JP3254829B2 - Method and apparatus for time-based extension reading of digital audio signal - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for reading out a digital audio signal on a time axis and a device for reading out a digital audio signal on a time axis so that the pitch of the audio signal can be reduced in the state of a digital signal.

[0002]

2. Description of the Related Art It is well known that sound signals have been conventionally edited for various purposes, and analog signals which have been widely used as sound signal editing apparatuses have been known. In addition to audio signal editing devices in the form of digital signals, audio signal editing devices in the form of digital signals are also used. The audio signal editing device described above is, for example, a recording and reproduction function, fade-in,
It is configured to have fade-out, cross-fade, editing and playback functions based on the edit list, and many other functions. That is, the pitch of the acoustic signal has been conventionally reduced.

[0003]

Conventionally, as a means for reducing the pitch of a digital audio signal, there is a method of reducing the sampling frequency to an analog signal. After converting to an analog signal, an anti-alias filter is used to remove aliasing noise. In this conventional method, when returning to a digital signal having a sampling period before the pitch is reduced after the pitch is reduced, an analog-to-digital converter for resampling is required, and the digital signal is converted to an analog signal. There is a drawback that high-precision conversion cannot be performed because deterioration of the signal for returning the signal is inevitable. The above disadvantages can be solved if the signal processing is performed in the state of the digital signal, but if the signal processing in the state of the digital signal simply performs zero interpolation or hold interpolation of data, aliasing noise occurs. For this reason, when the pitch is reduced while the digital signal remains unchanged, a signal processing method that can remove aliasing noise well is required.

[0004] In order to solve the above-mentioned problem, the present applicant has proposed that "the pitch of an acoustic signal is 1 in the form of a digital signal.
/ N (where n is an integer of 2 or more) is a digital signal processing method in which a sample value sequence of a digital signal whose pitch is to be reduced is multiplied by m (where m is 2
The oversampling filter for performing the oversampling of the above integer) and the successive sample values of the oversampled digital signal are repeatedly arranged for every N consecutive oversampling periods and arranged on the time axis. The sampling section of the digital signal whose pitch is to be reduced is expanded N times, and substantial hold interpolation is performed at the time position where the N times expanded sampling section is divided by MN. "A digital signal processing method comprising means for obtaining digital data of a sample value series in a state and an anti-alias filter for removing aliasing components in the digital data"
A patent application was filed on March 12 (see Japanese Patent Application No. 4-145013), and a signal processing method which does not cause the above problem is disclosed in, for example, a part of US Pat. No. 5,111,727. As described above, a prior art in which an interpolation signal is generated by linear approximation is also known.

However, in the above-described signal processing method,
In order to change the cutoff frequency in response to the change in the pitch of the sound signal, an anti-alias filter with a variable cutoff frequency is required, but when the filter characteristics are suddenly changed when the pitch is suddenly changed. The switching backlash occurs due to the switching noise occurring or the time until the filter characteristic is switched in response to the input at the time of the input for changing the pitch and the time at which the filter characteristic is switched. When such a signal processing method is applied to, for example, a jog device or a shuttle device for an acoustic signal, there is a problem that it is difficult to determine an edit point.

In the above-described conventional technique based on linear approximation in the signal processing method, since it can be implemented without using a filter, the problem in the above-mentioned signal processing method does not occur in the signal processing method. FIG. 6 shows a frequency response characteristic when a sampling frequency of 48 KHz and a sine wave signal of 1 KHz are converted into a signal having a pitch of 0.6 times by the signal processing method shown in FIG. Depending on processing method, sampling frequency is 48KH
The figure shows the frequency response characteristics when a 10 KHz sine wave signal is converted to a signal having a 0.6 times pitch in z.
6 and 7, a good approximation is obtained as shown in FIG. 6 for a sine wave signal having a low frequency of approximately 1/20 or less of the sampling frequency of 48 KHz. As shown in FIG. 7, a high-level subharmonic component appears in a sine wave signal having a high frequency close to half the sampling period. Since the high-level subharmonic component generated at a high frequency close to half the sampling period as described above causes a remarkably large distortion, an interpolated signal with little distortion is generated over a wide frequency range. This signal processing method cannot be applied to a jog device or a shuttle device for an acoustic signal that needs to be controlled.

[0007]

SUMMARY OF THE INVENTION According to the present invention, a pitch of an acoustic signal is reduced to 1 / N (where N is any positive or negative and any number of 1 or more) in a digital signal state. A method for reading out a time axis of a digital audio signal, comprising:
For each successive sampling period, a predetermined predetermined sequence that is continuous in the sample value sequence of the digital audio signal corresponding to the variation of the amplitude on the time axis of the audio signal targeted for pitch reduction. (Where M is a natural number of 4 or more) corresponding to an algebraic curve containing a sample value on the curve.
1) Interpolation according to the (M-1) degree polynomial by the coefficient of M terms obtained based on the M sample values in order to approximate the set of M sample values by the following polynomial A signal is obtained and output, and the signal ΔNk (where k is a natural number of 1 or more) indicating the rate of decrease 1 / N of the pitch of the acoustic signal given for each successive sampling cycle is accumulated. The set of M sample values used to obtain the interpolation signal in each successive sampling period when the magnitude of the accumulated value obtained is less than or less than 1 or less than 1 is not changed. When the magnitude of the accumulated value of the signal ΔNk exceeds 1, a value obtained by subtracting the magnitude 1 from the accumulated value is calculated by the following 1
It is used as the initial value of the accumulated value of the signal ΔNk in the sampling period, and instead of the set of M sample values used so far, the whole is shifted toward the new sample value by one sampling period. The present invention is to provide a time-base expansion reading method of a digital audio signal in which an interpolation signal according to the (M-1) -th order polynomial described above is obtained and output using a set of M sample values obtained as described above. . Further, there is provided a digital audio signal time-base expansion reading apparatus for reducing the pitch of an audio signal to 1 / N in the state of a digital signal (where N is any positive or negative number and an arbitrary number of 1 or more). And each one in turn
For each sampling cycle, a predetermined number M (continuously) in the sample value sequence of the digital audio signal corresponding to the variation of the amplitude on the time axis of the audio signal targeted for pitch reduction. However, in order to approximate the above set of M sample values by an (M-1) -order polynomial corresponding to an algebraic curve including a sample value of (M is a natural number of 4 or more) on the curve, Means for obtaining coefficient values of M terms on the basis of the sample values of the above, means for obtaining and outputting an interpolation signal according to the (M-1) -th order polynomial using the above-mentioned coefficient values, A signal ΔNk indicating the rate of decrease 1 / N of the pitch of the acoustic signal given for each sampling cycle ( where k is 1 or more)
Means for accumulating natural numbers) , and a set of M sample values used to obtain an interpolation signal in each successive sampling period when the magnitude of the accumulated value is 1 or less or less than 1 Is not changed, and when the magnitude of the accumulated value of the signal ΔNk exceeds 1, a value obtained by subtracting the magnitude 1 from the accumulated value is used as the signal ΔNk in the next one sampling period.
Is used as the initial value of the accumulated value of M, and instead of the set of M sample values used so far, M samples obtained by shifting the whole toward the new sample values by one sampling period are used. Means for using a set of sample values to read out a digital audio signal on a time axis.

[0008]

A predetermined number M (where M is a predetermined number) continuous in the sample value sequence of the digital audio signal corresponding to the variation of the amplitude on the time axis of the audio signal targeted for pitch reduction. (M is a natural number of 4 or more) on the curve and an algebraic curve corresponding to the (M-1) -order polynomial corresponding to the set of M samples to approximate the set of M samples. The coefficients of the M terms are calculated based on the values. An interpolation signal according to the (M-1) -order polynomial is obtained and output using the coefficients of the M terms described above. Further, the signal ΔNk indicating the rate of decrease 1 / N of the pitch of the acoustic signal given in each of the successive sampling cycles is accumulated, and the magnitude of the accumulated value is 1 or less or less than 1. No change is made to the set of M sample values used to obtain the interpolated signal in each successive sampling period in the case. When the magnitude of the accumulated value of the signal ΔNk exceeds 1, the value obtained by subtracting the magnitude 1 from the accumulated value is used as the initial value of the accumulated value of the signal ΔNk in the next one sampling cycle. Also, instead of the set of M sample values used so far, a set of M sample values obtained by shifting the whole to a new sample value by one sampling period is used. An interpolation signal according to the (M-1) -order polynomial described above is obtained and output. In this way,
It is possible to read out the digital audio signal by extending the time axis so that the pitch of the audio signal is reduced to 1 / N in the state of the digital signal (where N is any positive or negative and any number greater than 1). Easy.

[0009]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing an embodiment of a method and an apparatus for expanding and reading a digital audio signal on a time axis according to the present invention. FIG. 1 is a block diagram showing a schematic configuration of an apparatus to which a time axis extension reading method of a digital audio signal according to the present invention is applied. FIG. 2 is a diagram showing a time axis extension reading method of a digital audio signal according to the present invention. FIG. 3 is a block diagram when the present invention is applied to an apparatus, FIG. 3 is a waveform diagram for explaining a configuration principle and an operation principle of a digital audio signal time axis extension reading method of the present invention, and FIG. 4 is a time axis of the digital audio signal of the present invention. FIG. 5 is a diagram showing an output characteristic example of a device to which the decompression read method is applied, and FIG. 5 is a diagram showing an example of a change characteristic on the time axis of a signal ΔNk indicating a reduction rate 1 / N of a pitch of an audio signal; FIG. 4 is a diagram illustrating an example of output characteristics by a linear approximation method.

In the block diagram of FIG. 1 showing a schematic configuration of an apparatus to which the method of extending and reading the time axis of a digital audio signal according to the present invention is applied, each component indicated by reference numerals 3 and 4 is a part of the present invention. The signal processing unit 3 and the buffer memory 4 in the block diagram of FIG. 2 showing a configuration example when the time axis extension reading method of the digital audio signal of the invention is applied to a jog device or a shuttle device for an audio signal. The input / output terminals 3a, 3b, 3c in the signal processing unit 3 shown in FIG. 1 and the input / output terminals 4a, 4b, 4c in the buffer memory 4 are shown in FIG. Corresponding to the input / output terminals 3a, 3b, 3c of the signal processing unit 3 and the input / output terminals 4a, 4b, 4c of the buffer memory 4. The configuration of the signal processing unit 3 shown in FIG. 1 is only a part of the configuration of the signal processing unit 3 shown in FIG.

The input terminal 3a of the signal processing unit in FIG.
In the state of the digital signal, data of a digital audio signal targeted to reduce the pitch of the audio signal to 1 / N (where N is a positive or negative and an arbitrary number of 1 or more) is input. In FIG. 1, it is assumed that predetermined data read from the buffer memory 4 is given to the input terminal 3a of the signal processing unit from the output terminal 4a of the buffer memory 4. Further, a signal ΔNk indicating the rate of decrease 1 / N of the pitch of the acoustic signal is supplied from an external circuit to the input terminal 3c of the signal processing unit. By the way, after being read from the buffer memory 4 as described above, the coefficient calculation unit 31 in the signal processing unit 3 via the input terminal 3a of the signal processing unit 3
The digital audio signal data targeted for pitch reduction given to
A predetermined number M (where M is 4 or more) continuous in the sample value sequence of the digital audio signal corresponding to the variation of the amplitude on the time axis of the audio signal targeted for pitch reduction Is a natural number).

Now, white circles De, Dd, D in FIG.
c, Db, Da, and Dp are successive successive values in a sample value sequence of a digital audio signal having a digital value corresponding to an amplitude change mode on the time axis of the audio signal targeted for pitch reduction. FIG. 4 schematically shows an arrangement of data. And the above-mentioned white circles De, Dd, Dc, D
A curve shown by a solid line connecting b, Da, and Dp indicates a change in the amplitude on the time axis of the sound signal to be pitch-decreased, and a dashed-dotted line on the right in FIG. Curves and four (when M = 4) white circles Dd, Dc, Db, D
The curves shown by connecting the solid-line curve connecting a to the left dash-dot line curve in FIG. 3 are the four (in the case of M = 4) white circles Dd, Dc, and Db. , Da are algebraic curves including a curve shown by a solid line on the curve.

In other words, the algebraic curve described above corresponds to the digital audio signal having a digital value corresponding to the change of the amplitude on the time axis of the audio signal targeted for pitch reduction shown in FIG. Among consecutive data targeted for pitch reduction indicated by white circles De, Dd, Dc, Db, Da, and Dp which are continuous in the sample value series, four consecutive (M = 4 ), White circles Dd, D
It is a curve that approximates the curve shown by the solid line connecting c, Db, and Da. In the method for reading out a digital audio signal on a time axis in the time axis according to the present invention, the sample value sequence of the digital audio signal corresponding to the change mode of the amplitude on the time axis of the audio signal targeted for pitch reduction is described. Sample values (white circles De, Dd, Dc, Db, Da, D
.. p), a predetermined number M (where M is a natural number of 4 or more) of sample values corresponding to an algebraic curve containing the sample values on the curve. A set of sample values is approximated, and an interpolation signal according to the (M-1) -order polynomial is obtained and output by the coefficients of the M terms obtained based on the M sample values. . For example, the algebraic curve illustrated in FIG. 3 is composed of four consecutive white circles Dd, Dc, and D (when M = 4) as described above.
Since it is a curve approximating a curve shown by a solid line connecting b and Da, this algebraic curve is represented by a cubic polynomial y (t ') represented by the following equation (1).

[0014]

(Equation 1)

Thus, for example, continuous sample values (white circles De) in a sample value sequence of a digital audio signal corresponding to the variation of the amplitude on the time axis of the audio signal targeted for pitch reduction. , Dd, Dc, Db, Da, D
p) from the buffer memory 4 at each sampling period as a continuous predetermined number M of sample values.
When M = 4 as illustrated in the example, the coefficient calculation unit 31 in the signal processing unit 3 outputs the four sample values (for example, white circles Dd, Dc, and D) supplied thereto.
b, Da), the coefficients a, b, and 4 of the four terms in the third-order polynomial y (t ′) shown in the above equation (1)
Find c and d. Now, the four sample values supplied to the coefficient calculation unit 31 in the signal processing unit 3 are, for example, white circles D
If d, Dc, Db, and Da, the sample value Db at the center of the time axis shown in FIG.
The values of y (t) for the sample value Da before the sample value Dc and the sample values Dc and Dd after the sample value Db are y (0) and the sample value Db in FIG. Da is y (-
1) Since the sample value Dc is y (+1) and the sample value Dd is y (+2), in the third-order polynomial y (t ′) shown by the equation (1) in this case, Each coefficient a of the four terms,
b, c, and d are represented by the following equations (2) to (5).

A = y (+2) / 6−y (+1) / 2 + y (0) / 2−y (−1) / 6 (2) b = −y (0) + y (−1) / 2 + y (+1) / 2 (3) c = −y (+2) / 6 + y (+1) / 2−y (0) / 2−y (−1) / 3 (4) d = y ( 0)... (5) The coefficients a, b, c, and d are not adaptively determined by the coefficient calculation unit 31 in the signal processing unit 3 but are uniquely determined according to the above-described equations (2) to (5). And easily and accurately.

Based on the M sample values supplied from the buffer memory 4 to the signal processing unit 3 for each sampling period, the coefficients of the M terms calculated by the coefficient calculation unit 31 are calculated by the data calculation unit. 32. In the data calculation unit 32, the coefficients of the M terms calculated by the coefficient calculation unit 31 are substituted into, for example, an (M-1) -order polynomial expressed by Expression (1), and t '= T'n (where t'n = ΣΔNk)
The interpolation signal y (t'n) at the time point (see FIG. 3) is calculated. In a plurality of thin vertical lines shown in a section between time 0 and time +1 on the horizontal axis (time axis t) in FIG. 3, the interval between mutually adjacent ones is a signal every successive sampling period. It represents the magnitude of the signal ΔNk indicating the rate of decrease 1 / N of the pitch of the sound signal supplied to the input terminal 3c of the processing unit 3, and therefore represents the interval ΣΔNk between time 0 and time t'n in the figure. Of each signal ΔNk indicating the rate of decrease 1 / N of the pitch of the acoustic signal supplied to the input terminal 3c of the signal processing unit 3 at each of a plurality of sampling cycles (four sampling cycles in the illustrated example). This represents the accumulated value ΣΔNk.

The signal ΔNk indicating the rate of decrease 1 / N of the pitch of the acoustic signal supplied to the input terminal 3c of the signal processing unit 3 is
FIG. 2 is a block diagram showing an example of a block diagram in a case where the digital audio signal time axis extension readout method of the present invention is applied to an audio signal jog device or shuttle device. Supplied from the variable input unit 6a. FIG. 5 shows a signal .DELTA.Nk indicating the rate of decrease 1 / N of the pitch of an acoustic signal applied to the jog device or shuttle device when the jog device or shuttle device is operated.
5 is a variation on the time axis, and a curve J in FIG. 5 is a variation on the time axis of a signal ΔNk indicating a rate of decrease 1 / N of a pitch of an acoustic signal applied to the jog device during operation of the jog device. 5 is a curve showing an example, and a curve S in FIG. 5 is a change example on the time axis of a signal ΔNk indicating a rate of decrease 1 / N of a pitch of an acoustic signal given to the shuttle device during operation of the shuttle device. FIG.

In the data address management unit 34, the decimal part (accumulated from the accumulated value) of the signal ΔNk indicating the rate of decrease 1 / N of the pitch of the acoustic signal given for each successive sampling period is calculated. Integer minus the new signal Δ
Nk is added, and it is checked whether the value is 1 or more. If the magnitude of the accumulated value is 1 or less or less than 1,
Address information is supplied to the input terminal 4b of the address information of the buffer memory 4 via the output terminal 3b so that the set of M sample values used for obtaining the interpolation signal in each successive sampling period is not changed. . In addition, each one
Pitch reduction rate 1 of acoustic signal given for each sampling period
When the magnitude of the accumulated value obtained by accumulating the signal ΔNk indicating / N exceeds 1 and becomes 1 or more, the data address management unit 34 subtracts 1 from the accumulated value. Is used as the initial value of the accumulated value of the signal ΔNk in the next one sampling period, and the data read out from the buffer memory 4 is shifted toward the new sample value by one sampling period as a whole. Address information necessary for reading the set of M sample values in the shifted state from the buffer memory 4 and supplying the set to the coefficient calculation unit 31 of the signal processing unit 3 is output via the output terminal 3b. And supplies the address information to the input terminal 4b of the buffer memory 4.

The output of the above-mentioned data calculating unit 32 is output by a DC component removing unit 33 provided as necessary, when the signal reading speed is zero, such as when the pitch is 0, etc.
Alternatively, it is preferable that unnecessary DC components generated when the value is close to zero are removed before being output. The DC component removing unit 33 can be configured using a high-pass filter capable of removing a signal component of 10 Hz or less, for example. The signal processing unit 3 is configured using, for example, a digital signal processor. In the above-described case, whether the size of the accumulated value is 1 or more including 1 is checked. However, whether the size of the accumulated value exceeds 1 (excluding 1) or not is checked. In this case, if the magnitude of the accumulated value is 1 or less, the set of sample values is processed without being changed. If the magnitude of the calculated value exceeds 1, the processing may be performed with the set of sample values changed. Next, FIG. 2 is a block diagram showing a case where the time axis extension reading method of the digital audio signal of the present invention is applied to a jog device or a shuttle device for an audio signal. In FIG. 2, reference numeral 1 denotes an object of signal processing for lowering the pitch of the digital signal. Reference numeral 3 denotes a signal processing unit, 4 denotes a buffer memory, 5 denotes a recording / reproducing unit (for example, a recording / reproducing unit configured using a magneto-optical recording medium), and 6 denotes a digital signal input terminal. The control unit 7 is an output terminal. The recording / reproducing unit 5 stores information provided via the buffer memory 4 on a large-capacity recording medium, and outputs the stored information via the buffer memory 4. The digital signal having a sampling period of Ts supplied to the signal processing unit 3 through the input terminal 1 is converted into digital data of a predetermined format through the signal processing unit 3 and supplied to the buffer memory 4. The buffer memory 4 operates according to a control command from the control unit 6 and is used for a sample value interpolation operation. A sample value is read out according to address information given from a data address management unit 34 and signal processing is performed. The coefficient is supplied to the coefficient calculation unit 31 of the unit 3.

Further, the pitch reduction rate 1 / of the acoustic signal pitch given for each successive sampling period from the pitch variable input section 6a of the control section 6 to the input terminal 3c of the signal processing section 3 is described.
The signal ΔNk indicating N is supplied to the data calculation unit 32 and the data address management unit 34 as described above with reference to FIG. Depending on whether the device illustrated in FIG. 2 is an audio signal jog device or a shuttle device, the pitch is changed from the variable pitch input unit 6a to the input terminal 3c of the signal processing unit 3 at every one sampling period. As described above, the manner of change on the time axis of the signal ΔNk indicating the rate of decrease 1 / N of the pitch of the given acoustic signal differs from the curve J or the curve S in FIG. But each one in sequence
Pitch reduction rate 1 of acoustic signal given for each sampling period
/ N is input to the input terminal 3c of the signal processing unit 3.
Is supplied to the data calculation unit 32 and the data address management unit 34 via the data processing unit 32, thereby performing the signal processing operation described above with reference to FIG. 7 outputs a digital signal in the state of a digital signal which is expanded on the time axis. In addition, digital
The signal processing unit 3 configured using a signal processor uses a plurality of systems of sound signals (stereophonic sound) by a data address management program, a coefficient calculation program, a data calculation program, and the like provided therein. An interpolation signal is generated for each of the two signals in the case of a signal, and is output to the outside in a state synchronized with the external synchronization signal.

FIG. 4 shows an embodiment in which the digital audio signal time-base expansion reading apparatus of the present invention is configured with M = 4, and a 10 KHz sine wave signal (sampling frequency is 48 KH).
9 shows the frequency response characteristic of the reproduced signal when z) is pitch-converted to 0.6 times. A comparison between the characteristic curve shown in FIG. 4 and the case of the prior art described above with reference to the characteristic curve shown in FIG. In the case of (1), it can be seen that the level of the subharmonic that greatly affects the feeling of distortion is negligibly low, which is a major feature of the present invention. Although the harmonic components are still improved although they are considerably improved, the presence of the harmonic components hardly contributes to the distortion in the sense of hearing, so that the characteristics shown in FIG. 4 are obtained. This means that a good approximation can be obtained by the present invention. Needless to say, the present invention does not cause any problem for a low frequency (a frequency approximately equal to or less than 1/20 of the sampling frequency) which has no problem in the conventional example. In FIG. 4, as a result of the conversion, noise may appear in a frequency region exceeding 20 KHz, but this is an unnecessary part for hearing. Therefore, for example, a fixed coefficient for removing a frequency component of 20 KHz or more is used. The output signal may be output through the filter described above.

[0023]

As will be apparent from the above description, the method and apparatus for expanding and reading the time axis of a digital audio signal according to the present invention have the following advantages. The curve corresponds to an algebraic curve including a predetermined number M (where M is a natural number of 4 or more) of sample values that are continuous in a sample value sequence of a digital audio signal corresponding to the change mode on the curve ( M-1) To approximate the set of M sample values by the following polynomial, calculate coefficients of M terms based on the M sample values, and calculate the M terms An interpolation signal according to the (M-1) -th order polynomial is obtained and output using the coefficients of the above, and the signal ΔNk indicating the rate of decrease 1 / N of the pitch of the acoustic signal given at each of the above-mentioned successive one sampling periods is obtained. By accumulating, the magnitude of the accumulated value is, for example, 1 or less. The set of M sample values used to obtain the interpolated signal in each successive sampling period in the case was not changed, and the magnitude of the accumulated value of the signal ΔNk became greater than 1 In this case, a value obtained by subtracting 1 from the accumulated value is used as a signal ΔN in the next one sampling cycle.
Used as the initial value of the accumulated value of k, and instead of the set of M sample values used so far, M samples obtained by shifting the whole toward the new sample values by one sampling period In this embodiment, the interpolation signal according to the (M-1) -order polynomial described above is obtained and output using the set of sample values described above. , N is a positive or negative and any number greater than or equal to 1), the digital audio signal can be easily read out by extending the time axis with good approximation in a good approximation. There is no need to switch, so that it is possible to avoid the difficulty of detecting an edit point due to inconvenience such as generation of noise. In addition, in the signal processing method proposed above, a signal indicating a rate of decrease 1 / N of the pitch of an audio signal. When ΔNk is a small value very close to 0, However, there is a disadvantage that the scale of the filter to be used becomes so large that it is difficult to realize the filter. However, in the present invention, since the filter is not necessary, the above-described disadvantage is not only not present at all but also from the proposals already proposed. The operation scale in the present invention is always the same as the operation scale of the filter in the system even if the signal ΔNk indicating the rate of decrease 1 / N of the pitch of the sound signal is a small value very close to 0 or a large value. It is small and constant, and in the present invention, there is no distortion during output and the accuracy of searching for an edit point is good,
Further, in the present invention, the interpolation signal according to the (M-1) -th order polynomial described above is determined and approximated at each successive sampling period, so that time tracking is good and it is difficult to detect an edit point without occurrence of backlash. In addition, there is an advantage that search can be satisfactorily performed even when search in both the forward and reverse directions is mixed.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a schematic configuration of an apparatus to which a time-base expansion reading method of a digital audio signal according to the present invention is applied.

FIG. 2 is a block diagram in a case where the time axis extension readout method of a digital audio signal of the present invention is applied to a jog device or a shuttle device for an audio signal.

FIG. 3 is a waveform diagram for explaining a configuration principle and an operation principle of a method for reading out a digital audio signal on a time axis according to the present invention;

FIG. 4 is a diagram showing an example of output characteristics of a device to which a time-base expansion reading method of a digital audio signal according to the present invention is applied.

FIG. 5 is a signal Δ showing a rate of decrease 1 / N of a pitch of an acoustic signal.
It is a figure which shows the example of a change characteristic on the time axis of Nk.

FIG. 6 is a diagram illustrating an example of output characteristics according to a linear approximation method.

FIG. 7 is a diagram illustrating an example of an output characteristic according to a linear approximation method.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Input terminal, 3 ... Signal processing part, 4 ... Buffer memory, 5 ... Recording / reproducing part (for example, recording / reproducing part comprised using a magneto-optical recording medium), 6 ... Control part, 31 ... Calculation of coefficient Section, 32 ... data calculation section,

──────────────────────────────────────────────────続 き Continuation of the front page (56) References JP-A-60-95599 (JP, A) JP-A-2-66598 (JP, A) JP-A-5-281991 (JP, A) JP-A-5-281991 315891 (JP, A) (58) Field surveyed (Int. Cl. ⁷ , DB name) G10L 21/00-21/04 G11B 20/10

Claims (4)

(57) [Claims] 1. A time axis expansion read of a digital audio signal for reducing the pitch of the audio signal to 1 / N (where N is any positive or negative number of 1 or more) in the state of a digital signal. The method, wherein for each successive sampling period,
A predetermined number M (where M is 4 or more) continuous in the sample value sequence of the digital audio signal corresponding to the variation of the amplitude on the time axis of the audio signal targeted for pitch reduction To approximate the set of M sample values described above by an (M-1) -order polynomial corresponding to an algebraic curve including a sample value of (a natural number) on the curve, based on the M sample values By the obtained coefficient of M terms, (M-1)
A signal ΔNk (where k is a signal) indicating the rate of decrease 1 / N of the pitch of the sound signal given at each of the above-mentioned sequential sampling cycles, while obtaining and outputting an interpolation signal according to the following polynomial:
(1 or more natural numbers) The sum of M sampled values used to obtain an interpolation signal in each successive sampling period when the magnitude of the accumulated value obtained by performing the accumulation is 1 or less or less than 1 The set is not changed, and when the magnitude of the accumulated value of the signal ΔNk exceeds 1, the value obtained by subtracting the magnitude 1 from the accumulated value is used as the signal ΔNk in the next one sampling cycle.
Is used as the initial value of the accumulated value of M, and instead of the set of M sample values used so far, M samples obtained by shifting the whole toward the new sample values by one sampling period are used. A time-base expansion reading method of a digital audio signal in which an interpolation signal according to the (M-1) -th order polynomial described above is obtained and output using a set of sample values. 2. A time axis extension read of a digital audio signal for lowering the pitch of the audio signal to 1 / N (where N is any positive or negative number and a magnitude of 1 or more) in a digital signal state. An apparatus, wherein for each successive sampling period,
A predetermined number M (where M is 4 or more) continuous in the sample value sequence of the digital audio signal corresponding to the variation of the amplitude on the time axis of the audio signal targeted for pitch reduction To approximate the set of M sample values described above by an (M-1) -order polynomial corresponding to an algebraic curve including a sample value of (a natural number) on the curve, based on the M sample values Means for obtaining coefficient values of M terms; means for obtaining and outputting an interpolation signal according to the (M-1) -th order polynomial using the above-mentioned coefficient values; Means for accumulating a signal ΔNk (where k is a natural number of 1 or more) indicating the rate of decrease 1 / N of the pitch of the sound signal to be obtained, and a method for accumulating the signal when the magnitude of the accumulated value is 1 or less or less than 1 Used to obtain an interpolated signal in each successive sampling period If the magnitude of the accumulated value of the signal ΔNk exceeds 1, the value obtained by subtracting the magnitude 1 from the accumulated value is used as the next 1 It is used as the initial value of the accumulated value of the signal ΔNk in the sampling period, and instead of the set of M sample values used so far, the whole is shifted toward the new sample value by one sampling period. Means for using a set of M sample values obtained in the above manner. 3. A reduction in the pitch of an audio signal used to reduce the pitch of the audio signal to 1 / N (where N is a positive or negative and an arbitrary number of 1 or more) in the state of a digital signal. Signal ΔNk indicating rate 1 / N (where k is 1 or more
3. A means for variably inputting a natural number of
A digital audio signal time axis extension reading device. 4. The apparatus according to claim 2, further comprising means for removing a DC component.
A digital audio signal time axis extension reading device.