JPS60129797A - Pitch controller - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION FIELD OF THE INVENTION The present invention relates to a pitch control device which makes it possible, for example, to change the pitch of an audio signal at will.

BACKGROUND TECHNOLOGY AND PROBLEMS Conventionally, the following procedure has been used to change the pinch of a signal by operating on the time axis. For example, when there is a signal as shown in Figure 1A, in order to lower the pitch (pitch down), the signal is extracted intermittently according to the rate of decrease, and the extracted signal is shown in Figure 1B. As shown, the time axis is expanded, and these are connected and output.

However, in the case of such devices, discontinuities occur at the joints of signals, and although countermeasures have been taken, such as cross-fading, these discontinuities cannot be completely eliminated, and the audibility is affected. There was a feeling of discomfort.

Additionally, the information that was not retrieved intermittently was lost, making the signal incomplete.

OBJECTS OF THE INVENTION In view of these points, the present invention proposes an excellent pitch control device that has a simple configuration and is free from signal discontinuities and dropouts.

Summary of the Invention The present invention provides means for extracting a human-generated audio signal at arbitrary time intervals and arbitrary lengths of time, and means for performing Fourier transform on each extracted frame to convert the time axis into the frequency axis. , means for adjusting the phase of this converted signal to rM, means for performing inverse Fourier transform on this phase-adjusted signal to inversely transform the frequency axis into the time axis, and converting this inversely transformed signal into a desired time A pinch control device comprising means for sequentially synthesizing each interval, expanding and contracting the time axis so that the desired time interval becomes equal to the arbitrary time interval, and outputting the desired time interval. With a simple configuration, it is possible to perform good pitch control without signal discontinuity or loss.

Embodiment In FIG. 2, an audio signal supplied to an input terminal (1) is supplied to an AD conversion circuit (2), converted into a digital signal X(to)), and stored in a buffer memo (January 3). This memory (3) has, for example, a total length of L, and inputs are sequentially shifted and stored. Every time the contents of this memory (3) are shifted by R (R<<L), the contents of the memo (January 3) are taken out in parallel to the buffer memory (4).

As a result, signals are extracted from the memory (4) at arbitrary time intervals R and for arbitrary lengths of time.

Here, R is made sufficiently smaller than L by °C, and the signals are overlamped with each other.

The signal from this memo January 4) is fed to a multiplier (5) and multiplied by a predetermined window factor ham>°C. The signal multiplied by this window coefficient is supplied to the Fourier transform circuit (6).

This converts the time axis of the signal into the frequency axis.

This converted signal is supplied to the multiplier (7), and a phase adjustment corresponding to the shift amount R in Memo January 3) is performed. This phase-adjusted signal is supplied to a processing circuit (8).

In this processing circuit (8), the signals converted to the frequency axis by Fourier transformation are stored in each memory address for each predetermined frequency band. The stored signals are sequentially read out.

The signal output from this processing circuit (8) is sent to the multiplier (9
) is supplied to "C1" which is the shift amount R at the time of output, which will be described later.

m adjustment of the phase corresponding to is performed. This phase-adjusted signal is supplied to an inverse Fourier transform circuit Ql.

This converts the frequency axis of the signal into the time axis.

This converted signal is supplied to a multiplier (11), where it is multiplied by a window coefficient f (In) corresponding to the above-mentioned window coefficient h(m). The signal multiplied by this window coefficient is stored in the buffer memory.
(12). The contents of this memory (12) are supplied in parallel to the buffer memory (13).

The entire length of this memory (13) is, for example, L, and the contents are sequentially shifted and output. In addition, 0 is written in the blank space created by the shift.1. It will be done. Then, °ζ Each time the contents of this memory (13) are shifted R゛, the contents of the memory (12) are supplied and added to the previous contents. Furthermore, the memory (13) is the memo 1 mentioned above.
Clock frequency R at the time of signal extraction on month 4).

It is driven by a clock with a frequency of ζ and □.

As a result, the time interval R is obtained from the ζ memory (13).

This time interval R.

A signal whose time axis is expanded or contracted so that R- becomes equal to R- is output.

The signal from this memory (13) is sent to the DA conversion circuit (14)
The signal is supplied to the output terminal (15), converted into an analog signal, and taken out to the output terminal (15).

Furthermore, the signal from the input terminal (11) is supplied to a bypass filter (21) and a low-pass filter (22).

These outputs are supplied to a comparison circuit (23) to compare the energies of the signals in each band. This comparison output is supplied to a window coefficient h■)+f(Ill) selection circuit (24), and a window coefficient corresponding to each case is selected.

In this device, when a signal as shown in FIG. 3A is supplied to the input terminal ζ (11), this signal is extracted for each time interval R. This extracted signal is Fourier transformed. As shown in FIG. 3B, a spectrum is formed in which the time axis is converted to the frequency axis.

This signal is stored in each memory address of the processing circuit (8).
ζ phase adjusted. This phase-adjusted signal is subjected to inverse Fourier transform to form signals sequentially shifted by time intervals R as shown in FIG. 5A. These signals are sequentially supplied to the ζ mesori (13) through the memory (12) and are added.

For example, if the pitch is to be lowered (pinch-down), the frequency of the shift clock of the memory (13) is set lower than the frequency of the 'clock of the memory (4) at the time of signal extraction. On the other hand, the signal is supplied from the ζ memory (12) at every time interval R.

If the frequency of the Ro Zunawara clock is set to □ (R'< R) and the shift amount in the time interval R becomes Ro RX --= R'', then this shift 1
- Ia When addition is performed for each R', this signal becomes as shown in FIG. 4B, and the frequency band R' of this signal remains the same, but the time is shortened to □. Therefore, by extracting this signal with the second clock R', the frequency band becomes □ as shown in Figure 4C,
A signal that does not change over time can be obtained.

For example, when Ro=□-, the time remains the same but the frequency band becomes 10, resulting in a signal that is one octave lower than the pinch.

The same is true when raising the pitch (up yr), simply by setting R'>R.

As a result, the added signal becomes a signal R' with the frequency R' waveband remaining unchanged and the time extended to -, as shown in FIG. 5B. Therefore, by extracting this signal using the □ clock, the frequency R' number band becomes - as shown in FIG. 5C, and a signal that does not change in time can be obtained.

For example, if R'=2R, the time remains the same,
The frequency band is doubled, resulting in a signal that is one octave higher than the pinch.

Although the waveforms in FIGS. 3-1 to 5 are shown in analog form, they are actually processed as digital values.

Furthermore, in the device described above, the window coefficient h(匍, r(t)
) are related as follows. That is, the signal X(fn
) for x (m) h (SR-m) X (Ill) However,
S becomes a lazy integer, and by Fourier transforming it, we get X2 (SR+ ω) −Σ h (SR−m) x<
to) a−”■g F+ −00 Furthermore, perform inverse Fourier transform to S−1■ Since it is sufficient that this is equal to x <m>, Σ r (m,
rrt -3R) h (SR-m) "1. Vm
It is sufficient if S=-00.

- As described above, detect the spectral shape of the input signal and select the "C window coefficient h (g) + fQIl)
For example, if the low frequency component is smaller, ha
n+P1 f +m)-0,5-0,5cos (2yr n/N
-1) n=0,...N-] When the low frequency component is larger, ha+o=0.54-0.46 cos (2yr n
/N −1) n = Q, . . . N−1 fan+=2π/Σb j The sound quality can be improved.

Further, in the above-mentioned device, the phase δIi in the multiplier (9)
, 1 integral is done as follows.

First, the spectrum after Fourier transformation at time SR is X (or, ωk), and its real part is XR(SR, ωk) t, 11. r! B is Xl (SR, ωk) The 7 principal values of the phase are P (S11.ωk) However, -π≦P (SR, ωk) π and the phase continuous along time S is p (SR, ωk ) However, −■<p(SR, ωk) and ψ. At this time, phase continuity and phase transformation are performed as follows.

1) In the case of Sho (al First, by Fourier transform, X (SR, ω
k).

(Add P (51?, ωk) to bl change.

Here, when the sign of XR (Sll, ork) and Xl (SR, ωk) is (+, 10) or (+, -), P (SR
, ωk ), = tan-” (X ■ (31 ma,
ωk )/XR(SR, ωk)) When the sign is (-, +), P (Sll, ωk) = tan-1(Xr (S
R+ ωk)/XR(SR, ωk)) 1π When the sign is (-,-), °- P (SR, ork) = tan-” (Xr (S
R+(tok)/XR(SR,ωk))−π−.

(C1 further IP(SR, ωk> −P((S-1) R, ωk)
l<ε - However, it is determined whether ε is a constant.

(dl And if this stops, p (SR, ωk) = p((S-1)R, ωk) −
1-P (SR, ωk)-P((S-1) It, ωk
).

(d') Also, (when C1 is not constant, first, when the sign of the absolute value is (-), p (SR, ωk) - p ((S-1)R, ωk) IP
(SR, ωk)-P((S-1) R, ωk)+2π 1 when the sign is (+)) (SR, ωk)-p((S-1)R, ωk) I
P (SR, ωk)-P((S-1) R, ωk)-2
Let it be π.

Due to buried soil (the phase is made continuous. Furthermore, the shift at the time of synthesis described above), when the history of Et is changed,
In code and decoding, interference between hands is prevented (in order to prevent interference between hands, p (Sll, ωk) −p (SR, ωk) −-
−. This prevents noise from occurring due to phase discontinuity.

This is how pitch control is performed,
According to this device, the signals converted to the frequency axis by Fourier transform are synthesized after adjusting the phase δI +, l, so that extremely high quality signals can be obtained, and there are no discontinuities or omissions in the signal. A good pitch control 1 call with no etc. is made.

Furthermore, by setting Ro arbitrarily, the pinch can be varied over an extremely wide range, and the deterioration in sound quality during that time is extremely small.

Effects of the Invention According to the present invention, it has become possible to perform good pitch-1 control without signal discontinuity or dropout with a simple configuration.

[Brief explanation of the drawing]

FIG. 1 is a diagram for explaining a conventional device, FIG. 2-I is a configuration diagram of an example of the present invention, and FIGS. (]) is an input terminal, (2) is an AD conversion circuit, (3), (
4), (12), (13) are sofa memories, (5),
(7), (9), (11) are multipliers, (6) is a Fourier transform circuit, (8) is a processing circuit, α0) is an inverse Fourier transform circuit, (14) is ■] Δ transform times 1 to 3 , (15) are output terminals. 1. Skewered beef surface 1981 Patent Application No. 238692 2. Title of the invention Pitch control I: 1-13, di '+l-
4. Relationship with 15 Patent applicants Location: Kitashin J1, Tokyo Parts Store, 7-35
υ- Name (21B) 1 representative representative of Sony Corporation Person f)゛ Nori /, jl+4, agent (j, number of inventions increased by 31 to sleeve 11-+11 Amend the scope of claims as shown in the attached sheet) (2) In the specification, the phrase ``input pitch control device'' from page 2, line 20 to page 3, line 10 should be corrected as follows. means for multiplying and extracting each signal by a window function having a length and coefficient greater than the time limited by the time interval, and means for performing Fourier transform on each extracted signal to convert the time axis to the frequency axis; means for adjusting the phase of the converted signal; means for performing inverse Fourier transform on the phase-adjusted signal to inversely transform the frequency axis into the time axis; means for sequentially synthesizing each desired time interval by multiplying it by a window function having a time length and a coefficient, and expanding and contracting the time axis so that the desired time interval is equal to the arbitrary time interval and outputting the result. (3) Same, page 3, line 15 to page 4, line 15 [Figure 2...
Supplied. ” is corrected as follows. [In FIG. 2, an audio signal is supplied to the input terminal 11), which has been converted into an electrical signal by a microphone or the like and passed through a low-pass filter with a cut-off frequency of 3.2 kHz. This input audio signal is sequentially converted into digital data of 12 bits per word at the rate of this clock pulse by the AD converter (2) of 12 bits per word driven by a 5.4 kHz (periodic 158 μs) conversion clock. converted. The AD converter (2) is a 256a# consisting of 1 word of 12 bits driven by a 6.4kHz clock.
The shift register (3) is connected to the shift register (3) of t, and every time one pulse of the driving clock is supplied to the shift register (3), the shift register (3) outputs one word, right in FIG. The words "J left, I right" are used to mean left and right in Figure 2), and the output data of the AD converter (2) is shifted one word to the left of the shift register (3). Then, it enters the shift register (3). In other words, the shift register (3) has an AD converter (2)
The shift register (3) inputs a series of 256 words of digital data generated by the AD converter (2).
Go is shifted to the right and its contents are updated. Before explaining the specific flow of the signals below (4) in Figure 2, let us first explain the short-time Fourier analysis.
ζ Let me state some general matters. For example, if we consider the audio signal "Aiueo", the time when the sound "1a" is being emitted and the time when the sound "i" is being emitted differ depending on the mouth of the person making the sound. The shape of the vocal tract is different. In other words, the audio signal ``Aiueo'' is a signal emitted from a physical entity whose characteristics change over time, and cannot be regarded as a stationary signal. In this way, the characteristics of the physical entity that emits audio signals, music signals, etc. change over time.
In general, it cannot be regarded as a stationary signal, and it is impossible to directly apply Fourier spectrum analysis to stationary signals. However, in the example above, regarding 1゛aiueo'', 1゛a'', ``i'', ``u'',
[picture". The shapes of the human mouth and vocal tract remain fairly constant during the time each sound of 1"o" is uttered, and if the signal is limited to that time, it can be regarded as a steady signal. Therefore, if we limit the region to be Fourier transformed to a time interval that can be considered stationary, perform Fourier transform, and use the Fourier spectrum obtained by updating that interval one after another, it is possible to Fourier analysis becomes possible for audio signals and music signals that are stationary in the interval. Such Fourier analysis is called short-time Fourier analysis. Let's explain further using mathematical formulas. The input signal x (tl) is the data string obtained by sampling (xlml) (m
= (L 1+ 2+...), the above-mentioned matter is a subsequence of data 1 x (m+S
R) lm=0+L...; 5=OA+-...-(
For a variable m of (R, M are certain integer constants), a finite subsequence (x(m+SR)) m=0.1. ..., the window coefficient (h (
-rn) ) (m=0.1...,M-1), performs a discrete Fourier transform on the variable m, and obtains the short-time Fourier spectrum X (SR,k) (
S=0,1. ...-, M-1; k=0.1.2.・
...? 1-1). tr As is clear from FIG. 6, when analyzed, R is the update amount of the section, and has the following constraints. From equation (A), if W m + SR= f, -(B) Window coefficient (h (-m) ) (m=0.1.2...
・, M-1) is expanded to -Q~10ψ for m, then 2π In other words, X(SR,k) is the first
For the th variable SR, the data string (h (m) ) is convolved with the data string X (S, k) (S=0.1
．． 2. . It can be interpreted as Therefore, as shown by the sampling theorem, the update amount R'X 1 of the interval to be analyzed must be within the band for the first variable m. (X (+n,k)) (m-0,1,2,...=
The band Ill of ) depends on (m-(LL2+...), but its upper limit is the low-pass characteristic of a linear digital system with an impulse response (h (m) >) in the figure. Since it is suppressed, in the band for the if-th variable m of ] ≧2 X ((h (m) ) (m=0.1, 2.=
) = CD > In other words, R must be −(E). For example, if M=256. (h (m) ) is the Hamming window coefficient, the window coefficient h (m) = 0. 54
−0.46 cos(2πm/255) (m=0.1
．． ..., 255'), then (h
(m) ) (m=o, 1.2...-, 255) in the low-pass band is attenuated to about 42 dB, so from the relationship in the above equation, R must be R≦□=64. Must be. In Figure 2, (41, (51, +61, (71
The short-time Fourier transform described above is performed. M
= 256, Hamming window coefficient h(
m) -0,54-0,46Xcos (2πm/255
) (m=(LL2゜..., 255), R
= 64 and °ζ. As is clear from the above example, R-
64 satisfies formula (E). The details will be explained below. The contents of the shift register (3), which consists of a 12-bit, 2561-bit 1i crystal, is one pulse of the clock obtained by dividing the drive clock of the AD converter (2) by 64 (i.e., 64X
(Driving clock period of AD conversion (2), approximately 158μ5e
c) (seconds)) also 1ii12 bits, 25
It is latched into a shift register (4) consisting of six words. The latched 256 words of data are transferred to the shift register (4).
At the clock timing of 8 MIIz (period: 125 n 5 ec) supplied to the multiplier (5 ) is sent to one of the input terminals. On the other hand, at the timing of this same clock, the Hamming window coefficient h (m)-
0,54-0,46cos (2mm/255) (
m = 0 + 1 + ..., 255), each - word, m = 0
．． 1.2. ..., and the product of these two inputs is set to the output terminal of the multiplier (5) 100 n seconds after the input data is set as the output of the multiplier (51). The 23-bit output result of the multiplier (5) is
F F T (Fast Fourier Tr
ansform) is sent to the g converter (6). FF
T converter (6) When the data of 23 bits per word sent in this way becomes 256 words, perform FFT on this data of 23 bits and 256 bits, and both the real part and the imaginary part consist of 16 bits. , generates 256 words of complex data. Now, the 256 times the input data to the FFT converter (6) is (y(m)) (m=0.1.-..., 255)
Output data (Y (k) (k=o, 1, 2...
, 255), then from the definition of FFT, n On the other hand, this human data (y (m) ) (m=0+1
From equation (A), the short-time Fourier spectrum of
255) and (X (64s, k ) > (k=0.
1.2. Old..., 255) has the following relationship: 2π (k=0.1.2..., 255>-(Summer 1). Therefore, the FFT odd #!!4+63
2π The short-time Fourier spectrum of the input data X(m) is obtained. This is done by the multiplier (7). In other words, the 256 words of complex data generated by the FFT converter (6), consisting of 16 bits for both the real and imaginary parts, are as follows:
At the timing of a clock with a period of 125 n sec, one of the multipliers (7) has two complex data input terminals with both real and imaginary parts of 16 bits, and one output terminal with both real and imaginary parts of 16 bits. is sent to the input terminal of On the other hand, at the timing of this same clock, the above-mentioned coefficients, 2π 2, ..., 255) prepared in advance are sent to the other input terminal of the multiplier (7) by -1. In addition, the product of these two inputs is set as the output of the multiplier (7), and the human data is set as "(100 n sec from
Later, it is sent to the output terminal of the multiplier (7). This output result is sent word by word to the spectrum modification circuit (8), 256 words in total, at the timing of the clock that sends the input data to the multiplier (7). (4) Same, ff14Ji (line 16, page 7, line 1, 14th
In lines 14 and 15 of pages, the words "processing circuit" are corrected to "spectrum modification circuit." (5) Same, page 4, line 20 to page 6, line 7 [This processing circuit... is taken out. ” is corrected as follows. [256 words of complex data in which the string is transformed by the spectrum transformation circuit (8) and consists of 16 bits in both the real part and the imaginary ff1l are +91. all, (11), (
12), (13). The signal is converted into a time domain signal in (14). Before specifically explaining the flow of (9) to (14), (9)
) to (14), the general relationships will be described below. As mentioned earlier, the transformed short-time Fourier spectrum x (SR', l (S=0.1.2...
・;k=0.1,2. ..., M-1) is the short-time Fourier spectrum X (S, k) (S=0.
1.2. ...; ni = O, l. 2, . . . , M-1) is resampled for the first variable S every R'-1 data. Therefore, the transformed short-time Fourier spectrum 'zo(SR', k) (S=O, 1, 2,...1k
=0°1.2. ..., 卜1), to create a time domain signal, X (SR', k) (S=0.
1.2. ...1k=0.1,2. ..., Tol)
By interpolating X (S, k) (S-0, 1, 2,...
...i k-0,1,2,..., M-1), and create X (S, k) (S=0.1,2..., k
= 0.1.2. .... M-1) can be subjected to inverse discrete Fourier transform, that is, regarding the first variable of X (SR', k),
For each, data X(S,k) is created by inserting R'-1 0s between adjacent data, and M(hard data(f(m))(m=0.
1. −..., Mfl) as the impulse response, and then
6 From the definition of the formula
Perform inverse discrete Fourier transform to obtain output signal (y(S))(S
-0, 1, 2,...) is obtained. If we also write this as a formula,
It will look like this: -(1) R' -R and when the spectrum is not manipulated, the input signal must become the output signal as it is. For that purpose, from the above formula, V (S) −x (S);
) ・ x <i>lIπ−■ −≠−■ By the way, since , let jl=s−pM (p: variable) −(J) Therefore, (h (m) ) and (f (m) ) It is necessary that for all S, °ζ, I11+ −■ −(K). Now, from equation (1), if we write -) (f (m) ) is m = O+1+2+" ・
: + Since it is not 0 only at M-1, f (S-*R') ・x (d', S) is S=m
R', mR' +l, group, d'+M-1 (m=S
Only the part −w+R' + n=(M+old・+M-1) is not 0, so as R', r-R'=M
(r: positive integer constant) and M are chosen to divide evenly, (m-1)R'+M≦S≦mR'+M-1 (m=0.1
,2. ), and (y(S) ) (S=0.1.2
．． ...) is found one after another.In addition, in order to use FFT when calculating x (+mR,S), the relationship of equation (11) is established between the FFT-transformed data and the short-time Fourier spectrum data. Since some 2π j−R'kS, o M (k−0, l
,..., M-1i S-0,1,2,...;
R', an integer constant) and then perform FFT. This will be explained in detail in Figure 2. Note that in the following description, R'=64. The real part, transformed by the spectrum transformation circuit (8),
256-word short-time to Fourier spectrum X (64S, k) (k=0.1,
2゜..., 255) has a period of 125 n se
At the clock timing of e, k = 0.1, 2.・
A multiplier with two complex data input terminals each consisting of 16 bits for both the real and imaginary parts, and one output terminal consisting of 16 bits for both the real and imaginary parts. On the other hand, the coefficients and coefficients mentioned above are prepared in advance at the same clock timing.
, 255) Crab = o, l, 2. ...(7) Sequentially, one word at a time, is sent to the other input terminal of the multiplier (9),
The product of these two inputs is the output of the multiplier (9), and the input data is multiplied by 1i! 19), but 10
After 0 n sec, it is set to the output terminal of the multiplier (9). This output result is sent one word at a time, a total of 256 words, to the inverse FFT converter Ql at the timing of the clock that sends the input data to the multiplier (9). The inverse FFT transformer pA (with it) performs an inverse FFT on this data when the data sent in this way, consisting of 16 bits for both the real and imaginary parts, reaches 256, and converts the data into 256 words, each word consisting of 16 bits. Generate time domain data. These 256 words of data, each word consisting of 16 bits, are generated at a clock timing of 125 nsec.
It is fed into one input terminal of a multiplier (11) having two input terminals of 16 bits and one output terminal of 16 bits. On the other hand, at the timing of this same clock, the multiplier (11
), the above-mentioned relational expression (K)2π 0.46cos (-g) ), m=0. L2.
・−・・+255 )55 is m=0.1, 2. The product of these two inputs is inputted one word at a time in the order of .
After e, it is sent to the output terminal of the multiplier (11). This output result is 256 words in total, every single word is generated at the timing of the clock that sends the input data to the multiplier (11).
It is sent to the shift register (12). The shift register (12) has 16 bits per word, 25
It consists of 6 words and sends the multiplication result of the multiplier (11). It is driven by an intermittent clock with a period of 125 n see, and is shifted to the right by 11 times each time the multiplication result is transferred from the multiplier (11) by one word. In this way, the shift register (12) has a 256-word multiplier (11
), the shift register (12) becomes
25 of shift register (12) becomes shift prohibited state.
6it! ;6<, la! Each of the shift registers (13) consisting of 16 hits and 256g8 are added for each corresponding word, and the addition result is put into the corresponding word of each of the shift registers (13). This shift register (13) has 6.4kl! which drives the AD converter (2)! A clock of z is supplied, and when the above addition is completed, the shift register (13) performs 1 every l pulse of this clock of 4kllz.
The word is shifted to the right and sent to the 16-bit AD converter (14).
■Data is sent. On the other hand, due to this shift, 1M data having a value of 0 is input into the shift register (13) from the left. Thus register (13) to shift 1
performs the shift R' = 64 times and converts the 64 output data to D
Transfer to A converter. 16 bits DA conversion 1 (14) is 6.4kHz
The 1ixe bit data sent at the timing of J7 is sequentially converted into an analog voltage value, and the output terminal (1
5) Output. (6) Same, page 7, line 2, 1 ゛°C phase...signal.''
Correct the statement to read, ``This signal is.'' (7) Same page, line 12, Suesha, page 8, lines 1, 2, 10
(8) Same, page 9, line 9, after “for” [window coefficient h
Multiply by (m) and add ". (9) Same page, line 12 (after converting, add 1゛spectrum Same, page 10, lines 6-13 “h(m)=1...
can. ” is corrected as follows. rh (m) = 1 f (m) = 0.5-0.5 cos (2πm/
(N l )) m=0,...N-1 When the low frequency component is larger, h (m) =0.54-0.46 cos (2πm
/(N-1))rn=0,...N-1 f(m)=R'/Σh'', the sound quality can be improved. Note that h (m)=0 corresponds to doing nothing with the multiplier (5). "(12) Same, page 11, lines 5-8, 1 value of F phase, . . . phase" should be corrected as follows. ``In this case, if one value of the phase is P (SS1.ωk), P (SR, ωk) takes a value of 1 ≤ P (SR, ωk) > π, and the phase becomes discontinuous. Therefore, the phase after removing this discontinuity is corrected to be one discontinuity after "(13), page 12, line 10 [further"]. (15) In the drawings, Figure 5 will be corrected as shown in the attached sheet. (16) Figures 6 and 7 will be added to the drawings as shown in the attached sheet. What is claimed is: a means for multiplying and extracting an input audio signal for each arbitrary time interval by a window function having a length and a coefficient longer than the time limited by the time interval, and for each extracted signal. means for converting the time axis into the frequency axis by Fourier transform; means for adjusting the phase of this converted signal; and means for inversely converting the frequency axis into the time axis by performing inverse Fourier transform on the phase-adjusted signal. Then, this inversely transformed signal is multiplied by a window function having a time length and a coefficient defined by the above window function to sequentially synthesize each desired time interval, and this desired time interval is equal to the above arbitrary time interval. A pinch control device comprising a means for expanding and contracting a time axis and outputting the same.

Claims (1)

[Claims] A means for extracting an input audio signal at an arbitrary time interval and an arbitrary length of time, a means for Fourier transforming each extracted frame to convert the time axis into a frequency axis, and the converted signal means for adjusting the phase of the signal, means for performing inverse Fourier transform on the phase-adjusted signal to inversely transform the frequency axis into the time axis, and sequentially synthesizing the inversely transformed signals at each desired time interval. a pitch control device 6' comprising means for expanding and contracting the time axis so that the desired time interval is equal to the arbitrary time interval and outputting the result;