JP2002359560A - Coder and coding method, decoder and decoding method, recording medium and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To code, without enlarging the scale of a code sequence table as on the occasion using multi-dimensional variable-length codes. SOLUTION: A controller 31 checks the value of a quantized spectrum 401, computes the length (the number of specified values) of successively lined specified values (value 0 in this example) and controls a coder 32 to code the number thereof. The controller 31 controls the coder 33 to code other quantized spectra than the specified values (other than value 0). The coder 32 and a coder 33 execute a coding process, based on the specified code sequence table under the control of the controller 31.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an encoding apparatus and method, a decoding apparatus and method, a recording medium, and a program, and performs encoding, transmission, recording, and reproduction of input digital data by so-called high-efficiency encoding. The present invention relates to an encoding device and method, a decoding device and method, a recording medium, and a program suitable for use in obtaining a reproduced signal by decoding.

[0002]

2. Description of the Related Art There are various methods for high-efficiency encoding of signals such as audio or voice. For example, an audio signal or the like on a time axis is divided into a plurality of frequency bands and encoded without being blocked. Sub-band coding (sub-band coding: SBC (Subband Coding)), which is a non-blocking frequency band division method, or converting a signal on the time axis to a signal on the frequency axis (spectral conversion), A divided frequency band division method in which the frequency band is divided into frequency bands and encoded for each band, so-called conversion encoding, or the like can be used.

[0003] Also, a high-efficiency coding method combining the above-mentioned band division coding and transform coding has been considered.
In this case, for example, after band division is performed by band division coding, the spectrum of the signal in each band is converted into a signal on the frequency axis, and the spectrum-converted band is encoded. Here, as a filter for band division described above, for example, a QMF filter (Quadrature Mirror Filter)
For example, 1976 RECrochier
e Digital coding of speech in subbands, Bell Sys
t.Tech. J. Vol.55, No.8, 1976.

Also, ICASSP 83, BOSTON Polyphase Quadr
ature filters-A new subband coding technique, Jose
ph H. Rothweiler describes an equal bandwidth filter partitioning technique. Here, as the above-mentioned spectral transform, for example, an input audio signal is divided into blocks in a predetermined unit time (frame), and a discrete Fourier transform (DFT), a discrete cosine transform (Discrete Cosine Transform) is performed for each block. (DCT), Modified Discrete Cosine Transfo
rm) (MDCT) or the like, there is a spectrum conversion that converts the time axis to the frequency axis. For MDCT, for example, ICASSP 1987 Subband / Transform Coding Using Filt
er Bank Designs Based on Time Domain Aliasing Canc
ellation, JPPrincen ABBradley Univ. of Surr
ey Royal Melbourne Inst. of Tech.

[0005] By quantizing the signal divided for each band by the filter or the spectrum conversion as described above,
A band in which quantization noise is generated can be controlled, and a more efficient coding can be perceptually performed by utilizing properties such as a masking effect. Further, if the normalization is performed for each band, for example, with the maximum value of the absolute value of the signal component in the band before the quantization is performed, more efficient encoding can be performed.

[0006] As a frequency division width for quantizing each frequency component divided into frequency bands, for example, band division is performed in consideration of human auditory characteristics. That is, an audio signal may be divided into a plurality of bands (for example, 32 bands) in a band generally called a critical band (critical band) such that the higher the band, the wider the bandwidth. When encoding data for each band at this time, predetermined bits are allocated to each band, or encoding is performed by adaptive bit allocation (bit allocation) for each band.

For example, when the coefficient data obtained by the MDCT processing is encoded by the bit allocation, adaptively assigning to the MDCT coefficient data for each band obtained by the MDCT processing for each block. The encoding is performed with the number of bits. The following two methods are known as bit allocation methods.

The first method is Adaptive Transform Codin
g of Speech Signals, R. Zelinskiand P. Noll, IEEE Tr
ansactions of Accoustics, Speech, and Signal Process
ing, vol. ASSP-25, No. 4, August 1977. Here, bit allocation is performed based on the magnitude of the signal for each band. In this method, the quantization noise spectrum is flattened and the noise energy is minimized, but the actual noise sensation is not optimal because the masking effect is not used for the sense of hearing.

The second method is ICASSP 1980 The crit
ical band coder-digital encodingof the perceptual
requirements of the auditory system, MAKransner
Disclosed to MIT. Here, a method is described in which a necessary signal-to-noise ratio is obtained for each band by using auditory masking and fixed bit allocation is performed. However, in this method, even when the characteristic is measured with a sine wave input, the characteristic value is not so good because the bit allocation is fixed.

[0010] In order to solve these problems, all bits that can be used for bit allocation include a fixed bit allocation pattern predetermined for each small block and a bit allocation depending on the signal size of each block. A high-efficiency coding apparatus has been proposed in which a division ratio is used depending on the amount to be performed, and the division ratio depends on a signal related to an input signal, and the smoother the spectrum of the signal, the larger the division ratio into the fixed bit allocation pattern. ing.

According to this apparatus, when energy is concentrated on a specific spectrum such as a sine wave input, a large number of bits are allocated to a block including the spectrum, thereby significantly improving the entire signal-to-noise characteristic. Can be improved. In general, human hearing is extremely sensitive to signals having steep spectral components. Therefore, using such a method to improve the signal-to-noise characteristics merely improves the numerical values measured. In addition, it is effective in improving sound quality in terms of hearing.

[0012] Many other methods of bit allocation have been proposed. In addition, if an auditory model is refined and the capacity of the encoding device is increased, a code that is more efficient in terms of auditory sense will be provided. Becomes possible.

The present inventors have also disclosed in Japanese Patent Application No. 5-15286.
No. 5 has previously proposed a method of separating a tone component, which is particularly important in terms of audibility, from a spectral signal and coding the component separately from other spectral components. As a result, it is possible to efficiently encode an audio signal or the like at a high compression rate with almost no audible deterioration.

When the above-described DFT or DCT is used as a method of converting a waveform signal into a spectrum, M independent real number data can be obtained by performing conversion using a time block including M samples. In order to reduce the connection distortion between time blocks, each block usually overlaps the neighboring blocks with M1 samples each, so on average, DF
In T or DCT, M real number data is quantized and encoded for (M-M1) samples.

On the other hand, when the above-mentioned MDCT is used as a method of converting into a spectrum, independent 2N samples are overlapped with N times each of the adjacent times.
Since M pieces of real number data are obtained, on average, the MDCT quantizes and codes M pieces of real number data for M samples. In the decoding device, it is possible to reconstruct the waveform signal from the code obtained by using the MDCT in this way by adding the waveform elements obtained by performing the inverse transform in each block while interfering with each other. it can.

In general, by extending the time block for the transformation, the frequency resolution of the spectrum is increased,
Energy concentrates on specific spectral components. Therefore, the DFT is performed by using a MDCT in which the conversion is performed with a long block length by overlapping the neighboring blocks by half each and the number of obtained spectral signals does not increase with respect to the number of original time samples. It is possible to perform more efficient coding than when DCT or DCT is used.
In addition, by providing a sufficiently long overlap between adjacent blocks, distortion between blocks of a waveform signal can be reduced.

In constructing an actual code sequence, first, quantization accuracy information and normalization coefficient information are encoded with a predetermined number of bits for each band in which normalization and quantization are performed. What is necessary is just to encode the quantized spectrum signal.

For encoding a spectrum signal, a method using a variable length code such as a Huffman code is known. For Huffman codes, for example,
DavidA. Huffman, "A Method for the Construction of
Minimum-Redundancy Codes ", Proceedings of the I.
RE, pp 1098-1101, Sep., 1952.

Further, a plurality of spectral signals are collected into one
A method using a multidimensional variable length code expressed by two codes is known. Generally, in an encoding method using a multidimensional variable length code, as the order of the code is larger, more efficient encoding can be performed in terms of compression efficiency. However, as the order increases, the size of the code string table dramatically increases, which causes a problem in practical use. In practice, an optimal order according to the purpose is selected in consideration of the compression efficiency and the size of the code string table.

In general, in an acoustic waveform signal, energy is often concentrated on a fundamental frequency component and a frequency component that is an integral multiple of the fundamental frequency, that is, a so-called harmonic component, and the level of a spectrum signal around the frequency is higher than that of a so-called harmonic component. Since it is very small, the probability of being quantized to zero increases. In order to efficiently encode such a signal, it is sufficient to encode a spectrum signal quantized to 0 having a large occurrence probability with a minimum amount of information.
When a one-dimensional variable length code is used, even if each spectrum signal is encoded with the shortest code length of 1 bit, N
The spectrum of the book requires N bits of information. Order N
In the case of using the multi-dimensional variable-length code, since N spectra can be encoded with the shortest code length of 1 bit, efficient encoding can be performed on a signal having frequency components as described above. Can be performed.

However, in the encoding method using a multidimensional variable length code, increasing the order of the code has a considerable advantage in terms of compression efficiency, but in consideration of practical use, the code of the code is not limited. It is impossible to increase the order.

Normally, a code string table is prepared for each piece of quantization accuracy information set for each band. If the quantization precision is low, the value of the spectral signal that can be represented is small, so increasing the order does not increase the size of the code string table so much, but if the quantization precision is high, the value of the spectral signal that can be represented Therefore, even if the order is increased by one, the size of the code string table is significantly increased.

The above will be further described with reference to specific examples. Now, it is assumed that an input signal is subjected to MDCT conversion and a spectrum as shown in FIG. 1 is obtained. FIG. 1 shows the absolute value of the spectrum of the MDCT with the level converted to dB. The input signal is converted into 64 spectral signals for each predetermined time block, and these are combined into eight coding units [1] to [8], and normalization and quantization are performed.

By changing the quantization accuracy for each coding unit depending on the distribution of the frequency components, it is possible to minimize the deterioration of the sound quality and perform audio-efficient coding. The quantization accuracy information required in each coding unit can be obtained, for example, by calculating a minimum audible level and a masking level in a band corresponding to each coding unit based on an auditory model. The normalized and quantized spectrum signal is converted into a variable length code, and is encoded in each encoding unit together with quantization accuracy information and normalization information.

FIG. 2 is a diagram for explaining a method of expressing quantization accuracy information. When the quantization accuracy information code is expressed by 3 bits, up to eight types of quantization accuracy information can be set. In this example, 1, 3, 5, 7, 15, 31, 63, or 1
The quantization is performed by any of the eight steps of 27 steps. Here, being quantized in one step means that the spectral signals in the coding unit are all quantized to a value of zero.

FIG. 3 is a diagram for explaining a conventional variable-length coding method. The spectrum signal is quantized based on quantization accuracy information determined for each coding unit, and a quantized spectrum is obtained. When encoding the quantized spectrum, it is converted into a corresponding code string by referring to a code string table as shown in FIG.
As is clear from FIG. 4, a code string table is prepared for each piece of quantization accuracy information.

In FIG. 3, in the coding unit [1], a code “011” is selected as quantization accuracy information. Therefore, as shown in FIG. 4, quantization is performed in seven steps, and the values of the quantized spectrum signals (quantized spectrum) are in order from the lowest frequency.
3, -1, 2, and 2. When these are converted into code strings using the code string table in which the quantization accuracy information code of FIG. 4 is “011”, they are 1110, 101, 1100, and 110, respectively.
0, and the code lengths are 4, 3, 4, and 4, respectively.

In the coding unit [2], a code "010" is selected as quantization accuracy information. In this case, as shown in FIG. 4, quantization is performed in five steps. In this example, the quantization spectrum is -2, 1, 0, 1 in order from the lowest frequency. These are converted into code strings using the code string table with the quantization accuracy information code of “010” in FIG.
0, 0, 100, and the code lengths are 3, 3, 1,
It becomes 3.

Similarly, in the coding unit [3], the code “001” is selected as the quantization accuracy information.
The quantization is performed by the number of steps in the step, the quantized spectrum is 0, -1, 0, 0, the code sequence is 0, 11, 0, 0, and the code length is 1, 2, 1, 1.

FIG. 5 is a diagram for explaining a two-dimensional variable length coding method. FIG. 5 shows the same encoding unit as in FIG.

Assuming that a two-dimensional code string table as shown in FIG. 6 is used as the code string table when the quantization accuracy information code is “001”, the coding unit [3] in FIG.
Are grouped into one group by two and converted into one code string. Therefore, the quantized spectra of the four spectral signals 0, -1, 0,
0s are grouped into two groups of (0, -1) and (0, 0) and are converted into two code strings of 101 and 0.

When the spectrum signal of the coding unit [3] is coded with a one-dimensional variable length code as shown in FIG. 3, the required information amount is 1 + 2 + 1 + 1 = 5 bits. On the other hand, as shown in FIG. 5, when encoding is performed using a two-dimensional variable length code, the information amount is 3 + 1 = 4 bits. That is, it can be seen that when the degree is increased, encoding can be performed with a smaller amount of information.

[0033]

As described above,
Quantized spectra of a plurality of (N) spectral signals are grouped into one group to form N-dimensional data,
By encoding this into a variable length code according to the code string table, the code length can be reduced as compared with the case where a one-dimensional variable length code is used. However, when the order (the value of N) increases, the size of the code string table significantly increases, and there is a problem that practical use becomes difficult.

For example, when a two-dimensional variable length code is used, the quantization accuracy information code is "001", "01".
As shown in FIGS. 6 to 8, the code sequence table in the case of 0 ″ and “110” has a larger code sequence table than the code sequence table in the case of using a one-dimensional variable length code (FIG. 4). I have.

The present invention has been made in view of such a situation, and performs encoding with a short code length without increasing the size of a code string table as in the case of using multidimensional variable length encoding. Is what you can do.

[0036]

An encoding apparatus according to the present invention comprises:
Calculating means for detecting each numerical value constituting the numerical value sequence to be encoded and calculating the length of the sequence of the specific values; first encoding means for encoding the length of the sequence of the specific values; And second encoding means for encoding the numerical value of

The first encoding means and the second encoding means are:
The encoding can be performed alternately.

The numerical sequence to be encoded is a quantized acoustic waveform signal, and the specific value can be a value 0.

The first encoding means can encode the length in which the specific values are arranged into a variable-length code string.

The second encoding means encodes a numerical value other than the specific value into a variable-length code string.

The variable-length code string can be a Huffman code.

According to the encoding method of the present invention, each numerical value constituting a sequence of numerical values to be encoded is detected, and a calculation step of calculating the length of the specific values is arranged, and the length of the specific value is encoded. It is characterized by including a first encoding step and a second encoding step of encoding a numerical value other than the specific value.

The program of the first recording medium of the present invention comprises:
A calculating step of detecting each numerical value constituting the numerical value sequence to be encoded and calculating a length in which the specific values are arranged; a first encoding step of encoding the length in which the specific values are arranged; And a second encoding step of encoding the numerical value of.

A first program according to the present invention detects each numerical value constituting a numerical value sequence to be coded, and calculates a length in which specific values are arranged, and encodes a length in which specific values are arranged. The present invention is characterized by causing a computer to execute a process including a first encoding step of performing encoding and a second encoding step of encoding a numerical value other than a specific value.

In the encoding apparatus and method of the present invention, the program of the first recording medium, and the first program, each numerical value constituting the numerical value sequence to be encoded is detected, and the length in which specific values are arranged. Is calculated, the length in which the specific values are arranged is encoded, and a numerical value other than the specific value is encoded.

A decoding device according to the present invention decodes a code string having a length in which specific values are arranged, and determines a sign of a signal component based on the result, and a decoding means for decoding the sign of the signal component. A decoding device comprising:

When the value obtained by decoding the code string having the length in which the specific values are arranged is the value 0, the determination means determines whether the value other than the specific value inputted next to the code string having the length in which the specific values are arranged is obtained. If the value obtained by decoding the code string of the value and determining the resulting value as the sign of the signal component and decoding the code string having a length in which the specific value is arranged is not the value 0, the value is obtained. Can be determined to be a specific value.

The specific value can be the value 0.

The decoding method of the present invention decodes a code string having a length in which specific values are arranged, and determines a sign of a signal component based on the result, and a decoding step of decoding the sign of the signal component. It is characterized by including.

The program of the second recording medium of the present invention comprises:
It is characterized by including a determining step of decoding a code string having a length in which specific values are arranged and determining a code of a signal component based on the result, and a decoding step of decoding a code of the signal component.

A second program according to the present invention decodes a code string having a length in which specific values are arranged, determines a code of a signal component based on the result, and decodes the code of the signal component. And causing the computer to execute the processing including the steps.

In the decoding apparatus and method of the present invention, the second recording medium, and the second program, a code string having a length in which specific values are arranged is decoded, and the code of the signal component is determined based on the result. And the sign of the signal component is decoded.

[0053]

FIG. 9 shows an example of the configuration of an encoding apparatus to which the present invention is applied.

The signal waveform 10 input to this encoding device
1 is converted into a signal frequency component 102 by a conversion circuit 1, each component is encoded by a signal component encoding circuit 2, and a code sequence is generated by a code sequence generation circuit 3. This code string is transmitted to a predetermined transmission path or recorded on the information recording medium 4.

FIG. 10 shows a configuration example of the conversion circuit 1.

The signal components 211 and 212 divided into two bands by the band division filter 11 are converted into spectral signal components 221 and 222 by forward spectrum conversion circuits 12 and 13 such as MDCT in each band. It has been done. The bandwidth of the signals 211 and 212 is １ ／ of the bandwidth of the signal 201 (signal 2
01 has been decimated to 1/2).

Note that 201 in FIG. 10 is replaced with 101 in FIG.
221 and 222 in FIG. 10 respectively correspond to 102 in FIG.

A large number of conversion circuits other than this embodiment are conceivable. For example, the input signal may be directly converted to a spectrum signal by MDCT, or DF instead of MDCT.
It may be converted by T or DCT.

Although it is possible to process a signal without converting it into a spectrum simply by dividing a signal into band components by a band division filter, the method of the present invention is particularly effective when energy is concentrated at a specific frequency. Therefore, it is convenient to adopt a method of converting a large number of frequency components into frequency components by spectral conversion that can be obtained with a relatively small amount of calculation.

FIG. 11 shows a configuration example of the signal component encoding circuit 2.

Each signal component 301 is normalized by the normalizing circuit 21 for each predetermined band, and then input to the quantizing circuit 22 as a signal 302, where the quantization precision determining circuit 2
3 is quantized on the basis of the quantization precision signal 303 calculated in Step 3 and is output as a signal 304. 11 corresponds to 102 in FIG. 9, and 304 in FIG. 11 corresponds to 103 in FIG. 9, respectively.
Here, 304 (103) includes normalization coefficient information and quantization accuracy information in addition to the quantized signal components.

FIG. 12 shows a configuration example of the code string generation circuit 3.

The control unit 31 checks the value of the quantized spectrum signal value (quantized spectrum) 401 input to the code string generation circuit 3 and determines the quantum of a specific value (in this case, value 0). In addition to detecting the length (the number of quantized spectra having a value of 0) in which the encoded spectra are arranged, the encoding unit 32 controls and encodes the number.

The control unit 31 controls the encoding unit 33 to encode a quantization spectrum other than the specific value (a quantization spectrum other than the value 0 in this example).

The encoding unit 32 and the encoding unit 33 execute an encoding process based on a predetermined code string table (described later) under the control of the control unit 31, and the code strings 402 and
403 is output to a predetermined transmission path or the information recording medium 4 as appropriate.

Note that reference numeral 401 in FIG.
In addition, 402 and 403 in FIG. 12 correspond to 104 in FIG. 9, respectively.

Next, the operation of the encoding device (FIG. 9) will be described. The acoustic waveform signal 101 (201) is input to the band division filter 11 (FIG. 10) of the conversion circuit 1, and a lower frequency signal component 211 and a higher frequency signal component 2
12 are divided. Lower frequency signal component 211
Is input to the forward spectrum conversion circuit 12 and converted into a spectrum signal component 221. Similarly, the higher frequency signal component 212 is input to the forward spectrum conversion circuit 13, converted into a spectrum signal component 222, and output.

That is, with reference to FIG. 1, the forward spectrum conversion circuit 12 unitizes a lower frequency spectrum signal to generate coding units [1] to [6]. Further, the forward spectrum conversion circuit 13
Unitizes the higher frequency spectral signal components to generate coding units [7], [8].

The spectrum signal components 221 and 222 (102) output from the forward spectrum conversion circuits 12 and 13
Is the normalization circuit 21 of the signal component encoding circuit 2 (FIG. 11)
Is input to the quantization accuracy determination circuit 23. Normalization circuit 2
1 divides the value of each signal by its maximum value among a plurality of spectral signal components in the encoding unit,
Perform normalization. Then, the obtained normalization coefficient 302 is supplied to the quantization circuit 22.

The quantization precision determination circuit 23 determines the quantization precision of the input spectral signal component by coding unit by calculating the minimum audible level and masking level in the band corresponding to each coding unit. I do. The quantization circuit 22 quantizes the normalization coefficient 302 supplied from the normalization circuit 21 with the quantization precision 303 supplied from the quantization precision determination circuit 23, and obtains the obtained quantization spectrum 3
04 (103) is supplied to the code string generation circuit 3.

The code string generation circuit 3 converts the quantized spectrum into a code string. Details of the processing will be described with reference to the flowchart of FIG.

In step S1, the code string generation circuit 3
The control unit 31 has a built-in counter i for counting the number of spectrum signals in the encoding unit and a counter run for counting the length of the quantization spectrum of a specific value (in this example, value 0). Is initialized to the value 0.

Next, in step S2, the control unit 31
Is a spectrum signal corresponding to the value of the counter i (hereinafter, referred to as
The quantization spectrum 401 (referred to as QSP (i)) is examined, and it is determined whether or not the quantization spectrum 401 has a value of 0. If it is determined that the value is 0, the process proceeds to step S3.

In step S3, the control unit 31 increases the values of the counter i and the counter run by 1, and in step S4, determines whether the value of the counter i is less than the number M of the spectrum signals in the coding unit. If it is determined that the number of spectrum signals is less than the number M, the process returns to step S2 and the subsequent processing is executed.

If it is determined in step S4 that the value of the counter i is not less than the number M of the spectrum signals (several M
If it is the above, the process proceeds to step S5 and the control unit 31
Controls the encoding unit 32 to encode the value of the counter run. Accordingly, the encoding unit 32 converts the value of the counter “run” into a code string corresponding to the value by referring to the code string table in FIG.

If it is determined in step S2 that the quantized spectrum 401 of the QSP (i) is not 0, the process proceeds to step S6, where the control unit 31 controls the encoding unit 32 to execute the counter run. The value is encoded, and in step S7, the value of the counter run is reset to a value of 0.

Next, in step S8, the control unit 31
Controls the encoding unit 33 to encode the quantized spectrum (a value other than the value 0) 401 of the QSP (i). Thereby, the encoding unit 33 converts the quantized spectrum of QSP (i) into a code sequence corresponding to the value by referring to the code sequence table in FIG.

In the present invention, as described above, the number of successively arranged quantized spectra (the value of the counter run) (hereinafter referred to as a run-length run) is coded as described above. (Steps S5 and S6), the value 0 itself is not encoded. That is, the code string table of FIG. 15 does not include a code string for a quantized spectrum having a value of 0.

In this example, the encoding of the run length run (steps S5 and S6) and the encoding of the quantized spectrum other than the value 0 (step S8) are performed alternately. Is, after all, run-length ru
The code sequence of n and the code sequence of the quantized spectrum other than the value 0 are alternately arranged. In a decoding device described later,
A decoding process using this principle is performed.

Returning to FIG. 13, in step S9, the control unit 31 increases the value of the counter i by 1, and in step S10, determines whether the value of the counter i is less than the number M of the spectrum signals of the coding unit. It is determined whether or not it is less than the number M of spectrum signals, and the process returns to step S2 to execute the subsequent processing.

After the processing in step S5 or when it is determined in step S10 that the value of the counter i is equal to or more than M, the processing ends and the next coding unit is processed.
Steps S1 to S10 are executed in the same manner.

Next, as an example, the coding unit shown in FIG. 16 in which 16 (= M) spectral signals exist (the coding unit [7] or [8] in the example of FIG. 1) will be described above. The operation of the code string generation circuit 3 will be described again.

When the counter i = 0 and the counter run = 0 (step S1), the quantized spectrum of QSP (0) (spectral signal of the lowest frequency among the encoding units) (shown at the left end in the figure) (Quantized spectrum to be obtained) is 0 (step S2), the values of the counter run and the counter i are increased by 1 to 1
(Step S3). Since the quantization spectrum of QSP (1) is also a value of 0, the value of the counter run, i is further increased by 1 to a value of 2.

The quantization spectrum of QSP (2) has a value of 2
Therefore, the value 2 of the counter run at that time is
Is encoded into a 3-bit code string "011" (step S6). Then, after the value of the counter run is reset to the value 0 (step S7), the QSP
The quantization spectrum (value 2) of (2) is encoded into a 2-bit code string “10” based on the code string table in which the quantization accuracy information code shown in FIG. 15 is “010”.

The other spectrum signals (QSP (3) through QSP (3)
The quantization spectrum of SP (15) is also encoded in the same manner.

That is, in the present invention, for example, when the quantization precision information code is “011”, 1 is used to encode the value of the counter “run” shown in FIG.
A code string table of a total of 22 words, that is, a code string table of 6 words and a code string table of 6 words whose quantization accuracy information code is “011” shown in FIG. 15 is used.

On the other hand, for example, in the case of a conventional two-dimensional variable length code, a code string table of 49 words shown in FIG. 8 is required. Although not shown, in the case of a four-dimensional variable length code, a code string of 2401 words is required.

In the present invention, the encoding unit shown in FIG. 16 is a 3-bit code string "011" (a code string of a run-length run having a value of 2) and a 2-bit code string "10".
(Code sequence of QSP (2) quantized spectrum having value 2),
5 bit code string “11010” (run length of value 9)
run code), 2-bit "01" (QSP of value -1)
(12) Quantized spectrum code string)
011 "(code string of run-length run of value 2) and 2-bit" 10 "(code string of quantized spectrum of QSP (15) of value 2), the code length of which is 17 (= 3 + 2 + 5 + 2 + 3 + 2) On the other hand, in the conventional two-dimensional variable length code, the code length of the encoding unit in FIG. 16 is 21 as shown in FIG. The long code has 22 bits as shown in FIG.

That is, in the present invention, the encoding unit in FIG. 16 is encoded with a shorter code length than a conventional one-dimensional or two-dimensional variable length code.

The code string table used for coding the run-length run shown in FIG. 14 has a code string corresponding to the number of spectrum signals of the coding unit. It is also possible to use a code string table of less than the number of code strings. In this case, the counter run
Is the maximum value, the value of the counter run is 0
And the process returns to step S2.

FIG. 19 shows an example of the configuration of a decoding device that decodes and outputs an audio signal from a code sequence generated by the coding device of FIG.

The information has been transmitted or the information recording medium 4
The code of each signal component is extracted by the code sequence decomposition circuit 41 from the code sequence 501 reproduced from
2 by the signal component decoding circuit 42
3 is decoded, and an acoustic waveform signal 504 is generated and output by the inverse conversion circuit 43.

FIG. 20 shows an example of the configuration of a code string decomposition circuit 41 for converting the code string of each encoding unit into a signal component code.

The control unit 51 appropriately reads the code string 601 of the coding unit input to the code string decomposition circuit 41,
The decoding unit 52 controls the decoding unit 52 to decode the read run-length run code string, and determines the sign of the signal component based on the decoding result.

For example, when the value 0 is obtained by the decoding of the decoding unit 52, the control unit 51 sets the code sequence input after the run-length run code sequence (the code sequence of the quantized spectrum other than the value 0). Is decoded by controlling the decoding unit 53, and the value obtained as a result is determined as the sign of the signal component. When a value other than the value 0 is obtained by the decoding of the decoding unit 52, the control unit 51 determines the sign of the signal component corresponding to the value to be a specific value (the value 0 in this example). The control unit 51 outputs the code of the determined signal component to the signal component decoding circuit 42 as a code 602 as appropriate.

The decoding unit 52 performs a decoding process based on the code string table corresponding to FIG. 14 according to the control of the control unit 51.
Is performed based on the code string table corresponding to.

Note that reference numeral 601 in FIG. 20 denotes 501 in FIG.
20 corresponds to 502 in FIG. 19, respectively.

FIG. 21 shows a configuration example of the inverse conversion circuit 43. This corresponds to the conversion circuit 1 in FIG.
The signals 701, 701 are output by the inverse spectrum conversion circuits 61, 62.
The signals 711 and 712 of each band obtained from 702 are synthesized by the band synthesis filter 63 and output as a signal 721. 701 and 7 in FIG.
02 corresponds to 503 in FIG. 19, and 721 in FIG.
04 respectively.

Next, the operation of the decoding device will be described.

The code string decomposition circuit 41 performs a code string decomposition process as shown in the flowchart of FIG.

In step S21, the control unit 51
The value of a built-in counter i for counting the number of spectrum signals in the encoding unit is initialized to a value of zero.

In step S22, the control unit 51
The code string of the encoding unit is sequentially read, and the read run-length run code string is controlled and decoded by the decoding unit 52. In step S23, it is determined whether or not the resulting value is a value 0. If it is determined that the value is 0, the process proceeds to step S24.

In step S24, the control unit 51
The code string of the encoding unit is further read, and the read code string is decoded by controlling the decoding unit 53.
The code (quantized spectrum) of (i) is used.

Next, in step S25, the control unit 5
In step S2, the value of the counter i is increased by one.
In 6, it is determined whether or not the value of the counter i is less than the number M of spectrum signals in the coding unit. If it is determined that the value is less than M, the process returns to step S22, and the subsequent processing is executed. .

In step S23, the run length run is
If it is determined that the value is not 0, the process proceeds to step S27, and the control unit 51 sets the sign of QSP (i) to QSP (i + run-length run-1) to the value 0.

Next, in step S28, the control unit 5
1 increases the value of the counter i by the value of the run length run, and in step S29, the value of the counter i is
It is determined whether or not the number of spectrum signals is less than M. If it is determined that it is less than M, the process returns to step S24, and the subsequent processing is executed.

In step S26 or step S29,
If it is determined that the value of the counter i is not less than the number M of the spectrum signals of the coding unit (if it is not less than M), the process ends, and the code string of the next coding unit is subjected to step S21. Steps S29 to S29 are performed in the same manner.

The code 502 of each signal component extracted by the code sequence decomposition circuit 41 in this way is input to the signal component decoding circuit 42 and decoded. This signal component decoding circuit 42 performs a process reverse to that of the signal component encoding circuit 2 shown in FIG.

The signal 503 output from the signal component decoding circuit 42 has a lower frequency spectrum signal component 701 input to the inverse spectrum converter 61, and the higher frequency spectrum signal component 702 has an inverse frequency spectrum signal component 702. The signal is input to the spectrum conversion circuit 62. The inverse spectrum conversion circuits 61 and 62 convert the input spectrum signal components 701 and 702 into acoustic signals 711 and 711 on the time axis, respectively.
712, and outputs the result to the band synthesis filter 63. The band synthesis filter 63 outputs the lower frequency sound signal 71.
1 and an audio signal 712 of a higher frequency are synthesized and output as a synthesized audio signal 721 (504).

In general, an acoustic waveform signal often has energy concentrated in a fundamental frequency and a frequency component that is an integral multiple of the fundamental frequency, that is, a so-called harmonic component, and the spectral signal of other frequency components is zero even if quantized. There are many. The higher the fundamental frequency, the greater the distance between the so-called harmonic components, and therefore the higher the probability of occurrence of a quantized spectrum of 0. In addition, a signal having such a frequency component, unless quantization is performed with a higher quantization precision, a sufficient amount of information is consumed because a sufficient signal-to-noise ratio in audibility cannot be secured. In particular, the amount of information consumed is large in a band in which the width of the coding unit is wide.

Therefore, in order to efficiently encode such a signal, it is important how to encode the spectrum signal quantized to 0 with a small amount of information.
The best known method is to use a multi-dimensional variable-length code, but as described above, a huge amount of code string table is required, which is not practical. However, the method according to the present invention achieves efficient encoding without increasing the size of the code sequence table.

As described above, a signal obtained by subjecting a signal once subjected to a band division filter to spectrum conversion by MDCT is used as a band division circuit, and an inverse MDCT (IMD
Although the embodiment using the inverse spectrum transformed by (CT) and applied to the band synthesizing filter has been described, it is needless to say that the MDCT transform and the IMDCT transform are performed directly without using the band dividing filter and the band synthesizing filter. You may do it. Also, the type of spectrum conversion is not limited to MDCT, and DFT, DCT, etc. may be used.

Further, the band division and the band synthesis may be performed only by the band division filter and the band synthesis filter without using the spectrum conversion. In this case, a band divided by the filter or a band obtained by combining a plurality of these bands is defined as an encoding unit. However, the method of the present invention can be applied efficiently by performing spectral conversion such as MDCT and converting into a large number of spectral signals, and then configuring the coding unit as described using the embodiment. .

In the above description, the case of processing an acoustic waveform signal has been described. However, the method of the present invention can be applied to other types of signals, for example, to image signals. It is possible. However, the present invention realizes efficient encoding when there are many spectral signals quantized to 0 by using the characteristics of the acoustic waveform signal, and therefore, when the acoustic waveform signal is used for the acoustic waveform signal. Demonstrate power.

Further, the method according to the present invention can be used in combination with a conventional method using a variable length code. For example, when the quantization precision is low, that is, when the number of quantization steps is small, encoding is performed using a two-dimensional or four-dimensional code sequence table by a conventional method, and the quantization precision is high, that is, the number of quantization steps is large. In the case where is large, it is more effective if the encoding is performed using the method according to the present invention in consideration of the scale of the code string table.

The above-described series of processing can be realized by hardware, but can also be realized by software. When a series of processing is realized by software, a program constituting the software is installed in a computer, and the program is executed by the computer, so that the above-described encoding device and decoding device are functionally realized. Is done.

FIG. 23 is a block diagram showing the configuration of an embodiment of the computer 101 functioning as the above-described encoding device or decoding device. CPU (Central Pro
), an input / output interface 116 is connected to the CPU 111 via a bus 115.
When a user inputs a command through the input / output interface 116 from an input unit 118 including a keyboard, a mouse, and the like, for example, the ROM (Read Only Memory) 11
2. A program stored in a recording medium such as a magnetic disk 131, an optical disk 132, a magneto-optical disk 133, or a semiconductor memory 134 mounted on the hard disk 114 or the drive 120 is stored in a RAM (Random).
Access Memory) 113 for execution. Thus, the various processes described above (for example, the processes shown by the flowcharts in FIGS. 13 and 22) are performed. Further, the CPU 111 transmits the processing result to an LCD (Liquid Crystal Diode) via the input / output interface 116, for example.
splay) or the like is output as necessary. The program is stored on the hard disk 114 or RO
M112 is stored in advance and provided to the user integrally with the computer 101, the magnetic disk 131, the optical disk 132, the magneto-optical disk 133, the semiconductor memory 1
34, packaged media such as satellite,
It can be provided to the hard disk 114 via a communication unit 119 from a network or the like.

In the present specification, the step of describing a program provided by a recording medium may be performed not only in chronological order but also in chronological order in the order described. This also includes processing executed in parallel or individually.

[0119]

According to the encoding apparatus and method of the present invention, the program of the first recording medium, and the first program, each numerical value constituting the numerical sequence to be encoded is detected, and the specific value is detected. Since the length of the line is calculated, the length of the specific value is encoded, and numerical values other than the specific value are encoded, so that, for example, without increasing the scale of the code string table,
Encoding with high compression efficiency can be performed.

According to the decoding apparatus and method of the present invention, the second recording medium, and the second program, a code string having a length in which specific values are arranged is decoded, and a code of a signal component is decoded based on the result. Decision and decoding of the code of the signal component, for example, without increasing the scale of the code string table,
Encoding with high compression efficiency can be performed.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating a spectrum of an encoding unit.

FIG. 2 is a diagram illustrating a method of expressing quantization accuracy information.

FIG. 3 is a diagram illustrating an encoding method using a variable length code according to the related art.

FIG. 4 is a diagram illustrating an example of a code string table.

FIG. 5 is a diagram illustrating an encoding method using a multidimensional variable length code according to the related art.

FIG. 6 is a diagram showing an example of a multidimensional code string table according to the related art.

FIG. 7 is a diagram showing an example of another multi-dimensional code string table according to the related art.

FIG. 8 is a diagram showing an example of another multidimensional code string table according to the related art.

FIG. 9 is a block diagram illustrating a configuration example of an encoding device to which the present invention has been applied.

FIG. 10 is a block diagram illustrating a configuration example of a conversion circuit in FIG. 9;

11 is a block diagram illustrating a configuration example of a signal component encoding circuit in FIG. 9;

12 is a block diagram illustrating a configuration example of a code string generation circuit in FIG. 9;

FIG. 13 is a flowchart illustrating an operation of the code sequence generation circuit.

FIG. 14 is a diagram illustrating an example of a code string table used when encoding a run.

FIG. 15 is a diagram illustrating an example of a code string table used when encoding a quantized spectrum other than a specific value.

FIG. 16 is a diagram illustrating an encoding method according to the present invention.

FIG. 17 is another diagram illustrating an encoding method according to the related art.

FIG. 18 is another diagram illustrating an encoding method according to the related art.

FIG. 19 is a block diagram illustrating a configuration example of a decoding device to which the present invention has been applied.

20 is a block diagram illustrating a configuration example of a code string decomposition circuit in FIG. 19;

21 is a block diagram illustrating a configuration example of the conversion circuit of FIG.

FIG. 22 is a flowchart illustrating the operation of a code string decomposition circuit.

FIG. 23 is a block diagram illustrating a configuration example of a computer 101.

[Explanation of symbols]

Reference Signs List 1 conversion circuit, 2 signal component coding circuit, 3 code string generation circuit, 4 information recording medium, 11 band division filter, 12 forward spectrum conversion circuit, 13 forward spectrum conversion circuit, 21 normalization circuit, 22 quantization circuit, 23 Quantization accuracy determination circuit, 31 control unit,
32 encoding unit, 33 encoding unit, 41 code string decomposition circuit, 42 signal component decoding circuit, 43 inverse conversion circuit, 51 control unit, 52 decoding unit, 53 decoding unit, 61, 62 inverse spectrum conversion circuit, 63
Band synthesis filter

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Minoru Tsuji F-term in Sony Corporation 6-7-35 Kita-Shinagawa, Shinagawa-ku, Tokyo F-term (reference) 5D044 AB06 GL01 GL28 5D045 DA00 5J064 AA02 AA04 BA08 BA09 BA16 BC01 BC11 BC14 BC18 BC25 BD03

Claims (15)

[Claims] 1. A calculating means for detecting each numerical value constituting a numerical value sequence to be encoded and calculating a length in which specific values are arranged, and a first encoding for encoding the length in which the specific values are arranged Means, and second encoding means for encoding the numerical value other than the specific value. 2. The encoding apparatus according to claim 1, wherein the first encoding unit and the second encoding unit perform encoding alternately. 3. The encoding apparatus according to claim 1, wherein the encoded numerical sequence is a quantized acoustic waveform signal, and the specific value is a value of zero. 4. The encoding apparatus according to claim 1, wherein the first encoding unit encodes a length in which the specific values are arranged into a variable-length code string. 5. The encoding apparatus according to claim 1, wherein the second encoding unit encodes a numerical value other than the specific value into a variable-length code string. 6. The encoding apparatus according to claim 5, wherein the variable-length code string is formed of a Huffman code. 7. A calculating step of detecting each numerical value constituting a numerical value sequence to be encoded and calculating a length in which specific values are arranged, and a first encoding for encoding the length in which the specific values are arranged. And a second encoding step of encoding the numerical value other than the specific value. 8. A calculating step of detecting each numerical value constituting a numerical value sequence to be encoded and calculating a length in which specific values are arranged, and a first encoding for encoding the length in which the specific values are arranged. And a second encoding step of encoding the numerical value other than the specific value. A recording medium in which a computer-readable program is recorded. 9. A calculating step of detecting each numerical value constituting a numerical value sequence to be encoded and calculating a length in which specific values are arranged, and a first encoding for encoding the length in which the specific values are arranged. A computer-readable storage medium storing a program for causing a computer to execute a process including: a step of encoding a numerical value other than the specific value; 10. Decoding a code string having a length in which specific values are arranged,
A decoding device comprising: a determination unit that determines a code of a signal component based on a result thereof; and a decoding unit that decodes a code of the signal component. 11. When the value obtained by decoding a code string having a length in which the specific values are arranged is a value 0, the determining means is input next to the code string having a length in which the specific values are arranged. A code string having a value other than the specific value is decoded, and the value obtained as a result is determined as the sign of the signal component. If not 0,
11. The decoding device according to claim 10, wherein a sign of a signal component corresponding to the value is determined to the specific value. 12. The decoding device according to claim 10, wherein the specific value is a value 0. 13. Decoding a code string having a length in which specific values are arranged,
A decoding method, comprising: determining a code of a signal component based on the result; and decoding the code of the signal component. 14. Decoding a code string having a length in which specific values are arranged,
A recording medium in which a computer-readable program is recorded, comprising: a determining step of determining a code of a signal component based on a result thereof; and a decoding step of decoding a code of the signal component. 15. Decoding a code string having a length in which specific values are arranged,
A program for causing a computer to execute a process including a determining step of determining a sign of a signal component based on the result and a decoding step of decoding the sign of the signal component.