JP3636056B2 - Waveform data processing method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a waveform data processing method for creating a waveform of a loop portion that is repeatedly reproduced during waveform reproduction of a plurality of waveform data related to each other.
[0002]
[Prior art]
As a waveform memory sound source, a musical sound signal when a musical instrument is played is recorded in stereo by a microphone, and the obtained stereo original waveform data of Lch (left channel) and Rch (right channel) is processed and stored in the waveform memory. Some stereo sound waveform data are read in parallel in response to note-on to generate a 2-channel musical sound waveform.
Here, parallel reading of the original stereo waveform data is performed using two time-division sound generation channels of the waveform memory sound source, and a musical sound waveform is generated based on the read waveform data.
The waveform memory sound source normally stores waveform data of an attack part and a loop part. After the attack part is read out in response to the note-on, the loop part is repeatedly read out a plurality of times, and a musical sound waveform is generated based on the read waveform data.
[0003]
In general, the portion following the attack portion of the musical tone waveform does not change completely in a steady manner, and the frequency, phase and amplitude of each frequency component slightly vary with time.
Therefore, there is no guarantee that the waveform can be connected smoothly when the loop waveform readout position returns from the loop end point to the loop start point. However, natural connection of the loop portions is important for generating high-quality musical sounds. Therefore, for example, Japanese Patent Application Laid-Open No. 2000-10565 and Japanese Patent Application No. 2000-056774 disclose that the loop portion is automatically taken out so that the connection is not unnatural.
[0004]
FIG. 9 is a hardware configuration diagram illustrating an example of a conventional waveform data processing apparatus. In the figure, 31 is a CPU, 32 is a ROM storing various programs and various control information, 33 is a RAM used as a work area, a buffer area or an area for storing various programs, and 34 is a clocking operation or timer interrupt. It is a timer for performing.
Reference numeral 35 denotes a panel switch provided with various operation switches, and reference numeral 36 denotes a panel display for displaying an original waveform, a frequency component and the like. 37 is a MIDI interface, and 38 is a drive device for accessing a recording medium 39 such as a CD-ROM, HD (hard magnetic disk), MO, FD or the like.
[0005]
Reference numeral 40 denotes a waveform memory. 41 is an access management unit for managing access time slots of the waveform memory 40 so that accesses from the writing circuit 43, the sound source unit 45, and the CPU 31 do not collide with each other, 42 is an external waveform input terminal, and 43 is input from the outside. A write circuit 44 that samples the original waveform data to be written and writes it in the waveform memory 40, 44 is for transferring the waveform data written in the waveform memory 40 from the recording medium 39 or RAM 33 or the waveform data read out from the waveform memory 40. It is a buffer.
45 is a sound source unit, 46 is a sound system that outputs a musical tone signal, and 47 is a bus line used for exchanging information between each element.
A communication interface circuit may be provided to download waveform data, various programs, etc. from a server on the network. A keyboard operator may be provided.
[0006]
The CPU 31 controls the entire apparatus according to external inputs from the panel switch 35 and the MIDI interface 37 in accordance with various control programs stored in the ROM 32 and RAM 33.
During the sound source waveform data creation process, data in the waveform memory 40 is read and written via the buffer 44, and the waveform data is read, analyzed, processed, edited, and written again into the waveform memory 40, or the recording medium 39 or communication. Waveform data supplied from the network is written into the waveform memory 40, or conversely, supplied from the waveform memory 40 to the recording medium 39 or the communication network.
Further, at the time of performance processing execution, the tone generation state of the tone generation channel is controlled according to performance information supplied from the MIDI interface 37, the recording medium 39, the RAM 33, and the like. The tone generator 45 generates a musical tone using the waveform data selected from the waveform memory 40 using the assigned sound generation channel.
[0007]
FIG. 10 is a flowchart showing an outline of processing of a conventional waveform data processing apparatus.
In S51, the original waveform data is written into the waveform memory 40. In S52, waveform data analysis processing is performed. In S53, the waveform data of the attack part and the loop part are processed and synthesized based on the analysis value of the frequency component in S52, and the synthesized sound source waveform data is stored in the waveform memory 40. Write.
In S54, performance control information is created based on the MIDI data using the sound source waveform data created in S53, and supplied to the sound source unit 45. The tone generator 45 reads the musical sound waveform data from the waveform memory 40 based on the performance control information, reads the loop waveform repeatedly for the loop portion as it is for the attack portion waveform, gives a predetermined envelope, effects processing, etc. To generate and synthesize musical tones for each channel and supply them to the sound system 46.
[0008]
Next, the waveform data analysis process of S52 described above will be described.
The original waveform data is frequency-analyzed by short-time Fourier analysis (STF analysis) using a time window centered on a plurality of analysis points. The original waveform data is multiplied by a window function, and frequency, frequency and amplitude data of frequency components are obtained by FFT (Fast Fourier Transform). Detect all frequencies where the amplitude peaks. The analysis point is moved, and the above-described processing is repeated to extract an amplitude peak at each analysis point. When the frequency, phase, and amplitude of each frequency component forming a peak are compared at the previous and next analysis points and adjacent peaks are associated and connected, a plurality of peak loci are obtained.
[0009]
Unnecessary components such as noise are removed from this trajectory, and a trajectory of a desired frequency component such as a fundamental tone component, overtone component, or anharmonic component is selected, and a sine wave signal corresponding to each frequency component is selected. By generating and synthesizing a sine wave, a deterministic waveform is synthesized from the original musical sound waveform data. Further, a residual waveform is obtained by subtracting a waveform obtained deterministically from the original musical sound waveform. The frequency component of the selected locus is modified, and the remaining waveform is modified by equalizer, FFT or other signal processing. By adding the two modified waveforms, a desired tone waveform is obtained.
[0010]
FIG. 11 is a diagram showing the phase trajectory of the fundamental component for explaining the loop point determination process and the smoothing process for cutting out the loop portion from the original waveform data.
The phase means a phase of a waveform corresponding to each frequency component. Therefore, it is not a phase in the sense of a phase difference of each frequency component obtained by frequency analysis with respect to a reference phase.
The loop start point is determined, and then the optimum loop end point is determined in relation to the loop start point while relatively moving the temporarily input loop end point.
Based on the phase of the fundamental component at the loop start point as a reference, the position at which the phase of the fundamental component is separated by 2Nπ (N is a positive integer value that can be arbitrarily specified in advance) is defined as the loop end point.
[0011]
However, even if the position where the phase of the fundamental component is separated by 2Nπ from the loop start point is set as the loop end point, the frequency ratio between the m harmonic component and the fundamental component (m = 1) constituting the original waveform data is not necessarily m. Since it is not doubled, the phase difference between the loop start point and the loop end point may be shifted from 2 mNπ depending on the frequency component.
Therefore, the smoothing process shown in FIG. 11B is executed for phases other than the fundamental component.
In the smoothing process, the phase difference between the loop start phase and the loop end point becomes an integer multiple of 2π by changing the phase gradient of the harmonic component between the loop start point and the loop end point. In this way, the phase is made continuous. Here, it is not necessarily 2 mNπ. As long as the phase difference is an integral multiple of 2π at the harmonic frequency, phase continuity is ensured. Therefore, the phase difference is set to the nearest integer multiple of 2π so that the minimum phase inclination adjustment is sufficient.
[0012]
Next, a specific example of the smoothing process will be described.
The frequency component of each analysis point is read from the RAM 33 between a preset loop start point and a loop end point determined from the phase continuity of the fundamental component. The read phase data of each analysis point is uniformly multiplied by an appropriate coefficient so that the phase difference between the loop start point and the loop end point is an integral multiple of 2π.
The smoothing process described above may be performed for one or more desired frequency components. Note that the fundamental tone component can be similarly smoothed when the phase difference is not strictly 2Nπ.
[0013]
If the above-described extraction of the loop waveform and phase adjustment in the loop part are separately performed for each of the stereo Lch and Rch waveform data, appropriate loop start points and loop end points are automatically set for each waveform data. be able to. However, there is no guarantee that the Lch and Rch loop start points and loop end points will be the same. This is considered to be caused by a difference in the degree of modulation with respect to the pitch and recording conditions for each channel.
[0014]
Here, it is not always necessary to match the loop start point and the loop end point in both channels, and each point may be freely shifted in order to correct the recording condition or aim for a special sound field effect.
However, it was found that the time from the loop start point to the loop end point (loop length: loop length) should be the same.
The reason will be described next. In order to indicate the address of the sound source waveform data stored in the waveform memory at the time of sound generation, the F number calculation method is adopted. This F number is a parameter that controls the speed at which the address counter advances. The F number that is proportional to the pitch frequency (musical pitch) is accumulated by the counter every predetermined clock period, and the integer part is obtained. The sound source waveform data is read out as the address of the sound source waveform data. Since the same pitch frequency is specified for the Lch and Rch waveform data, the same F number is used for reading the sound source waveform of each channel. However, if the loop lengths are not equal, the loop repetition periods of the two sound generation channels are different.
[0015]
Since the sound source waveform with a length that is an integral multiple of the pitch period of the fundamental tone component is stored in the loop part, the sound of the loop part is based on the difference between the loop lengths and the basic sound waveform of both channels that are sounded. The pitch frequency will be different. In addition, as the number of repetitions of the loop portion increases, the positions in the loop waveform read by both channels are shifted from each other. In general, in stereo reproduction, it is known that if the phase relationship between the Lch and Rch waveforms changes, the localization will be affected. In this case as well, the sense of localization of the generated stereo waveform becomes unstable.
[0016]
Here, if the read-out pitch of the sound source waveform of the Lch and Rch loop portions is adjusted, the above-described problem is solved. For example, the F number for reading the sound waveform data of the short loop is adjusted to be smaller than the F number for reading the sound waveform data of the long loop.
However, the number of digits of the F number is finite. Even if fine adjustment is possible with the F number, a slight error remains in the adjustment. As a result, the period for reading out the waveform does not completely match, and a slight error remains. Since these slight errors are accumulated and the phase relationship gradually shifts, the sense of localization of the reproduced stereo sound signal becomes unstable.
[0017]
[Problems to be solved by the invention]
The present invention has been made in order to solve the above-mentioned problems. From a plurality of mutually related original waveform data such as a stereo musical sound waveform, the waveform data of the loop portion is obtained without impairing the mutual phase relationship. It is an object of the present invention to provide a waveform data processing method for creating a waveform.
[0018]
[Means for Solving the Problems]
The present invention according to claim 1 is a waveform data processing method for creating a waveform of a loop portion that is repeatedly reproduced during waveform reproduction based on original waveform data, wherein the original waveform data is A loop length setting step for setting a loop length of a reference loop portion that is repeatedly generated during waveform playback, which is created based on the related reference original waveform data, and the loop portion for the original waveform data A loop position determining step for determining the position of the loop section so that the loop length of the reference loop section is equal to the loop length of the reference, and at least the phase of the fundamental component of the frequency components of the original waveform data is , The at least fundamental component of the original waveform data in the loop portion so as to continuously change from the end position to the start position of the loop portion. And it has a loop phase adjustment step of adjusting the.
Therefore, since the loop length of the original waveform data can be matched with the loop length of the reference loop part of the original original waveform data, the waveform of the loop part and the reference loop part waveform can be reproduced during waveform playback. Can be reproduced without impairing the mutual phase relationship.
If the waveform of the loop part is created by waveform synthesis based on a frequency component obtained by frequency analysis of the original waveform data, a desired component among the frequency components included in the original waveform data is obtained. The waveform of the loop portion can be freely created by selecting only the waveform or removing the residual waveform component.
Further, in the loop position determination step, the start phase in the loop portion of the fundamental component of the original waveform data and the start phase in the loop portion of the fundamental component of the reference original waveform data are matched. If the start position of the loop part of the original waveform data is determined, the time difference between the time when the loop part is started and the time when the reference loop part is started is left as it is. And the phase difference between the waveform of the reference loop portion. As a result, the phase relationship between the loop part and the reference loop part can be easily controlled during waveform reproduction.
[0019]
The invention according to claim 2 is a waveform data processing method for creating a waveform of a loop portion that is repeatedly reproduced during waveform reproduction based on the original waveform data, the reference being related to the original waveform data and A loop length setting step for setting a loop length of a reference loop portion which is created based on the original waveform data and which is repeatedly reproduced at the time of waveform reproduction, and the loop length of the loop portion with respect to the original waveform data, The loop position determining step for determining the position of the loop part so as to be equal to the loop length of the reference loop part, and the reference for the phase of the fundamental component of the original waveform data in the loop part In-loop phase adjustment step for giving the same phase gradient as the phase gradient in the reference loop portion of the fundamental component of the original waveform data It is intended.
Therefore, since the loop length of the original waveform data can be matched with the loop length of the reference loop portion, it is possible to reproduce the waveform without damaging the mutual phase relationship.
[0020]
According to a third aspect of the present invention, in the waveform data processing method according to the second aspect, the loop position determining step matches the loop position of the loop part with the loop position of the reference loop part. Yes,
Further, the phase of the fundamental waveform component of the original waveform data is shifted from the position before the loop section to the start position of the loop section from the slope of the phase of the fundamental waveform component of the original waveform data. It has a pre-loop phase adjustment step that gradually changes to the phase gradient of the component.
Therefore, it is possible to make the gradient of the phase of the fundamental component continuous when the position before the loop portion of the original waveform reaches the start position of the loop portion.
[0021]
According to a fourth aspect of the present invention, in the waveform data processing method according to the third aspect of the present invention, in the phase adjustment before loop phase, the phase of the fundamental component of the original waveform data at the start position of the loop portion is The original waveform data is shifted with respect to the reference original waveform data so that the phase of the fundamental waveform component of the original waveform data is substantially maintained before the phase is changed. While changing the phase, the position of the loop portion of the original waveform data is shifted.
Therefore, the start phase of the fundamental sound component when the position before the loop portion of the original waveform reaches the start position of the loop portion can be held at the value before the phase adjustment.
[0022]
According to a fifth aspect of the present invention, in the waveform data processing method according to any one of the first to fourth aspects, at least a phase of a fundamental component among frequency components of the reference waveform data is the reference and A reference loop position determining step for determining the position of the reference loop portion of the original waveform data to be continuously changed from the end position of the loop portion to the start position.
Therefore, the waveform connection from the end position to the start position of the reference loop portion can be improved for the reference waveform data.
[0023]
DETAILED DESCRIPTION OF THE INVENTION
FIG. 1 is a functional block configuration diagram of a first embodiment of a waveform data processing apparatus of the present invention.
In this embodiment, the Lch and Rch waveforms are subjected to STF analysis, the loop start point and the loop end point are determined for each, and the processed waveform data of Lch and Rch are stored in the storage device. As the hardware configuration, the configuration of the conventional processing apparatus shown in FIG. 9 may be used.
In the figure, 1 is an Lch original waveform data storage unit, and 8 is an Rch original waveform data storage unit. Reference numeral 2 denotes an Lch STF analysis (short-time Fourier analysis) unit, and 9 denotes an Rch STF analysis unit, which respectively perform short-time Fourier analysis on the original waveform data. A deterministic component (hereinafter referred to as an “STF component”) for adding and synthesizing a deterministic waveform is separated from a residual waveform. This STF component has amplitude, phase, and frequency data of the peak frequency component. On the other hand, the residual waveform is a value obtained by subtracting a waveform obtained deterministically from the original waveform.
[0024]
Reference numeral 3 denotes an input unit that stores start point, loop start point, and temporary loop end point address information for the Lch original waveform data, start point for the Rch original waveform data, and address information for the loop start point. Is the input part. The start point is the start position of the attack portion, and the same address information is set for Lch and Rch.
Instead of setting input by the user, a function block that can automatically set the address information described above may be provided.
Reference numeral 4 denotes an Lch loop address determination unit, which is based on the Lch loop start point, each address information of the temporary loop end point, and the Lch STF component in the same manner as in the prior art. Each address information and loop length (loop size) of the loop end point is determined. The start point is output to the Lch residual waveform loop processing and synthesis unit 6 as it is.
Further, the input unit 3 may specify a value as to how many times the period of the fundamental component is to be used instead of the temporary loop end point.
[0025]
Reference numeral 5 denotes an Lch loop adjustment unit, which is the same as the “smoothing process” in the prior art. From the determined loop end point to the loop start point, the STF component is processed so that the phases of the fundamental component and each harmonic component are continuously connected.
Reference numeral 6 denotes a loop processing and synthesis unit for the Lch residual waveform. First, a residual waveform suitable for the loop portion is synthesized based on the original residual waveform. This is hereinafter referred to as a synthesized residual waveform. Next, an area having a predetermined width is set before the loop start point. The original residual waveform from the start point to this predetermined width region is used as it is for the synthesis described later. The original residual waveform in the predetermined width region is used for synthesis after cross-fading with the waveform of the predetermined width region in front of the loop end point of the above-described synthesized residual waveform. From the loop start point to the loop end point, the synthesized residual waveform is used for synthesis. As a result, a residual waveform from the start point through the loop start point to the loop end point is finally created.
Next, the finally created residual waveform is synthesized with a deterministically obtained waveform that is sine wave synthesized based on the STF component after the loop adjustment from the start point to the loop end point. Reference numeral 7 denotes a processed waveform data memory for an Lch waveform memory sound source, which stores combined waveform data composed of an attack portion and a loop portion. Note that the waveform data of the loop portion may be only a waveform obtained deterministically.
[0026]
Reference numeral 10 denotes an R channel loop address determination unit. The address information of the Rch loop end point is determined from the address information of the Rch loop start point and the loop length previously determined for the Lch. The start point is output to the Rch residual waveform loop processing and synthesis unit 12 as it is.
Reference numeral 11 denotes an Rch loop adjustment unit, which has the same function as the Lch loop adjustment unit 5, but is used differently. Reference numeral 12 denotes an Rch residual waveform loop processing and synthesis unit. Reference numeral 13 denotes an Rch processed waveform data storage unit, which is the same as the Lch side.
[0027]
FIG. 2 is an explanatory diagram of an operation of extracting a loop waveform and a smoothing process in the first embodiment of the present invention. The reference channel is processed as Lch.
As shown in FIG. 2A, as a step 1, the Lch loop start point LSP _L Enter. Next, a temporary loop end point is input, and the loop start point LSP is moved from the temporary loop end point forward or backward on the time axis based on the STF analysis. _L To loop end point LEP _L Loop end point LEP suitable for loops such that the phase difference of the fundamental component up to 2Nπ is _L Lch loop length LL _L To decide. FIG. 2B shows the locus of the phase of the fundamental component of the Lch.
As step 2, loop start point LSP for Rch waveform data _R Enter. Next, the Lch loop length LL determined by the Lch loop address determination unit 4 _L Rch loop length LL to be equal to _R Set the loop end point LEP _R Address information is determined. Lch loop length LL _L And Rch loop length LL _R Must be matched exactly. FIG. 2C shows the locus of the phase of the fundamental component of Rch.
[0028]
If the loop lengths are made equal, the temporal phase relationship of each channel can be kept constant by making the F number values of the sound channels of Lch and Rch the same.
However, as a result of forcibly adjusting the Rch loop length to the Lch loop length, the Rch loop start point LSP is also obtained for the fundamental component. _R To loop end point LEP _R The phase difference up to is not necessarily 2Nπ.
Therefore, when it does not become 2Nπ, the conventional loop smoothing process is performed for the fundamental component as well as the harmonic component and the anharmonic component.
[0029]
For example, as shown in FIG. 2 (c), the loop start point LSP _R And loop end point LEP _R For the fundamental component in between, multiply the phase value by a coefficient and expand or compress in the phase axis direction so that the phase difference of the fundamental component becomes an integer multiple of 2π closest to the original phase difference. adjust. As a result, the loop length becomes N times the period of the fundamental component for both Lch and Rch. For the frequency components other than the fundamental component of Rch, the loop start point LSP is the same as in the past. _R And loop end point LEP _R Then, the phase difference is adjusted so as to be an integer multiple of 2π closest to the original phase difference.
Note that the time positions (address information) of the loop start points of the Lch and Rch waveforms do not have to match. It is only necessary that the start phase of each fundamental sound component at the loop start point of each waveform always maintains a predetermined phase relationship during the repetition of the loop portion.
[0030]
FIG. 3 is an explanatory diagram of an operation for extracting a loop waveform and a smoothing process in the second embodiment of the present invention.
In this embodiment, the loop smoothing process is performed by compressing and expanding on the time axis.
Step 1 is the same as that in FIG. In the step 2 shown in FIG. 3A and the diagram shown in FIG. 3B, the loop end point LEP is also applied to the fundamental component of Rch in the same manner as before. _R1 And set the loop length to LL _R And
In step 3 shown in FIG. 3A and the diagram shown in FIG. 3C, the loop length LL _R Is Lch loop length LL _L Is equal to the loop start point LSP _R To loop end point LEP _R1 The STF component up to is compressed or expanded. As a result, the loop length matches between Rch and Lch.
[0031]
FIG. 4 is an explanatory diagram of an operation of extracting a loop waveform in the third embodiment of the present invention.
In this embodiment, the loop length LL of each channel is based on the Lch and Rch waveform data. _L , LL _R , And the Lch and Rch common loop length LL based on the two temporarily determined loop lengths _X Is finally determined.
[0032]
In steps 1 and 2, address information of the loop start point and the loop end point is determined independently for Lch and Rch by the same method as in the prior art. However, it is necessary to match between both channels how many times the period of the fundamental component is to be the loop length. This address information is tentatively determined, and the Lch and Rch loop lengths LL _L , LL _R Is also a tentative decision.
In step 3, the final loop length LL is determined according to the input control value α as follows: _X To decide.
LL _X = LL _L × α + LL _R × (1-α)
That is, the loop length LL _L , LL _R The final loop length LL is obtained by interpolating (or extrapolating) _X To decide. This determined loop length LL _X And loop start point LSP _L , LSP _R Loop end point LEP based on the address information of _L , LEP _R Address information is determined.
[0033]
In this embodiment, smoothing processing is also performed for the fundamental components of Lch and Rch. In this embodiment, for example, when different pitch modulation or the like is applied to the original waveforms of Lch and Rch and the pitch is shifted, the fundamental component of the channel with the higher pitch is expanded in the time axis direction to loop length LL. _X By compressing the fundamental component of the channel with the lower pitch in the time axis direction, the loop length LL _X Is effective in reducing the pitch difference.
[0034]
FIG. 5 is a partial view of the functional block configuration of the fourth and fifth embodiments of the present invention. This partial diagram is an internal configuration diagram of the fundamental tone component phase synthesizing unit 14 indicated by a broken line in the functional block configuration diagram shown in FIG.
FIG. 6 is a schematic diagram showing a temporal change in the phase differential value of the fundamental component in the functional block configuration shown in FIG. The phase differential value is a frequency but also a phase gradient.
In this embodiment, the fundamental component is modified for the waveform from the loop start point on the Rch side to the loop end point on which the loop length equal to the loop length of the Lch is set, and the slope of the phase is changed to Lch. To match the phase gradient of the fundamental tone component (phase gradient fitting). Since the continuity of the phase of the fundamental component from the loop end point to the loop start point is secured on the Lch, naturally the continuity of the phase of the fundamental component from the loop end point to the loop start point is secured on the Rch side as well. Is done.
However, as it is, the phase gradient of Rch is often discontinuous at the loop start point, and if the degree of the phase gradient discontinuity is large, it may be felt as noise during music reproduction.
[0035]
In FIG. 5, 21 is a differential processing unit that outputs the phase gradient of the Lch fundamental component, and 22 is a differential processing unit that outputs the phase gradient of the Rch fundamental component. FIG. 6 shows the differential outputs as Lch and Rch. The Lch and Rch loop start points and loop end points are matched.
23 is a cross-fade synthesizing unit that gradually approaches the phase gradient of the Lch fundamental component from the phase gradient of the Rch fundamental component. From the loop start point to the loop end point, the phase of the Lch fundamental component It has a function to make it completely coincide with the inclination of.
[0036]
Reference numeral 24 denotes an integration processing unit, which initially sets the phase of the fundamental component of Rch at the crossfade start point at a position a predetermined distance before the start point of the loop part (for example, the position immediately after the waveform attack ends). As a value, the phase of the synthesized fundamental tone component is output by integrating the slope of the synthesized phase. This output is used as the phase of the fundamental component of Rch in FIG. Note that the amplitude of the fundamental component of the Rch is used as it is for the output of the STF analysis unit 9.
For harmonic components other than the fundamental tone component, smoothing processing is performed so that the phase is continuously connected from the loop end point to the loop start point.
[0037]
FIG. 7 is a schematic diagram for explaining the operation of synthesizing the phases of the loop waveforms in the above-described fourth embodiment of the present invention. However, FIG. 7 is not drawn so as to accurately match the phase inclination shown in FIG.
FIG. 7 shows the phases of the fundamental sound components of Lch and Rch and the trajectory of the phase of the fundamental sound component of Rch synthesized by cross-fading. The phase of the fundamental component of the Rch synthesized by the crossfade gradually approaches the phase of the fundamental component of the Lch from the crossfade start point. From the loop start point to the loop end point, the phase of the fundamental component of the Lch And the phase trajectory of the fundamental component of Rch are parallel with a constant phase difference. That is, between the loop start point and the loop end point, the phase difference of the Lch fundamental component and the phase difference of the Rch fundamental component are equal.
[0038]
Since the phase of the fundamental component of Lch is continuous from the loop end point to the loop start point, the phase of the fundamental component of Rch is also continuous. As a result, an appropriate loop can be cut out for both channels at the loop start point and loop end point determined for one channel.
By performing the cross-fading process in this way, the Rch phase gradient when entering the loop loop start point from the front of the loop portion can be made continuous.
[0039]
However, when the cross-fading process is performed, in FIG. 7, the phase of the synthesized fundamental component of the Rch is slightly shifted from the phase of the fundamental component of the Rch before synthesis at the loop start point.
At this time, the phase of the synthesized fundamental component of the Rch matches the phase 0 of the original fundamental component of the Rch before the synthesis at the loop start point. The point is returned to the left by (Δ <0). Depending on the phase relationship between the fundamental component of Lch and the original Rch, there may be a point (0 <Δ) that advances to the right.
[0040]
FIG. 8 is a schematic diagram for explaining the operation of synthesizing the phases of the loop waveforms in the fifth embodiment of the present invention.
In this embodiment, according to the fourth embodiment described above, the crossfade synthesis is calculated, the value of Δ is calculated in advance, and the STF component of the original Rch is delayed by this Δ. Thus, cross-fade processing and phase gradient fitting processing in the loop portion are performed.
In FIG. 7, since Δ <0, the position of the cross-fade start point is returned to the waveform start end side by Δ from the original Rch black circle position to the white circle position shown in the figure. FIG. 6 also shows the positions of black circles and white circles.
Even with the above-described processing, there is no guarantee that the phase of the fundamental component of the original Rch and the phase of the synthesized fundamental component of the Rch completely match at the loop start point. However, they can be matched better than the case shown in FIG.
[0041]
In each of the above-described embodiments, the phase difference between the start phases at the time of loop start of the fundamental components of Lch and Rch is not considered. However, if the channel start point of each channel is determined so that the start phase of the Rch fundamental component at the loop start point matches the start phase of the Lch fundamental component at the loop start point, the synthesized musical sound is produced. This is suitable for musical tone control when the sound is played.
In each of the above-described embodiments, the loop start point of the fundamental sound component of the Lch is input to determine the loop end point. Conversely, the loop end point may be input to determine the loop start point. In the first to third embodiments described above, the loop start point of the Rch fundamental component is input, but a loop end point may be input conversely.
[0042]
In the above description, the loop length of the reference Lch is determined such that the phase of the fundamental component obtained by frequency analysis of the original waveform data changes continuously from the loop end point to the loop start. However, the loop end point can be determined by searching for a loop end point that does not cause phase discontinuity, including not only the fundamental component but also multiple desired harmonic components, fluctuations in the residual waveform, and the fundamental component. The loop end point may be determined in consideration of the amplitude fluctuation period.
Also, without analyzing the frequency of the reference Lch waveform data, the loop start point where there is a loop end point that can be connected to the loop end point without any noise in terms of only the waveform amplitude is searched, and as a result The reference Lch loop length may be determined.
[0043]
In the above description, the waveform of the Rch loop portion is created by synthesizing the waveform based on the frequency component obtained by frequency analysis of the waveform data. However, for the waveform data obtained by cutting out the Rch waveform data as it is so as to be equal to the Lch loop length, the fundamental component is only shown in the above-described embodiments for improving the loop connection. Waveform processing may be performed.
In the above description, the processing of the waveform data of the Lch and Rch stereo music signals related to each other has been described. However, even when the number of channels exceeds two, waveform data can be processed in the same manner based on one waveform. Moreover, it is applicable not only to a musical sound signal but also to a general waveform such as a human voice signal or an acoustic signal.
[0044]
【The invention's effect】
As is apparent from the above description, the present invention has an effect that it is possible to synthesize waveform data of a loop portion from a plurality of related waveform data without impairing the phase relationship between waveforms. .
When applied to stereo musical sound waveform data, the phase change of the fundamental tone of the loop portion matches between Lch and Rch, so that a stable sound image with a sense of localization can be obtained, and the difference between Lch and Rch is related to the phase of the overtone component. Rich stereo feeling remains because it is preserved.
[Brief description of the drawings]
FIG. 1 is a functional block configuration diagram of a first embodiment of a waveform data processing apparatus of the present invention.
FIG. 2 is an explanatory diagram of an operation of extracting a loop waveform and a smoothing process in the first embodiment of the invention.
FIG. 3 is an explanatory diagram of an operation of extracting a loop waveform and a smoothing process in the second embodiment of the present invention.
FIG. 4 is an explanatory diagram of an operation of extracting a loop waveform in the third embodiment of the present invention.
FIG. 5 is a partial view of a functional block configuration according to fourth and fifth embodiments of the present invention.
6 is a schematic diagram showing a temporal change in a phase differential value of a fundamental component in the functional block configuration shown in FIG.
FIG. 7 is a schematic diagram illustrating an operation of synthesizing phases of a loop waveform in the fourth embodiment of the present invention.
FIG. 8 is a schematic diagram for explaining an operation of synthesizing phases of a loop waveform in the fifth embodiment of the present invention.
FIG. 9 is a hardware configuration diagram illustrating an example of a conventional waveform data processing apparatus.
FIG. 10 is a flowchart showing an outline of processing performed until a performance is executed in a conventional waveform data processing apparatus.
FIG. 11 is a diagram showing a phase trajectory of a fundamental component for explaining conventional loop point determination processing and smoothing processing;
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 ... Lch original waveform data storage part, 2, 9 ... STF analysis part, 3 ... Input part, 4,10 ... Loop address determination part, 5, 11 ... Loop adjustment part, 6, 12 ... Loop processing of residual waveform and Synthesis unit, 7 ... Lch processed waveform data memory, 8 ... Rch original waveform data storage unit, 13 ... Rch processed waveform data storage unit

Claims (5)

A waveform data processing method for creating a waveform of a loop portion that is repeatedly reproduced during waveform reproduction based on original waveform data,
A loop length setting step for setting a loop length of a reference loop portion that is created based on the original waveform data that is related to the original waveform data and is repeatedly reproduced during waveform reproduction;
A loop position determining step for determining the position of the loop portion so that the loop length of the loop portion is equal to the loop length of the reference loop portion with respect to the original waveform data;
Among the frequency components of the original waveform data, at least the fundamental component of the original waveform data in the loop portion so that at least the phase of the fundamental component continuously changes from the end position of the loop portion to the start position. In-loop phase adjustment step to adjust the phase,
A waveform data processing method characterized by comprising: A waveform data processing method for creating a waveform of a loop portion that is repeatedly reproduced during waveform reproduction based on original waveform data,
A loop length setting step for setting a loop length of a reference loop portion that is created based on the original waveform data that is a reference related to the original waveform data and that is repeatedly reproduced during waveform reproduction;
A loop position determining step for determining the position of the loop portion so that the loop length of the loop portion is equal to the loop length of the reference loop portion with respect to the original waveform data;
A phase in the loop that gives the same slope of the phase of the fundamental component of the reference original waveform data as the slope of the phase in the reference loop part of the fundamental waveform component of the original waveform data to the phase in the loop part Adjustment steps,
A waveform data processing method characterized by comprising: In the loop position determination step, the loop position of the loop portion is matched with the loop position of the reference loop portion,
Further, the phase of the fundamental waveform component of the original waveform data is shifted from the position before the loop section to the start position of the loop section from the slope of the phase of the fundamental waveform component of the original waveform data. Pre-loop phase adjustment step that gradually changes to the phase gradient of the component,
The waveform data processing method according to claim 2, wherein: In the pre-loop phase adjustment step, the phase of the fundamental component of the original waveform data at the start position of the loop portion is substantially maintained at the phase before the phase of the fundamental component of the original waveform data is changed. After shifting the original waveform data with respect to the reference original waveform data, the phase of the fundamental component of the original waveform data is changed, and the position of the loop portion of the original waveform data is shifted,
The waveform data processing method according to claim 3. Among the frequency components of the reference waveform data, at least the fundamental component phase becomes the reference of the reference original waveform data so that the phase of the reference loop portion continuously changes from the end position of the reference loop portion to the start position. A reference loop position determining step for determining the position of the loop portion;
5. The waveform data processing method according to claim 1, wherein the waveform data processing method is provided.

Images