JPH06252772A - Coder or decoder and recording medium - Google Patents

Abstract

PURPOSE:To reduce the information quantity and the redundancy by coding a spectrum signal of an optional block depending on a spectrum signal of other block. CONSTITUTION:Time series sample data X21 stored in a time series sample buffer 132 are converted into spectrum signals X22-X24 by an MDCT calculation circuit 132 and stored in spectrum buffers 134-136. Then the spectrum signals X22-X24 are used and when a spectrum signal of one block is coded by a parameter calculation circuit 137, an energy calculation circuit 138, an energy discrimination circuit 139 and a parameter selection circuit 140, a parameter X28 to apply coding depending on spectrum information of two remaining blocks is selected out. Thus, highly efficient coding is attained by sending the two spectrum signals and the proper parameter X28 in place of sending the three spectrum signals X22-X24.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a coding device for coding digital signals such as voice, audio signals and image signals, a decoding device for decoding the coded signals, and the coded signals. The present invention relates to a recording medium on which is recorded.

[0002]

2. Description of the Related Art Conventionally, transform coding using so-called spectrum transform has been known as a kind of high-efficiency coding in which time-series sample data signals such as audio signals are bit-compressed and coded with high efficiency. . The transform coding is a spectrum conversion of an input signal in units of blocks to code, and a discrete cosine transform (DCT) is a typical spectrum transform. In this transform coding,
Block distortion is a problem in which discontinuous seams between blocks are perceived as noise, and in order to reduce this, it is common practice to overlap the ends of blocks with adjacent blocks. . Here, so-called MDCT (Modified DCT)
Is suitable for high-efficiency coding because it does not perform double transmission on samples in the overlapping part while having half (half block) overlap with blocks on both sides of an arbitrary block.

Regarding encoding and decoding using such MDCT and its inverse transform IMDCT, for example, Mochizuki, Yano, and Nishitani's "Filter Constraint Conditions for MDCT Mixed with Multiple Block Sizes", IEICE Technical Report, CAS90
-10, DSP90-14, pp.55-60, or "Adaptive block length adaptive transform coding using MDCT (ATC-ABS)" by Hazu, Sugiyama, Iwadari, Nishitani, 199.
Proceedings of the 0th IEICE Spring National Convention, A-1
97, etc. Hereinafter, the above MDCT and I
MDCT will be briefly described with reference to FIG. 7.

In FIG. 7, an arbitrary block of time-series sample data, for example, the J-th block, is the (J-
1) The block and the (J + 1) th block each have a half (50%) overlap. When the number of samples of the Jth block is N (N is a natural number), there is an overlap of N / 2 samples with the (J-1) th block and also with the (J + 1) th block. It has an overlap of N / 2 samples. Each of these blocks, for example, an arbitrary Jth block input time series sample 101 is subjected to a preprocessing filter or a forward conversion window Wh to obtain N time series data 102.

The characteristics of the preprocessing filter or the forward conversion window Wh are selected so that the power concentration degree of the converted data is the highest in accordance with the statistical properties of the input signal. By performing MDCT linear forward transform processing on the N-series time-series data 102, N / 2 independent spectrum data 103, which is half the number of input samples, can be obtained on the frequency axis. By performing a linear inverse transform process of IMDCT on the N / 2 pieces of spectrum data 103, N pieces of time series data 104 are obtained.
Is obtained. The time-series data 104 is subjected to a synthesis filter or an inverse conversion window Wf to obtain the time-series data 105, which is then added to the output results of the preceding and following blocks to restore the original input time-series sample data.

[0006]

By the way, in the conventional encoding / decoding method for recording or transmitting the spectrum data of all the blocks as described above, the redundancy is still high in consideration of further efficient encoding. There is a problem that it is expensive.

Therefore, the present invention has been made in view of the above-mentioned circumstances, and an encoding apparatus that realizes reduction of redundancy and further high-efficiency encoding, and a decoding apparatus corresponding thereto, It is also an object of the present invention to provide a recording medium for recording an encoded signal.

[0008]

DISCLOSURE OF THE INVENTION The present invention has been proposed in order to achieve the above-mentioned object, and an encoding device of the present invention divides an input signal into blocks, and converts the signal of each block into a spectrum signal. Is a coding device for coding the spectrum signal of each block, and when coding the spectrum signal of an arbitrary block, performs coding depending on the spectrum signal of another block. It is the one. Here, at the time of the above encoding, specifically, the following is performed. That is, when the spectrum signal in the arbitrary block is encoded, the difference signal between the spectrum signal obtained depending on the spectrum signal of the other block and the spectrum signal in the arbitrary block is encoded. In addition, the code of the spectrum signal of one block among the spectrum signals after the spectrum conversion of the J-th block, the (J + M) -th block, and the (J + 2M) -th block of the time-series sample data of the input signal. At the time of coding, coding is performed depending on the spectrum information of the remaining two blocks. A modified discrete cosine transform is used to convert the input signal into a spectrum signal. Furthermore, the spectrum signal of each block to be encoded is a spectrum signal obtained by dividing the spectrum signal of each block into a plurality of units and normalizing each unit. Further, when encoding the spectrum signal of the arbitrary block, the spectrum signal of the arbitrary block is divided for each band, and the spectrum of the same band of the other block with respect to the spectrum signal of each band. Encodes depending on the signal. Furthermore, when encoding the spectrum signal of each band in the arbitrary block, the encoding depends on the coefficient information obtained for each band in the arbitrary block together with the spectrum signal of the same band of another block. Give. The coefficient information is encoded by a non-linear quantizing means. The division of the band is performed based on the shape of the spectrum signal.

Next, the decoding device of the present invention divides the input signal into blocks, converts the signal for each block into a spectrum signal, and encodes the spectrum signal for each block to obtain information obtained. , A decoding device for decoding, which, when decoding encoded information of a spectrum signal of an arbitrary block, performs decoding depending on a spectrum signal of another block. . here,
In the above decoding, the following is specifically performed. That is, when decoding the encoded information of the spectrum signal of the arbitrary block, the difference signal between the spectrum signal obtained depending on the spectrum signal of the other block and the spectrum signal of the arbitrary block To decrypt. In addition, one of the J-th block, the (J + M) -th block, and the (J + 2M) -th block of the time-series sample data, which is one of the encoded spectrum signals after the spectrum conversion, is encoded. When the spectrum signal is decoded, the decoding is performed depending on the spectrum information of the remaining two blocks. At the time of the above decoding, the inverse transform of the modified discrete cosine transform is used to transform the spectrum signal into the time series sample data.
Further, the spectrum signal of each block is divided into a plurality of units, and the encoded information of the normalized spectrum signal is decoded for each unit. Further, when the encoded information of the spectrum signal of the arbitrary block is decoded, the spectrum information of the arbitrary block is divided for each band, and the other block is divided with respect to the spectrum information of each band. Decoding is performed depending on the spectrum information of the same band. Furthermore, when decoding the encoded spectrum information of each band of the arbitrary block, the coefficient information obtained for each band of the arbitrary block together with the spectrum information of the same band of the other block is used. Dependent decryption is performed. The coefficient information is decoded by a non-linear inverse quantization means. The division of the band is performed based on the shape of the spectrum signal.

Further, the recording medium of the present invention is one in which the information encoded by the encoding device of the present invention is recorded. That is, the recording medium of the present invention is a recording medium that blocks an input signal, converts the signal of each block into a spectrum signal, and records information obtained by encoding the spectrum signal of each block. Therefore, when the spectral signal of an arbitrary block is encoded, the information obtained by performing the encoding depending on the spectral signal of another block is recorded. Specifically, it is as follows. That is, at the time of encoding the spectrum signal in the arbitrary block, the information obtained by encoding the difference signal between the spectrum signal obtained in dependence on the spectrum signal of the other block and the spectrum signal in the arbitrary block is obtained. Record it. In addition, the code of the spectrum signal of one block among the spectrum signals after the spectrum conversion of the J-th block, the (J + M) -th block, and the (J + 2M) -th block of the time-series sample data of the input signal. The information obtained by performing encoding depending on the spectrum information of the remaining two blocks at the time of conversion is recorded. It should be noted that the information is recorded by using the modified discrete cosine transform for the conversion of the input signal into the spectrum signal. Further, the spectrum signal of each block is divided into a plurality of units, and the information obtained by performing the above encoding on the spectrum signal normalized for each unit is recorded. Further, the spectrum signal of the arbitrary block is divided for each band, and the information obtained by performing the encoding depending on the spectrum signal of the same band of the other block on the spectrum signal of each band is recorded. It will be. Furthermore, the spectrum signal of each band in the arbitrary block is subjected to encoding depending on the coefficient information obtained for each band in the arbitrary block together with the spectrum signal of the same band of the other block. The obtained information is recorded. Information obtained by encoding the coefficient information by a non-linear quantizing means is recorded. In addition, the information that the band division is performed based on the shape of the spectrum signal is recorded.

[0011]

According to the present invention, when the input signal is divided into blocks, the signal of each block is converted into a spectrum signal, and the spectrum of each block is encoded and decoded, the spectrum of an arbitrary block is converted. By encoding and decoding the signal depending on the spectrum signal of another block, the amount of information can be reduced and thus the redundancy can be reduced. Further, by recording this information, it becomes possible to record more information on the recording medium.

[0012]

Embodiments of the present invention will be described below with reference to the drawings.

Here, the schematic configuration of the encoding apparatus of the present embodiment is shown in FIG. 1 and the schematic configuration of the decoding apparatus of the present embodiment is shown in FIG. 3, but prior to the description of these specific configurations, The principle of encoding and decoding of the present invention will be described. That is, in the encoding and decoding device of the present invention, in order to solve the above-mentioned problem, the spectrum signal of another block is used by using the spectrum signal obtained by performing the spectrum conversion of the time-series sample data of some blocks. Coding and decoding are performed using a method of predicting on the frequency axis.

First, consider the case where the input signal is a sine wave in the encoding using MDCT for spectrum conversion. Here, the MDCT in the J-th block is defined by the mathematical expression (1) of Equation 1.

[0015]

[Equation 1]

In the equation (1), x is the input waveform, h is the window function for forward conversion, N is the block length, and X _J (k) is the kth MDCT output signal. n is an integer from 0 to N-1, and k is 0 to N / 2-
It is an integer up to 1.

Here, when the sine wave x _p (n) shown in the equation (2) is input, the MDCT in the Jth block is input.
The result is represented by the mathematical expression (3) of the equation 2.

X _p (n) = A _p sin (ω _p n + θ _p ) n ≧ 0 (2)

[0019]

[Equation 2]

Therefore, the MDCT result in the (J + M) th block is as shown in the equation (4) of the equation 3.

[0021]

[Equation 3]

However, at this time, the equation (5) and the equation (6) of the equation 4 are used.

C _{p, M} = cos (ω _p MN / 2), S _{p, M} = sin (ω _p MN / 2) (5)

[0024]

[Equation 4]

From this, the MDCT result in the (J + 2M) th block is as shown in equation (7).

X _{p, J + 2M} (k) = C _{p, 2M} X _{p, J} (k) + S _{p, 2M} Y _{p, J} (k) = (C ² _{p, M} -S ² _{p, M} ) X _{p, J} (k) + 2C _{p, M} S _{p, M} Y _{p, J} (k) ・ ・ ・ (7)

Further, since the formula (8) is established by the formula (4), substituting it into the formula (7) gives the formula
It becomes (9).

S _{p, M} Y _{p, J} (k) = X _{p, J + M} (k) −C _{p, M} X _{p, J} (k) (8)

X _{p, J + 2M} (k) = (c ² _{p, M} -S ² _{p, M} ) X _{p, J} (k) + 2C _{p, M} (X _{p, J + M} (k) -C _{p, M} X _{p, J} (k) = -X _{p, J} (k) + 2C _{p, M} X _{p, J + M} (k) ・ ・ ・ (9)

From this, the equations (10) and (11) can be obtained with X _{p, J + M} ≠ 0 and C _{p, M} ≠ 0.

C _{p, M} = (X _{p, J} (k) + X _{p, J + 2M} (k)) / 2X _{p, J + M} (k) (10)

X _{p, J + M} (k) = (X _{p, J} (k) + X _{p, J + 2M} (k)) / 2C _{p, M} (11)

Therefore, from the above equations (10) and (11), from the spectral data of two blocks among the J-th, (J + M) -th, and (J + 2M) -th blocks,
Spectral data for the remaining one block can be calculated.

Next, consider the case of a general input signal. The input signal is represented by the mathematical expression (12) of Expression 5.

[0035]

[Equation 5]

At this time, M in the Jth block
The DCT result is the same as when the sine wave is input,
Can be expressed as

[0037]

[Equation 6]

If equation (14) is used, equation (1
2) can be expressed as the mathematical expression (15) of the equation 7.

Ρ _{n, k} = (π (2k + 1) (2n + N / 2 + 1) / 2N), φ _p = ω _p (n + JN / 2) + θ _p (14)

[0040]

[Equation 7]

Therefore, the MDCT result in the (J + M) th block is as shown in the mathematical expression (16) of the equation 8 as in the case where the sine wave is used as the input signal.

[0042]

[Equation 8]

Here, if the expression (17) is set, the expression (16) can be expressed as the expression (18) of the expression 9.

Ψ _p = ω _p MN / 2 (17)

[0045]

[Equation 9]

Therefore, the single sine wave x _{p in} equation (2)
When inputting (n), the value of equation (3) is calculated as X0 _{p, J} (k),
When the value of the equation (6) is expressed as Y 0 _{p, J} (k), the equation (18) is transformed into the equation (19) of the equation 10.

[0047]

[Equation 10]

Therefore, in the (J + 2M) th block, the mathematical expression (20) of the mathematical expression 11 is obtained. Therefore, by modifying the mathematical expression as in the case of a single sine wave, the mathematical expression (2) of the mathematical expression 12 is obtained.
It can be expressed as 1).

[0049]

[Equation 11]

[0050]

[Equation 12]

Here, the spectrum in which energy is concentrated when a single sine wave x _p (n) is input is k _p . k
Considering the value of the equation (21) when = k _p , the Q of the equation (21) is
It is considered that all the other terms are close to 0 except the terms near p = P. From this, in k near k _p , the C _{p, M} is used, and the J-th, (J + M) -th, and (J + 2M) -th are the same as when a single sine wave is input. Using the spectra of two blocks among the three blocks, the spectra of the remaining blocks can be approximately obtained.

As described above _, by using the above C _{p, M} , the spectra of two blocks among the three blocks of the J-th, (J + M) -th, and (J + 2M) -th blocks are used, and the remaining The spectrum of the block can be approximately calculated.

Here, for example, considering the energies of m spectra each _, and using C _{p, M in the} vicinity where the energies of the m spectra are particularly concentrated, the Jth, (J + M) th, Using the spectra of two of the (J + 2M) th three blocks, the spectra of the remaining blocks can be approximately obtained.

Further, if there are a plurality of places where the energy is concentrated in this way, the remaining spectrum can be predicted by recording or transmitting C _{p, M} and the position thereof for each of them.

In this case, various methods of determining the degree of concentration of energy or setting parameters according to the method can be considered, but the method is not particularly limited here.

Therefore, according to the principle (method) of the present invention, without recording or transmitting all the spectrum data of these three blocks, the spectra of two blocks among them, C _{p, M} and appropriate parameters are recorded. Alternatively, the information can be encoded more efficiently by transmitting.

Alternatively, the parameters such as C _{p, M} ,
It is also conceivable to record information more efficiently by recording or transmitting the difference between the spectrum of the remaining blocks predicted by this method and the actual spectrum.

Further, a method is adopted in which this encoding is not applied to all of the input signals, but this encoding is adopted only in a block in which the encoding can be performed more effectively by this encoding. Can also be considered.

The present embodiment is based on the above-mentioned principle.

Here, before the concrete description of the encoding apparatus of the present embodiment of FIG. 1, the signal conversion procedure in the embodiment for performing the encoding according to the above-described principle of the present invention will be described using the flowchart of FIG. A brief description will be given.

In the first step S11 of the flow chart of FIG. 2, time-series sample data such as PCM audio data is divided into blocks every predetermined number of samples (N samples in this embodiment). Step S
After 11, the process proceeds to step S18. This step S18
Then, is the time-series sample data block the Jth block?
It is determined whether it is the (J + M) th or the (J + 2M) th,
For the data of the Jth block, step S12
The data of the (J + M) th block is sent to step S14, and the data of the (J + 2M) th block is sent to step S13.

In step S12, the process shown in FIG.
As shown in, each block is set such that the overlap amount between adjacent blocks is 50%, that is, each N / 2 sample overlaps each other, and the J-th block of the time-series data For the sample data, the forward conversion window Wh as shown in FIG. 7 described above.
Then, MDCT is applied to this to obtain N / 2 spectral data.

Also in steps S13 and S14, as in step S12, the forward transform window Wh is applied to the sample data of the (J + 2M) th and (J + M) th blocks, respectively, and the MDCT is applied to this.
To obtain N / 2 spectrum data.

In the next step S15, the above step S1
N / 2 parameter candidates for interpolating the spectrum data of the (J + M) th block are calculated from the spectrum data of the three blocks from S2, S13, and S14.

The sample data of the (J + M) th block of step S14 is also sent to step S16. In step S16, the energy of each spectrum of the (J + M) th block in which the spectrum itself is not recorded or transmitted is calculated.

After steps S15 and S16, the process proceeds to step S17. In this final step S17, of the N / 2 parameter candidates output from step S15, a parameter in the vicinity of the concentrated energy is selected to select the Jth and (J + 2M) th blocks. The spectrum data is output together with the processing, and the process ends.

The encoding apparatus of this embodiment, which is an example of the hardware configuration for implementing the encoding of the flowchart of FIG. 2, will be described below. In FIG. 1, the time series sample buffer 1 is supplied through the terminal 131.
The time-series sample data x21 of the three blocks stored in 32 is subjected to a forward conversion window by the MDCT calculation circuit 133 and converted into N / 2 spectral data x22, x23, x24 of each block.

These spectral data x22, x23,
x24 is the corresponding spectrum buffer 13
Once stored in 4,135,136, then taken out,
It is sent to the parameter calculation circuit 137.

The parameter calculation circuit 137 uses the converted spectrum data of each block to obtain N / 2.
The value x25 of each actual parameter is calculated.

The spectrum data x24 is also taken out from the spectrum buffer 136 and then sent to the energy calculation circuit 138.

In the energy calculation circuit 138, the (J
The energy x24 of the spectrum of the + M) block is summed with respect to, for example, every m pieces of spectrum data adjacent to each other, and normalized N / (2m) values x26 are output. This value x26 is sent to the energy determination circuit 139.

The energy judging circuit 139 selects a larger value than a certain value from the N / (2m) output values x26 of the energy calculating circuit 138, and maximizes the energy among the respective m pieces. The thing is combined with each energy, and it outputs as the value x27. Alternatively, if there is no output value x26 that is larger than a certain value, the maximum value is selected and the one with the maximum energy among the m output values is output as the value x27. This output value x27 is sent to the parameter selection circuit 140.

The output value x25 from the parameter calculation circuit 137 is also supplied to the parameter selection circuit 140. In the parameter selection circuit 140, the range to which each parameter is applied is calculated from each energy of the output value x27 of the energy determination circuit 139, and this is selected from the N / 2 output values x25 of the parameter calculation circuit 137. The parameter corresponding to is selected, combined with the value x27, and output as x28. This output is output from the terminal 142 as a parameter.

On the other hand, the spectrum buffers 134 and 1
The spectrum data x22 and x23 from the MDCT calculation circuit 133 via 35 are output from the terminal 141 via the output circuit 144.

The quantization of the above-mentioned energy and parameters at the time of recording or transmitting may be performed by a linear quantization means or a non-linear quantization means.

Before the description of the decoding apparatus of this embodiment shown in FIG. 3, the flow chart of FIG.
A signal conversion procedure in the embodiment of performing the decoding corresponding to the above-described encoding of the present invention will be described.

In the first step S31 of the flow chart of FIG. 4, the parameters of the parameters recorded directly from the encoding device or recorded on the package medium or the like or transmitted by the communication means and the spectra of the Jth and (J + 2M) th blocks. From the data, the spectral data of the (J + M) th block is interpolated.

In the next step S32, the interpolated (J
IMDCT on the spectrum data of the (+ M) th block
(Inverse transformation of modified discrete cosine transformation) is performed, and a window for inverse transformation is applied to generate time series data.

A detailed description will be given below with reference to FIG. 3 showing the decoding apparatus of this embodiment, which is an example of the hardware configuration for decoding in the flow chart of FIG.

In FIG. 3, the parameter value x27 from the encoder of FIG. 1 is sent to the parameter setting circuit 113 via the terminal 112. In this parameter setting circuit 113, the applicable range of each parameter is calculated using the parameter value x27 output from the encoder of FIG. 1, and N is calculated using each parameter of x28.
/ Parameter value x applied to each of the two samples
Set 40.

This parameter value x40 is sent to the (J + M) th block spectrum generation circuit 114. The (J + M) th block spectrum generation circuit 114 is also supplied with the spectrum data x22 and x23 from the encoding device of FIG. In the (J + M) th block spectrum generation circuit 114, the spectrum data x22 of the Jth block and the spectrum data x23 of the (J + 2M) th block, which are the outputs of the encoder of FIG.
And using N / 2 parameters x40, the (J + M) th
The spectral data x41 of the second block is output.
This spectrum data x41 is the IMDCT calculation circuit 11
Sent to 5.

On the other hand, the IMDCT calculation circuit 115 is also supplied with the spectrum data x22 and x23 from the encoding device of FIG. This IMDCT calculation circuit 115 has the above-mentioned (J +
M) IMDCT is applied to each of the output x41 of the block spectrum calculation circuit 114 and the spectrum data x22 and x23 of the Jth and (J + 2M) th blocks, and the inverse conversion window is applied to the time series sample data of each block. Output x24. This data x
24 is output from the terminal 116.

Next, a specific example of the high-efficiency coding apparatus to which the above-described coding apparatus of the present embodiment is applied will be described with reference to FIG. The specific high-efficiency coding apparatus shown in FIG. 5 uses techniques of band division coding, adaptive transform coding, and adaptive bit allocation. That is, in the high efficiency coding apparatus shown in FIG. 5, the input PCM
The frequency axis spectrum data obtained by dividing a digital signal such as an audio signal into a plurality of frequencies, selecting a wider bandwidth for higher frequencies, and performing the above-mentioned MDCT which is an orthogonal transformation for each frequency band, Bits are adaptively assigned and coded for each critical band.

That is, in FIG. 5, the input terminal 11
Is supplied with an audio PCM signal of 0 to 20 kHz, for example. This input signal is divided into a band of 0 to 10 kHz and a band of 10 to 20 kHz by a band dividing filter 12 such as a so-called QMF filter, and 0 to 10 kHz.
Similarly, the band signal is 0 to 5 kHz band and 5 to 10 kHz by the band division filter 13 such as a so-called QMF filter.
It is divided into a band and. 10 from the band division filter 12
The signal in the k to 20 kHz band is sent to the MDCT (Modified Discrete Cosine Transform) circuit 14 of the present embodiment described above, which is an example of the orthogonal transform circuit, and the signal from the band division filter 13 is converted into 5 signals.
The signal in the k to 10 kHz band is sent to the MDCT circuit 15,
The signal in the 0-5 kHz band from the band division filter 13 is M
By being sent to the DCT circuit 16, each MDC
T processed.

MD in each MDCT circuit 14, 15, 16
The spectrum data or the coefficient data on the frequency axis obtained by the CT processing is collected for each so-called critical band and sent to the adaptive bit allocation encoding circuit 17.

The critical band, that is, the critical band, is a frequency band divided in consideration of human auditory characteristics, and when a pure tone is masked by narrow band noise of the same strength in the vicinity of the frequency of a pure tone. It is the band that the noise has. This critical band is
The higher the band is, the wider the bandwidth is, and the entire frequency band of 0 to 20 kHz is divided into, for example, 25 critical bands.

The adaptive bit allocation coding circuit 17 is a block which is not interpolated, here, the Jth or (J +) th block.
In the 2M) th block, for example, for each critical band, each spectrum signal is normalized by the scale factor, that is, the maximum absolute value of the spectrum signal contained therein, and the normalized spectrum signal is , Quantizing noise is requantized with the number of bits enough to be masked by the signal of the critical band, and the requantized spectral signal is converted into the scale factor and the requantization obtained for each critical band. Output with the number of bits used. The data encoded in this way is taken out via the output terminal 18.

The adaptive bit allocation encoding circuit 17 is also provided.
In the block (here, the (J + M) th block) interpolated using the spectrum of another block, the spectrum signal of the other block (here, the Jth and (J + 2M) th) is input from the input terminal 19. The block spectrum signal) is received, the parameter values are calculated and quantized, and this data is output to the output terminal 18.

The interpolation of the spectrum signal may be performed for each band. Further, in this calculation, it is also possible to divide the spectrum signal of each block into several units and normalize the spectrum signal for each unit for calculation. This enables more accurate calculation with the same calculation word length.

Further, division of these bands or units may be variably performed according to the nature of the input signal.

When setting parameters in a block to be interpolated by using the spectrum of another block, the difference between the predicted spectrum data and the actual spectrum signal is adopted as a parameter, which is recorded or transmitted. Good.

The data from the output terminal 18 is recorded on the recording medium of the present invention (not shown). As the recording medium of the present invention, various recording media such as a magnetic tape, an optical disc, a magneto-optical disc, a phase change type optical disc, a semiconductor memory and a so-called IC card can be considered.

Next, a decoding device corresponding to the above-mentioned high efficiency coding device of FIG. 5 will be described with reference to FIG. The data read from the recording medium of the present invention is supplied to the terminal 21 of FIG.

In FIG. 6, a block which is not interpolated (here, the J-th and (J + 2) th blocks are not inserted into the input terminal 21.
In the (M) th block), the scale factor,
The number of bits used for requantization and the data obtained by encoding the requantized spectrum signal are input, and the spectrum decoding circuit 22 constructs a spectrum signal from these data.

In the block (here, the (J + M) th block in this example) interpolated from the spectrum signal of another block, the input terminal 21 has two other blocks (here, the Jth and (J + 2M) th blocks). The spectrum signal of the second block) and parameters for interpolation are input, and the spectrum decoding circuit 22 interpolates the spectrum signal to construct the spectrum signal.

Of these spectrum signals, 10 k-
The spectrum signal in the band of 20 kHz is the IMDCT circuit 23.
Therefore, the spectrum signal in the band of 5 kHz to 10 kHz is I
The MDCT circuit 24 converts the spectrum signal of the band of 0 to 10 kHz into the signal waveform data of each band by the IMDCT circuit 25. Of the signal waveform data in the three bands thus obtained, first, 0
The signal waveform data of ˜5 kHz and the signal waveform data of 5 kHz to 10 kHz are combined by the band integration circuit 26 to be 0 to 10
10k to 20k after converted to signal waveform data of kHz
The signal waveform data of Hz is combined with the band integration circuit 27, and the signal waveform data over the entire band is output from the output terminal 28.

The method of the present invention is not limited to the case of interpolating the j-th and (J + 2M) th blocks to the (J + M) th block as in the above embodiment. The case where the spectrum signal of the remaining one block is predicted from the spectrum signals of any two blocks of these three blocks is also included.

Further, the present invention is not limited to the above-mentioned embodiment, and, for example, the applied apparatus is the one shown in FIG.
And the high efficiency encoding / decoding device shown in FIG.
It can be applied to various transform coding devices, decoding devices for solving coding, and the like.

[0099]

According to the present invention, when an input signal is divided into blocks, the signal of each block is converted into a spectrum signal, and when the spectrum signal of each block is coded and decoded, an arbitrary block By encoding and decoding the spectrum signal depending on the spectrum signal of another block, the amount of information can be reduced, and thus the redundancy can be reduced. That is, according to the present invention, the amount of information required for encoding and decoding can be reduced, and more efficient encoding and decoding can be performed. Further, by recording the information by this encoding, it becomes possible to record more information on the recording medium.

[Brief description of drawings]

FIG. 1 is a block circuit diagram showing a schematic configuration of an encoding device according to an embodiment of the present invention.

FIG. 2 is a flow chart schematically showing a basic embodiment of the signal coding principle according to the present invention.

FIG. 3 is a block circuit diagram showing a schematic configuration of a decoding device of the present embodiment.

FIG. 4 is a flow chart schematically showing a basic embodiment of the signal decoding principle according to the present invention.

FIG. 5 is a block circuit diagram showing an example of a circuit configuration of a high-efficiency encoder using MDCT to which the encoder of this embodiment is applied.

FIG. 6 is an IMDCT to which the decoding device of this embodiment is applied.
FIG. 3 is a block circuit diagram showing an example of a circuit configuration of a high-efficiency code decoding apparatus using the.

FIG. 7: MDCT (Modified discrete cosine transform)
FIG. 6 is a diagram for schematically explaining the processing procedure of IMDCT, which is the inverse transform of the DCT and its inverse transform.

[Explanation of symbols]

 132 --- Time-series sample buffer 133 --- MDCT calculation circuit 134-136-spectrum buffer 136 --- parameter calculation circuit 138 --- Energy calculation circuit 139 ... Energy determination circuit 140 ... Parameter selection circuit 113 ... Parameter setting circuit 114 ... Spectrum generation circuit 115 ... ... IMDCT calculation circuit 12, 13 ... band division filter 14-16 ... MDCT circuit 17 ... ... adaptive bit allocation coding circuit 22 ... ... Spectrum decoding circuit 23-25 ... IMDCT circuit 26,27 ... Band integration circuit

Claims (36)

[Claims] 1. An encoding device for converting an input signal into blocks, converting the signal of each block into a spectrum signal, and encoding the spectrum signal of each block, when encoding the spectrum signal of an arbitrary block. The encoding device is characterized in that encoding is performed depending on a spectrum signal of another block. 2. When encoding a spectrum signal in the arbitrary block, a difference signal between the spectrum signal obtained depending on the spectrum signal of the other block and the spectrum signal in the arbitrary block is encoded. The encoding device according to claim 1, wherein the encoding device is an encoder. 3. The J-th time-series sample data of an input signal
Th block, (J + M) th block, (J
Among the spectrum signals after the spectrum conversion of the (+ 2M) th block, when the spectrum signal of one block is encoded, the encoding depending on the spectrum information of the remaining two blocks is performed. The encoding device according to claim 1. 4. The coding apparatus according to claim 1, wherein a modified discrete cosine transform is used for converting the input signal into a spectrum signal. 5. The spectrum signal of each block to be coded is a spectrum signal obtained by dividing the spectrum signal of each block into a plurality of units and normalizing each unit. Encoding device. 6. When encoding the spectrum signal of the arbitrary block, the spectrum signal of the arbitrary block is divided for each band, and the spectrum signal of each band is identical to that of the other block. The encoding device according to claim 1, wherein encoding is performed depending on a spectrum signal of a band. 7. The encoding of the spectrum signal of each band in the arbitrary block depends on the spectrum signal of the same band of another block and the coefficient information obtained for each band in the arbitrary block. The encoding device according to claim 1, wherein encoding is performed. 8. The encoding apparatus according to claim 7, wherein the coefficient information is encoded by a non-linear quantizing means. 9. The encoding device according to claim 6, wherein the division of the band is performed based on the shape of the spectrum signal. 10. A recording medium in which an input signal is divided into blocks, a signal for each block is converted into a spectrum signal, and information obtained by encoding the spectrum signal for each block is recorded. A recording medium characterized by recording information obtained by performing encoding depending on the spectrum signal of another block when encoding the spectrum signal of. 11. A difference signal between a spectrum signal obtained by depending on a spectrum signal of the other block and a spectrum signal of the arbitrary block when the spectrum signal is encoded in the arbitrary block is obtained by encoding. The recording medium according to claim 10, wherein the recorded information is recorded. 12. The spectrum of one block of the J-th block, (J + M) th block and (J + 2M) th block of the spectrum signal after the spectrum conversion of the time-series sample data of the input signal. 11. The recording medium according to claim 10, wherein the information obtained by performing the encoding depending on the spectrum information of the remaining two blocks when the signal is encoded is recorded. 13. The recording medium according to claim 10, wherein information is recorded by using a modified discrete cosine transform for converting the input signal into a spectrum signal. 14. The spectrum signal of each block is divided into a plurality of units, and the information obtained by performing the above coding on the spectrum signal normalized for each unit is recorded. Item 11. A recording medium within the range of Item 10. 15. Information obtained by dividing the spectrum signal of the arbitrary block for each band, and subjecting the spectrum signal of each band to coding depending on the spectrum signal of the same band of the other block. 11. The recording medium according to claim 10, wherein the recording medium is recorded. 16. A spectrum signal of each band in the arbitrary block is encoded together with a spectrum signal of the same band of the other block depending on coefficient information obtained for each band of the arbitrary block. The recording medium according to claim 10, wherein the obtained information is recorded. 17. The recording medium according to claim 16, wherein information obtained by encoding the coefficient information by a non-linear quantizing means is recorded. 18. The recording medium according to claim 15, wherein the information is recorded by dividing the band based on the shape of the spectrum signal. 19. The information obtained by dividing an input signal into blocks, converting the signal of each block into a spectrum signal, and encoding the spectrum signal of each block,
In the decoding device for decoding, when decoding the encoded information of the spectrum signal of an arbitrary block, the decoding device is characterized by performing the decoding depending on the spectrum signal of another block. 20. When decoding the encoded information of the spectrum signal of the arbitrary block, the spectrum signal obtained depending on the spectrum signal of the other block and the spectrum signal of the arbitrary block are used. 20. The decoding device according to claim 19, wherein the difference signal is decoded. 21. The Jth block, (J + M) th block, and (J + 2M) th block of the time-series sample data.
Among the coded spectrum signals after the spectrum conversion of the second block, when decoding the coded spectrum signal of one block, decoding depending on the spectrum information of the remaining two blocks is performed. 20. The decoding device according to claim 19, which is performed. 22. The decoding device according to claim 19, wherein an inverse transform of a modified discrete cosine transform is used for converting the spectrum signal into the time series sample data. 23. The spectrum signal of each block is divided into a plurality of units, and the encoded information of the normalized spectrum signal is decoded for each unit. Decoding device. 24. When decoding the encoded information of the spectrum signal of the arbitrary block, the spectrum signal of the arbitrary block is divided into bands, and the spectrum signal of each band is divided into 20. The decoding device according to claim 19, wherein decoding is performed depending on a spectrum signal of the same band of another block. 25. When decoding a coded spectrum signal of each band of the arbitrary block, a coefficient obtained for each band of the arbitrary block together with a spectrum signal of the same band of another block 20. The decoding device according to claim 19, wherein decoding is performed depending on information. 26. The decoding apparatus according to claim 25, wherein the coefficient information is decoded by a non-linear inverse quantization means. 27. The decoding apparatus according to claim 24, wherein the division of the band is performed based on a shape of a spectrum signal. 28. An arbitrary block in a recording medium for dividing an input signal into blocks, converting the signal for each block into a spectrum signal, and recording the information obtained by encoding the spectrum signal for each block. A recording medium on which information to be decoded depending on the spectrum signal of another block is recorded when the encoded information of the spectrum signal is recorded. 29. A difference between a spectrum signal obtained by depending on a spectrum signal of the other block and a spectrum signal in the arbitrary block when decoding encoded information of the spectrum signal of the arbitrary block. 29. A recording medium according to claim 28, characterized in that the signal records information to be decoded. 30. The J-th block, (J + M) -th block, and (J + 2M) -th block of time-series sample data
Of the coded spectrum signals after the above-mentioned spectrum conversion of the second block, when decoding the coded spectrum signal of one block, decoding is performed depending on the spectrum information of the remaining two blocks. 29. The recording medium according to claim 28, wherein the recording medium records information. 31. The recording medium according to claim 28, characterized in that information for which an inverse transform of a modified discrete cosine transform is used for converting a spectrum signal into time series sample data is recorded. 32. The spectrum signal of each block is divided into a plurality of units, and information for decoding the normalized spectrum signal for each unit is recorded. Recording medium in the range of. 33. When the encoded information of the spectrum signal of the arbitrary block is decoded, the spectrum signal of the arbitrary block is divided for each band, and the spectrum signal of each band is other than the above. 29. The recording medium according to claim 28, wherein information to be decoded is recorded depending on the spectrum signal of the same band of the block. 34. When the encoded information of the spectrum signal of the arbitrary block is decoded, the spectrum signal of the arbitrary block is divided for each band, and the spectrum signal of each band is other than the above. 29. The recording medium according to claim 28, wherein information to be decoded is recorded depending on the spectrum signal of the same band of the block. 35. The recording medium according to claim 34, wherein the coefficient information records information decoded by a non-linear inverse quantization means. 36. The recording medium according to claim 33, wherein the division of the band records information to be decoded based on the shape of the spectrum signal.