the partial waveform D,
In the introduction part of the continuation part, the mixed waveform of the partial waveforms A and B and the partial waveform D are used by cross-fading.

In S61 of FIG. 12, a partial waveform A of the "introduction portion of the attack portion to the sustain portion" is created from the target waveform A, and in the next S62, the "introduction portion of the attack portion to the sustain portion" is formed from the reference waveform B. Is created. Normally, the “introduction portion of the attack portion to the sustain portion” of the target waveform A and the reference waveform B may be cut out as it is, but the waveform may be processed. Next, in S63, a partial waveform D of "continuation portion to release portion" is created from the target waveform A and / or the reference waveform B. Normally, the reference waveform B “
Release section ”may be used. When the sustain portion of the sustain portion is repeatedly reproduced as a loop waveform, it is preferable to perform the waveform processing so that there is no rapid change in the frequency spectrum in the portion before and after the joint of the loop waveform.

In the next step S64, as shown in the flowchart of FIG. 3, a comparison range between the attack portion of the partial waveform A and the attack portion of the partial waveform B is set. In the next S65, the timings of the partial waveform A and the partial waveform B are matched as shown in the flow of FIG. Note that, similarly to the example of FIG. 11, a partially synthesized waveform A in which the phase spectrum is replaced may be created. In the next step S66, a comparison range between the mixed waveform of the partial waveform A and the partial waveform B (an appropriate mixing ratio) and the partial waveform D is set. The comparison range is, for example, a period from the time when the volume envelope reaches the maximum value to the reference position at which a predetermined ratio of the maximum value is reached for each of the mixed waveform and the partial waveform D. When the time lengths of the comparison range of the mixed waveform and the comparison range of the partial waveform D are different, for example, the start point side of the comparison range is shifted in accordance with the shorter time length.

In the next S67, the timing of the mixed waveform and the partial waveform D is adjusted. Although the timing adjustment described with reference to FIG. 4 relates to the attack part, the cross-correlation function is calculated by the same method as this, and
In the comparison range set in 6, the partial waveform D is shifted so that the correlation between the mixed waveform and the partial waveform D becomes high.

The waveform reproduction is performed as follows. In response to the note-on, the reproduction of the partial waveforms A and B of "the attack part to the introduction part of the sustain part" is started simultaneously. At this time, the mixing ratio of the partial waveform A and the partial waveform B is controlled according to the velocity value included in the note-on information. Partial waveforms A and B
At the end of the attack section, the reproduction of the partial waveform D of "continuation section to release section" is started. From the mixed waveform of the partial waveform A and the partial waveform B to the partial waveform D, it is cross-faded at the introduction part of the sustain part. The change in the volume for the crossfade is based on the partial waveforms A, B, and D stored in the waveform memory.
From the beginning. Alternatively, the synchronized partial waveforms A, B, and D are stored in a memory, and the volume envelope of each partial waveform is controlled using a weighting coefficient for crossfading during a crossfade period during reproduction. Is also good.

As described above, the “introduction portion from the attack portion to the sustain portion” is a waveform in which the mixing ratio of the partial waveform A and the partial waveform B is controlled. "Continuation department-release department"
In the partial waveform D, the mixed waveform of the partial waveforms A and B and the partial waveform D are connected in a cross-fade manner in the introduction portion of the sustaining portion. At this time, since the timing of the mixed waveform of the partial waveforms A and B and the timing of the partial waveform D are aligned so that the degree of correlation is the highest, noise or distortion of the waveform due to mixing is less likely.

In the above description, in the first applied example, the timing of the target waveform A and the reference waveform B are adjusted so that the correlation between the waveforms in the attack portion (rising portion) of the waveform is increased. Was. In addition, in the second application example, in addition, the timing of the mixed waveform of the partial waveforms A and B and the timing of the partial waveform D is increased so that the correlation between the waveforms becomes high in the introduction portion of the sustain portion after the volume envelope becomes maximum. Was combined. However, depending on the purpose of using the partial waveform, the waveform is not limited to the above-described waveform portion, but may be a sustained portion where the waveform is substantially flat or a release portion of the waveform.
The timing may be adjusted so as to increase the correlation. To set the comparison range for calculating the correlation when timing is adjusted in the release section, the last time the decay rate accelerated is detected in the volume envelope of both waveforms, and a predetermined comparison What is necessary is just to set a range.

In the above description, an approximate comparison range having a high correlation is set by paying attention to the volume envelope. In addition, or in lieu of that, the pitch envelope (the temporal change of the fundamental frequency) and the spectral envelope (the temporal change of the sum of the levels of the frequency components satisfying the specific conditions) are set for the comparison range. May be the basis for For example, when the timing of a slur waveform, a glide waveform, a vibrato waveform, etc., whose pitch changes transiently during one tone generation period, is to be aligned between the target waveform A and the reference waveform B, the timing is based on the pitch envelope. It is preferable to set the comparison range by using the following. Further, since the attack portion has many anharmonic components and the anharmonic component is attenuated when the attack portion ends, the comparison range of the attack portion can be set based on the spectrum envelope of the anharmonic component.

[0058]

As apparent from the above description, the present invention has an effect that the timing can be automatically adjusted so that the correlation between the two waveforms is increased. in advance,
Since the comparison range can be appropriately set based on the envelope of the waveform parameter, there is an effect that the timing can be effectively adjusted in the calculation of the degree of correlation. Further, even if the input target waveform A and the reference waveform B are in a relationship where the polarity is inverted, there is an effect that the polarity can be automatically detected and the timing can be adjusted.

[Brief description of the drawings]

FIG. 1 is a block diagram of a waveform editing apparatus that executes a waveform editing method according to the present invention.

FIG. 2 is a flowchart showing an outline of a process of analyzing a waveform by using the waveform editing apparatus shown in FIG. 1 and performing a performance by a person;

FIG. 3 is a first flowchart for explaining the operation of the embodiment of the waveform editing method of the present invention.

FIG. 4 is a second flowchart for explaining the operation of the waveform editing method according to the embodiment of the present invention;

FIG. 5 is a waveform diagram of a target waveform A and a reference waveform B.

FIG. 6 is a waveform diagram showing a volume envelope of a target waveform A and a reference waveform B;

FIG. 7 is an enlarged waveform diagram of a target waveform A and a reference waveform B.

FIG. 8 is a cross-correlation function R between a target waveform A and a reference waveform B;
_It is a diagram showing _AB .

FIG. 9 shows a polarity inverted waveform C obtained by inverting the polarity of the target waveform A.
FIG. 7 is a diagram showing a cross-correlation function R _CB between a reference waveform B and a reference waveform B;

FIG. 10 is a comparison diagram of the target waveform A and the reference waveform B after the start address of the target waveform A is shifted by the time difference Δt and the timing is adjusted.

FIG. 11 is a flowchart illustrating a first example of creating a sound source waveform for touch control.

FIG. 12 is a flowchart illustrating a second example of generating a touch control sound source waveform.

[Explanation of symbols]

1 CPU, 2 ROM, 3 RAM, 4 timer, 5
Panel switch, 6 panel display, 7 MIDI interface, 8 driving device, 10 waveform memory, 11
Access management unit, 12 external waveform input terminal, 13 writing circuit, 14 buffer, 15 sound source unit, 16 sound system, 17 bus line

Claims (2)

[Claims] 1. A waveform editing method for adjusting the timing of the target waveform and the reference waveform so that the correlation between the target waveform and the reference waveform is high, wherein: Detecting an envelope of two waveform parameters; setting a comparison range for the target waveform and the reference waveform based on the envelopes of the target waveform and the reference waveform; and Calculating the degree of correlation with the reference waveform, calculating the time difference between the target waveform and the reference waveform when the degree of correlation is maximized, relative to the target waveform by the time difference with respect to the reference waveform Performing a shifting process. 2. A waveform editing method for adjusting the timing of the target waveform and the reference waveform so that the correlation between the target waveform and the reference waveform is high, wherein a certain one of the target waveform and the reference waveform is provided. Detecting an envelope of the waveform parameter; setting a comparison range for the target waveform and the reference waveform based on the envelopes of the target waveform and the reference waveform; and Reference waveform,
And calculating a degree of correlation for the polarity-reversed waveform obtained by inverting the polarity of the target waveform and the reference waveform, and when the correlation is maximized, a time difference between the target waveform or the polarity-reversed waveform and the reference waveform. When calculating the correlation degree between the target waveform and the reference waveform, the correlation degree is maximized, while shifting the target waveform relative to the reference waveform by the time difference, When the degree of correlation between the polarity-reversed waveform and the reference waveform is the largest, the processing of replacing the target waveform with the polarity-reversed waveform and relatively shifting the reference waveform by the time difference. A waveform editing method.