JP2000003178A - Time stretch device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To time-stretch an input signal without injuring musicality corresponding to various types of musical sounds by changing the number of division of a band and a characteristic to be used of a filter according to the type of the input signal. SOLUTION: The type setting picture of the input signal displayed on a screen of a display device 5 is shown, and a user specifies either one between two kinds of a rhythm sound system and a melody sound system as a type of a musical sound by using an operation element such as a cursor key. The number of division of a band and a characteristic of a filter, e.g. the number of taps used by a time stretch means are changed. The input signal is inputted from an A/D converter 8 and an external storage device (HDD 6, CD-ROM 7) to be transferred to a RAM 3. These input data are read out temporarily from the RAM 3 to be analyzed under control of a CPU 1, and the analyzed data are stored again in the RAM 3 to be transferred to the HDD 6, etc., thereafter.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a time stretching apparatus capable of changing a tempo or the like of a series of sounds such as a phrase independently of a pitch. Such a time stretching device can be mounted on an electronic musical instrument and used.

[0002] A series of musical strings such as a certain natural melody line is called a phrase (also referred to as a phrase, a phrase, etc.). A simple method can be realized by storing the data of the phrase in a memory and changing the reading speed. However, in this method, when the tempo (that is, the reading speed) is changed, the pitch also changes accordingly.

Therefore, several methods for changing the tempo without changing the pitch have been proposed. For example, a method for detecting a silent section and adjusting the length of the section to change the tempo without changing the pitch, or a trapezoidal method. There is a method of using a window to cut out a waveform from a phrase.

However, in the method using a silent section, there is a limit in adjustment for a phrase having no or few silent sections. In the method using a trapezoidal window, the envelope of the trapezoidal window is heard as noise, which is not satisfactory.

Further, after the input signal is divided into a plurality of bands, the amplitude and phase change (pitch) of each band are detected, and a tempo change is realized using the amplitude data to adjust the phase change. There is a method of realizing a pitch change by performing the above.

FIG. 13 shows a time stretch (corresponding to time compression / expansion, that is, tempo adjustment) by this method.
Is shown. As shown in the figure, the input signal is divided into 100 bands (Bands) Band0 to Band99 as shown in FIG. 14 by using, for example, 100 band filters having different passbands, and the signals of the respective divided bands are divided. By extracting the amplitude component and the frequency component (pitch component) of each of the bands, and changing the time change speed of the amplitude component and the time change of the frequency component of each band in accordance with the tempo change instruction amount, The time axes of the amplitude component and the frequency component are respectively compressed / expanded as shown in FIG. 15, for example, and the original input signal is time-stretched and output by combining the amplitude component and the frequency component of each band. Things. Here, FIG.
5 (A) is the waveform of the original signal (amplitude component and frequency component),
FIG. 15B is a waveform of a signal (amplitude component and frequency component) after the time axis is expanded.

[0007] The principle of this method is that the signal of each band can have a sufficiently small bandwidth (hence the frequency change width within the band) by sufficiently increasing the number of band filters, and thus the time axis can be compressed / expanded. Even if the frequency component is hardly changed, the pitch does not change even if the time axis is compressed / expanded without changing the amplitude value of the amplitude component.

[0008]

In the conventional time stretching method based on band division, when a tone signal having a steep change (such as an attack portion) such as a rhythm sound is input, it is affected by the filter used. However, there is a problem that a noise called a pre-echo is generated in the front part of the attack part, so that the attack part becomes a dull sound and lacks in musicality. This is a phenomenon that occurs when the number of taps (filter order) of the filter that divides the band is large. By reducing the number of taps, the degree of dullness can be reduced.

On the other hand, the filter has the property that the bandwidth becomes wider as the number of taps is reduced. As a result, if the number of taps is reduced, the number of divisions into which the input signal is divided into bands must be reduced. No longer. When the number of divisions is reduced in this way, for a sound whose harmonic structure is important (a sound mainly used for a melody), the harmonic structure is not properly reproduced, and the music is also lacking in musicality.

As described above, in the conventional time stretching method based on band division, the number of taps of the filter and the number of band divisions by these filters have a trade-off relationship from the viewpoint of maintaining the musicality of the time-stretched musical sound. It has not been easy to time-stretch both rhythm-based music and melody-based music so as to be musically satisfactory.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned problems, and is capable of time-stretching an input signal without impairing musicality in correspondence with various types of musical sounds such as a rhythm system and a melody system. The purpose is to.

[0012]

In order to solve the above-mentioned problems, a time stretching apparatus according to the present invention comprises an instructing means for instructing an amount of time stretching of an input signal, and a type instructing for instructing a type of the input signal. Means for dividing the input signal into a plurality of bands by using a filter, converting the signals in each band into an amplitude component and a frequency component, and changing the speed of the time change (for example, tempo) to the designated amount designated by the designation means. Time-stretching means for time-stretching the input signal by resynthesizing the input signal after changing based on the input signal type designated by the type-designating means. Changing means for changing according to the situation.

[0013]

By designating the type of the input signal with the type designating means, the changing means changes the filter characteristic of the band division number used by the time stretching means, for example, the number of taps, according to the designated type. For example, the number of taps is controlled to be small when the input signal is a rhythm sound input signal, and the number of divisions is increased when the input signal is a melody sound signal. In this way, the attack portion of the rhythm sound input signal does not become dull, and the overtone structure of the melody sound input signal is properly reproduced. This makes it possible to perform time stretching on any type of input signal without impairing the musicality.

[0014]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 shows a configuration of a time stretching device as one embodiment according to the present invention. Such a time stretching device can be mounted on, for example, a sampler or an electronic musical instrument. In FIG. 1, a CPU 1 is a central processing unit that controls the entire apparatus, a ROM 2 is a read-only memory that stores a control program and a table of various data, and a RAM 3 is used as a working memory area. The operation unit group 4 is composed of various switches, and the display unit 5 is a liquid crystal display used to display parameters on a screen when selecting parameters. The hard disk drive 6 and the CD-ROM drive 7 are external storage devices. The A / D converter 8 converts an analog input signal into a digital signal and inputs the digital signal.
The A converter 9 converts a digital signal processed in the device into an analog output signal and outputs the signal.

Here, in this waveform converter, an input signal (input data) to be analyzed is a sample value (amplitude value) obtained by sequentially sampling a waveform of a phrase composed of a series of sound strings. Here, the original sampling frequency of the input data (or the sampling frequency of the system) is fso. Further, the input data is input from the A / D converter 8 and the external storage device (the hard disk drive device 6, the CD-ROM drive device 7) and
Transferred to M3. The input data is once read out from the RAM 3 and analyzed under the control of the CPU 1, and the analyzed data is stored again in the RAM 3 and then transferred to the hard disk drive 6 or the like.

FIG. 2 shows an input signal type setting screen displayed on the screen of the display unit 5. As shown
It is possible for the user to specify one of two types of musical sounds to be input signals, a rhythmic "rhythmic" and a melody "melodic", by using an operating element such as a cursor key. is there. The apparatus of this embodiment has an input signal type indication flag TYPE. When “rhythmic” is designated, the input signal type indication flag TYPE is set to “1”, and when “melodic” is designated, It is set to “0”. Further, when “rhythmic” is designated, the in-octave division number M and the decimation number DECI, which are variables described later, are each set to “4”. On the other hand, when "melodic" is designated, those two variables are set to "8".

The operation of the embodiment will be described below with reference to the accompanying drawings. First, the procedure of the input signal analysis processing will be described with reference to the flowcharts shown in FIGS. This input signal analysis process detects and saves the envelope of the input signal, flattens the amplitude of the input signal so as to eliminate a steep portion (for example, an attack portion), and then converts a plurality of flattened input signals. And converts the signals in each band into its amplitude component and frequency component.

FIG. 3 shows the concept of band division in this embodiment. Here, assuming that the sampling frequency of the system is fso, the frequency of the input signal is represented by an angular frequency ω, and the angular frequency ω = π corresponds to fso / 2 in frequency conversion. Here, the section where ω is from π to π / 2 is the first octave, the section from π / 2 to π / 4 is the second octave,
The section from / 4 to π / 8 is divided into tenth octaves in octave units, such as the third octave. Here, these bands are referred to as octave bands. The number of the octave band is identified by an octave number counter oct. Furthermore, each octave band is equally divided by the number M of divisions within an octave, and these divided bands (hereinafter, referred to as division bands within an octave).
Are identified by an in-octave division number counter m.
Here, as described above in the present embodiment, the division number M within an octave is “4” when the input signal type is “rhythmic” and “8” when the input signal type is “melodic”.
The example shown in FIG. 3 is for the former case of “rhythmic”.

In the analysis process, first, the input signal in
The envelope detection (i) and its flattening are performed (step S1). FIG. 6 shows a detailed procedure of this envelope detection and flattening. In the envelope detection and flattening process of FIG. 6, i is an input signal (for example, FIG. 10).
Is a counter indicating the order of sample values on the time axis, and cur is a current (current) amplitude value for performing envelope detection processing. Also, env (i)
Is an envelope (envelope) value detected from the input signal in (i). RELEAS is a release parameter (release coefficient), ATACK is an attack parameter (attack coefficient), and LEN is a data length (length of a phrase to be analyzed).

In FIG. 6, first, a counter i = 0 and a current amplitude value cur = 0 are set (step S21).
This current amplitude value cur is compared with the absolute value of the input signal in (i). That is, it is determined whether or not | in (i) |> cur. If the absolute value of the input signal in (i) is larger than the current amplitude value cur, the envelope portion of the current input signal rises sharply and can be estimated to be the attack portion. In this case,
The current amplitude value cur is calculated by the following formula: cur = cur + (| in (i) | -cur) * ATT
ACK (however, the mark “*” represents multiplication; the same applies hereinafter) is updated so as to increase in accordance with the attack coefficient ATACK (step S24).

On the other hand, when the absolute value of the input signal in (i) is equal to or smaller than the current amplitude value cur, since the envelope portion of the current input signal gradually falls, it can be estimated that the envelope portion is the release portion. In this case, the current amplitude value cur is calculated by the following formula: cur = cur + (| in (i) | -cur) * REL
It is updated by EASE so as to decrease according to the release coefficient RELEASE (step S24). The attack coefficient ATTACK and the release coefficient RELEASE
Are previously set by the user in another mode.

Thereafter, the updated current amplitude value cur is set as the envelope value env (i) of the input signal in (i) (step S25) and stored in the buffer (memory).
That is, env (i) = cur.

Next, the input signal in (i) is flattened. This means that the input signal in (i) is converted to the current amplitude value cu.
This is realized by dividing by r (therefore, the envelope value) (step S27). As a result, the input signal in (i) becomes "1" in principle, but actually the current amplitude value cur does not necessarily represent an accurate envelope value, and thus becomes a value close to "1". If the current amplitude value cur is determined to be 0 (step S2
6) Since division in step S27 cannot be performed, step S27 is skipped in this case.

Next, in order to update the position on the time axis of the input signal in (i), the counter i is incremented by one (step S28). That is, i = i + 1, and the same process is repeated until the updated counter i exceeds the data length LEN (step S29). Thereby, the envelope of the input signal for one phrase can be stored in the buffer for storing the envelope value.

FIG. 10 shows an envelope extracted from the waveform of the input signal by the above-described envelope detection processing. FIG. 11 shows a signal obtained by flattening the input signal after extracting the envelope with the extracted envelope. 3 shows the waveforms of FIG.

Returning to the flow of the analysis process in FIG. 4, when this envelope detection and envelope flattening are completed (step S1), various variables are initialized. That is, the octave number counter oct is set to “1”, the in-octave division number counter m is set to “0”, and the thinning counter d is set to “0” (step S2). That is, oct = 1 m = 0 d = 0.

Next, it is determined whether or not the octave number counter oct is "1" (step S3), thereby determining whether to perform downsampling. This downsampling changes the sampling frequency fs for each octave band obtained by dividing the input signal in in octave units. Specifically, the sampling frequency fs is lowered for the low frequency components of the input signal.
As a result, effects such as a reduction in the amount of processing data can be obtained. In this embodiment, the octave number counter oct
If the value exceeds “1”, downsampling is performed to halve the sampling frequency for the octave band of the octave number (oct) (step S4). That is, in FIG. 3, the original sampling frequency fso is used for the first octave, but after the second octave, the sampling frequency fs is halved every time the octave number counter oct increases. Specifically, the downsampling process reduces the sampling rate by half by thinning out one sample after passing the input signal through a low-pass filter having a cutoff frequency of π / 2.

Next, in steps S5 to S13, a thinning variable D is set for a plurality of sub-octave division bands obtained by dividing the current octave band (octave band indicated by the octave number counter oct) by the sub-octave division number M.
While thinning out the input signal data by the number (DECI-1) of sample values indicated by the ECI (that is, extracting the input signal data at every (DECI-1) on the time axis), the rotation and the filtering are performed. The obtained data is converted into its amplitude component and frequency component (pitch component), and the obtained data is stored in a corresponding bank memory (bank).
Is to be saved.

Here, the bank memory is provided corresponding to each of the divided bands within each octave of each octave band, and the bank corresponding to the divided band within the first octave, m = 0, is designated by the bank number. bank =
Bank numbers are assigned to the subbands within the octave which are arranged in the following order as 0, and the maximum number MA
X-BANK is [M * (MAX-OCT) -1]. Note that MAC-OCT is the maximum value of the octave number oct.

First, the following processing is performed in the rotation processing (step S5). That is, a complex number e for the input signal
^-jA , where A = π / 2 + (2m + 1) π / 4M],
Π / 2 + (m + 1) π / 2M to π
The band of / 2 + mπ / 2M is shifted from -π / 2M to π / 2M. Here, the octave division number counter m is 0
~ (M-1), where M is the number of divisions within an octave, and the user inputs "Melodic"
The value is determined by selecting either “Rhythmic” or “Rhythmic”.

The filtering process is realized by passing a signal whose band has been shifted by the rotation process through a low-pass filter (step S5). This low-pass filter is such that the type of the input signal specified by the user is "melodic" and therefore the number of divisions within an octave M
= 8, the cutoff frequency is π / 16 and the stop gain is −5.
A 0 dB low pass filter is used. On the other hand, when the type of the input signal specified by the user is “rhythmic”, and therefore, when the number of divisions within an octave M = 4, a low-pass filter having a cutoff frequency of π / 8 and a stop gain of −100 dB is used. Here, the low-pass filter having a cutoff frequency of “melodyc” of π / 16 and a stop gain of −50 dB has a cutoff frequency of “rhythmic” of π.
/ 8, the rejection gain is larger than that of the low-pass filter of -100 dB.

The filtered input signal is converted into an amplitude component and a pitch component (step S6).
Specifically, the real part R (t) and the imaginary part I (t) of the input signal after the rotation and filtering processing are converted into an amplitude component and a pitch component according to the following equation. That is, assuming that the input signal is R (t) + jI (t), the amplitude power (t) is obtained by power (t) = [R (t) ² + I (t) ² ] ^1/2 . On the other hand, pitch pitch which is a frequency component
(T) is: pitch (t) = [[I (t) ｛R (t−1) −R
(T) {-R (t) {I (t-1) -I (t)}] /
{R (t) ² + I (t) ² } + π / 2 + (2m + 1) π
/ 4M] * 2 ^{Determined by oct-1} .

Next, thinning of the input signal data is performed in steps S7 to S11. This means that the sample values of the input signal are thinned out by (DECI-1) pieces and stored (that is, (DCI-1)).
(Each one is stored every ECI-1)), thereby reducing the amount of data to be stored.

First, it is determined whether the value of the thinning counter d is "0" (step S7).
Amplitude power (t) and pitch pitch obtained in 6
The bank (b) corresponding to the current Taiichi using (t) as data
ank) (step S8). If it is not "0", the value of the thinning counter d is incremented by one (step S9), and the thinning counter d is set to (the thinning number DE
It is determined whether or not CI-1) has been reached (step S10). If it has reached, the thinning counter d is cleared to "0" (step S11). If (DECI-1) has not been reached, the intra-octave division number counter m is updated by one (step S12), and until the intra-octave division number counter m reaches the intra-octave division number M (step S1).
3) The processing from step S5 to step S12 is repeated. As a result, (DECI-1) input signal data are thinned out on the time axis for each divided band within an octave, and stored in the bank corresponding to each band.

When the process is completed for the number of divisions M in the octave (step S13), the octave number counter oct is updated by one, and the in-octave division number counter m is cleared to "0". (Step S1
4), octave number counter oct is octave number MA
Until X-OCT (which can be arbitrarily designated by the user, 10 in this embodiment) (step S15), the above-described steps S3 to S14 are repeated, and the processing is completed by the number of octaves MAC-OCT for the designated number. Then, this flow is finished.

Next, a process of recombining the signals divided into bands in the above-described analysis process will be described with reference to the flowcharts of FIGS. The flowcharts of FIGS. 7 and 8 show the procedure of the re-synthesizing process. In the re-synthesizing process, the time axis of the amplitude component and the pitch component of the signal divided for each band is roughly described. Is time-stretched (compressed / decompressed) by the amount specified by the user, and then re-combined by adding them together, and further extracted into the re-combined signal (flattened signal) by the previous analysis processing. As described above, the envelope is time-stretched by an amount designated by the user (that is, restored), and an output signal is generated by time-stretching the time axis of the original input signal.

When the synthesizing process is started, first, the bank number counter bank is cleared to "0" (step S3).
0). Then, the sampling ratio rate is set as rate =
The sampling frequency fs of the band / the sampling frequency fso of the system is obtained (step S31). Here, the "sampling frequency fs of the band" is the downsampled sampling frequency fs when downsampling is performed.

Next, the data counter cnt is cleared to "0" (step S32). Next, the time-stretched output signal (that is, the finally obtained output signal)
The amount of group delay generated by the downsampling and band-pass processing is put into a variable ptr indicating the position on the time axis of (step S33).

Further, pcos and psin used for frequency modulation are used.
Is set to the initial value (step S34). That is, pcos = pc = cos (pitch [bank]

[0]) psin = ps = sin (pitch [bank]

[0]). Here, the display of "pitch [bank] [n]" means that the data of the time number n is read out of the time-series pitch data stored in the bank memory of the bank number bank. Amplitude data po
The same applies to wer [bank] [n].

Next, the integer part of the variable ptr is changed to an integer part n.
And the decimal part of the variable ptr is entered in the decimal part x (step S35).

Thereafter, in order to time-stretch the time axis of the signal according to the current value of the variable ptr, processing such as data interpolation and extrapolation is first performed in steps S36 to S40. That is, it is checked whether or not the integer part n of the variable ptr is greater than “0” (step S36). If it is “0” or less, “0” is entered into the amplitude work buffer pw and “0” is entered into the frequency work buffer pt. (Step S3
7). On the other hand, if the integer part n is larger than "0", it is checked whether or not the integer part n exceeds the data length LEN (step S38). If not exceeded, a process of linearly interpolating the data is performed (step S39), and the data length LEN
Is reached, a process of extrapolating the data to a straight line is performed (step S40).

Here, the linear interpolation process of the data in step S39 is a process of generating and supplementing data between the left and right data [data of the (n + 1) th point and the nth point].
Specifically, the following formula: pw = (power [bank] [n + 1] -power
r [bank] [n]) × x + power [bank]
[N]) pt = (pitch [bank] [n + 1] -picc
h [bank] [n]) × x + pitch [bank]
[N]).

The linear extrapolation of the data in step S40 is based on the data (the last data and the second-to-last data) at the end of the data length LEN. Is predicted and compensated for. Specifically, the following calculation formula pw = (power [bank] [LEN-1] -po
wer [bank] [LEN-2]) * x + power
[Bank] [LEN-2]) pt = (pitch [bank] [LEN-1] -pi
tch [bank] [LEN-2]) * x + pitch
[Bank] [LEN-2]).

Next, a process of multiplying the frequency work buffer pt by the pitch change instruction amount pRate, that is, pt = pt * pRate is performed (step S41). This pitch change instruction amount pR
"ate" is instructed by the user according to the size of the pitch to be changed, and is also referred to as a pitch shift ratio. For example, if the pitch change instruction amount pRate is "1.0", the pitch is reproduced as it is, and if the pitch change instruction amount is "2.0", the reproduction is performed at twice the frequency, that is, one octave higher. As a result, the pitch of the input signal can be changed independently of the tempo (ie, time stretch).

Further, the variable ptr is set to the time change instruction amount t.
The process proceeds by the reciprocal of Rate, that is, ptr = ptr + rate / tRate is performed (step S42). This time change instruction amount tR
"ate" is instructed according to the size that the user wants to change the tempo (that is, time stretch), and is also referred to as a time stretch ratio. For example, if the time change instruction amount tRate is "1.0", the reproduction time (tempo) is unchanged, and if it is "2.0", the reproduction time is doubled.

Here, the sampling ratio rate in the above equation
Is the "ratio between the sampling frequency fs of the band and the sampling frequency fso of the system" obtained in step S31.
Therefore, for a band on which downsampling has been performed in the input signal analysis processing, the number of data (the number of sample values) thinned out by downsampling is calculated by the above equation using the original sampling number (and thus the sampling frequency f
so).

Thereafter, new pc and ps are obtained by the following equation: pc = pcos * cos (pt) -psin * sin (pt) ps = psin * cos (pt) + pcos * sin (pt) (step S43). And psin
Is updated by pcos = pcpsin = ps (step S44).

Thereafter, the synthesized data buffer out (cn
At t), the data of the current bank is expressed by the following expression out (cnt) = o
Add according to ut (cnt) + pw * pc (step S45). Then, the data counter cnt is incremented by one (step S46), and it is checked whether or not the updated data counter cnt has exceeded (LEN * tRate) (step S47). (LEN * tRate)
If not, the processing of steps S35 to S46 is repeated until cnt ≧ (LEN * tRate). When the data counter cnt becomes equal to or more than (LEN / tRate), next, the bank number counter bank is incremented by one (step S48), and it is checked whether or not the bank number counter bank exceeds the number of banks MAX-BANK (step S48). Step S49), MAX-BANK
If not, the processing of steps S35 to S48 is repeated until bank ≧ MAX-BANK.
By repeating this process, the synthesized data buffer out
At (cnt), a combined signal obtained by summing the signals of the respective bands can be obtained.

Finally, an envelope restoration process is performed (step S50). FIG. 9 shows a detailed procedure of this envelope restoration processing. In this envelope restoration process, the envelope extracted and stored in the input signal analysis process is compressed / expanded by the same amount by compressing / expanding the time axis of the input signal in the re-synthesis process. By applying the envelope to the synthesized signal synthesized by the re-synthesis processing, the time axis of the original input signal is compressed / expanded.

First, the data counter cnt is cleared to "0" and the variable ptr is cleared to "0", respectively (step S5).
1). Next, the integer part of the variable ptr is put into the integer part n, and the decimal part is put into the decimal part x (step S52).

Thereafter, in order to time-stretch the time axis of the envelope extracted in the analysis processing according to the current value of the variable ptr, first, processing such as interpolation and extrapolation of envelope data is performed in steps S53 to S57. . That is, it is checked whether the integer part n of the variable ptr is greater than “0” (step S53), and if it is “0” or less, “0” is entered into the work buffer work (step S54). On the other hand, if the integer part n is larger than “0”, the integer part n
Is checked to see if it exceeds the data length LEN (step S55). If not, processing for linearly interpolating the envelope data is performed (step S56), and the data length L
If it has reached EN, a process of extrapolating the envelope data to a straight line is performed (step S57).

Here, the linear interpolation process of the envelope data in step S56 is a process of generating and compensating for data between the left and right envelope data [data of the (n + 1) th point and the nth point]. Specifically, the following calculation formula work = (env [n + 1] −env [n]) × x +
env [n]) to determine an envelope value env (ptr).

The linear extrapolation of the envelope data in step S57 is based on the data at the end of the data length LEN (the last data and the second-to-last data), and further outside the data length LEN. Is a process of predicting and compensating for the data of the following equation. Specifically, the following calculation formula work = (env [LEN-1] -env [LEN-
2]) * x + env [LEN-2]) to determine the envelope value env (ptr).

Next, the variable ptr is set to the time change instruction amount t.
A process of proceeding by the reciprocal of Rate, that is, ptr = ptr + 1 / tRate is performed (step S58).

Thereafter, the synthesized data buffer out (cn
The time-stretched envelope value is multiplied by the following expression out (cnt) = out (cnt) * work to the composite signal (flattened signal) of (t) (step S59), and the composite signal is enveloped.

Then, the data counter cnt is incremented by one (step S60), and it is checked whether or not the updated data counter cnt has exceeded (LEN * tRate) (step S61). (LEN * tRat
If e) is not exceeded, the processing of steps S52 to S60 is repeated until cnt ≧ (LEN * tRate). Data counter cnt is (LEN * tRate)
When this is the case, the envelope restoration processing ends. FIG. 10 shows the waveform of the output signal restored using the envelope that has been time-stretched in this way.

In carrying out the present invention, various modifications are possible. For example, in the above-described embodiment, in the analysis process, the envelope of the waveform of the input signal is extracted, and the input signal after the envelope extraction is flattened so that the input signal is not adversely affected by the steep waveform. A division / filtering process is performed, and after re-synthesis, a pre-extracted envelope is re-attached. However, this is a more preferable embodiment and is not an essential requirement of the present invention. It is needless to say that the input signal may be subjected to band division and filtering processing without performing re-addition processing.

In the above-described embodiment, for example, the band of the input signal is first divided in octave units, and the octave band is divided at regular intervals. However, the division within each octave band is not regular. It may be performed on a cent axis.

Although the input signal is subjected to complex modulation as a method of realizing a complex BPF (band filter), the analysis filter may be subjected to complex modulation in advance, and analysis may be performed using the complex modulation. , (Fs / 2) or more may be used in which a filter is designed so as not to generate aliasing.

The filters used in this embodiment all use the minimum phase type FIR filter, but may use a linear phase FIR filter or a linear phase IIR filter.

The part for converting the real part and the imaginary part into the pitch is replaced by the pitch pit in addition to the method in the above-described embodiment.
ch (t) is the phase [pitch (t−) obtained from the previous data from the phase obtained from tan ⁻¹ [I (t) / R (t)].
1)] or tan ^-1 [｛R (t-1) * I (t) -I (t-1) * R
(T)｝ / ｛R (t-1) * R (t) -I (t-1) *
I (t)}].

In addition, the translation into this part and other parts
Parallel processing may be performed by a plurality of CPUs or DSPs.

Although a cosine value is used in the embodiment for the combining portion, it may be a sine value. Although the integration and the oscillation are performed simultaneously, the cosine or the sine may be obtained by simply integrating the pitch and then using the phase value.

[0064]

As described above, according to the present invention, a user or the like instructs a signal type (aspect) in advance, and changes the number of band divisions and the characteristics of a filter used for analysis according to the instruction. Thus, time stretching can be performed in a form suitable for the signal. As a result, it is possible to perform time stretching of an input signal for various types of musical sounds such as a rhythm system and a melody system without impairing musicality.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration of a time stretching device as one embodiment according to the present invention.

FIG. 2 is a diagram illustrating an example of an input signal type setting screen on a display device of the embodiment device.

FIG. 3 is a diagram for explaining the concept of band division in the apparatus according to the embodiment.

FIG. 4 is a flowchart (1/2) showing a processing procedure of an analysis process in the apparatus of the embodiment.

FIG. 5 is a flowchart (2/2) showing a processing procedure of an analysis process in the apparatus of the embodiment.

FIG. 6 is a flowchart illustrating a detailed processing procedure of an envelope detection process in an analysis process flow in the example apparatus.

FIG. 7 is a flowchart (1/2) illustrating a processing procedure of a re-synthesis process in the apparatus of the embodiment.

FIG. 8 is a flowchart (2/2) showing a processing procedure of a re-synthesis process in the apparatus of the embodiment.

FIG. 9 is a flowchart illustrating a detailed processing procedure of an envelope restoring process in a re-synthesis flow in the embodiment device.

FIG. 10 is a diagram illustrating an envelope detected from an input signal waveform by an envelope detection process of the embodiment device.

FIG. 11 is a diagram illustrating a signal waveform in which an envelope of an input signal signal is flattened by a flattening process of the apparatus according to the embodiment.

FIG. 12 is a diagram illustrating an output signal whose envelope has been restored by the envelope restoring process of the example device.

FIG. 13 is a diagram illustrating a configuration example of a conventional device that performs time stretching by band division and recombining.

FIG. 14 is a diagram illustrating the concept of band division in a conventional device that performs time stretching by band division and recombining.

FIG. 15 is a diagram illustrating a method of time stretching a divided band in a conventional device that performs time stretching by band division and re-synthesis.

[Explanation of symbols]

 Reference Signs List 1 CPU (central processing unit) 2 ROM (read only memory) 3 RAM (random access memory) 4 operator group 5 display 6 hard disk drive 7 CD-ROM drive 8 A / D converter 9 D / A converter

Claims (1)

[Claims] An input means for indicating an amount of time stretching of an input signal, a type indicating means for indicating a type of the input signal, an input signal is divided into a plurality of bands by using a filter, and a signal of each band is amplitude-modulated. Time-stretching means for converting the time-varying rate into a component and a frequency component based on the indicated amount indicated by the indicating means and re-synthesizing the input signal to time-stretch the input signal; And a changing means for changing the number of divisions and the characteristics used by the filter in accordance with the type of the input signal specified by the type indicating means.