WO2009093714A1 - 符号化方法、復号化方法、それらの装置、及びそれらのプログラムと記録媒体 - Google Patents
符号化方法、復号化方法、それらの装置、及びそれらのプログラムと記録媒体 Download PDFInfo
- Publication number
- WO2009093714A1 WO2009093714A1 PCT/JP2009/051123 JP2009051123W WO2009093714A1 WO 2009093714 A1 WO2009093714 A1 WO 2009093714A1 JP 2009051123 W JP2009051123 W JP 2009051123W WO 2009093714 A1 WO2009093714 A1 WO 2009093714A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- correction coefficient
- normalized
- signal sequence
- quantization
- input signal
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 65
- 238000013139 quantization Methods 0.000 claims abstract description 237
- 108010076504 Protein Sorting Signals Proteins 0.000 claims abstract description 193
- 239000013598 vector Substances 0.000 claims abstract description 184
- 238000010606 normalization Methods 0.000 claims abstract description 89
- 238000012937 correction Methods 0.000 claims description 354
- 238000004364 calculation method Methods 0.000 claims description 15
- 238000006243 chemical reaction Methods 0.000 claims description 5
- 230000006870 function Effects 0.000 description 22
- 238000007796 conventional method Methods 0.000 description 9
- 238000004088 simulation Methods 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 3
- 230000005236 sound signal Effects 0.000 description 3
- 238000001228 spectrum Methods 0.000 description 3
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 2
- 239000004065 semiconductor Substances 0.000 description 2
- 230000001174 ascending effect Effects 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 230000017105 transposition Effects 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M7/00—Conversion of a code where information is represented by a given sequence or number of digits to a code where the same, similar or subset of information is represented by a different sequence or number of digits
- H03M7/30—Compression; Expansion; Suppression of unnecessary data, e.g. redundancy reduction
- H03M7/3082—Vector coding
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L19/00—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/189—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the adaptation method, adaptation tool or adaptation type used for the adaptive coding
- H04N19/192—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the adaptation method, adaptation tool or adaptation type used for the adaptive coding the adaptation method, adaptation tool or adaptation type being iterative or recursive
- H04N19/194—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the adaptation method, adaptation tool or adaptation type used for the adaptive coding the adaptation method, adaptation tool or adaptation type being iterative or recursive involving only two passes
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/85—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
- H04N19/94—Vector quantisation
Definitions
- the present invention relates to a method of encoding a digital signal such as an audio signal or an audio signal such as music or an image signal, a decoding method, a device thereof, a program thereof, and a recording medium.
- a sequence of input original sound samples is divided into input signal sequences of fixed intervals of about 5 to 50 ms called frames, and the input signal sequence for each frame is divided.
- the vector quantization is performed after dividing the normalized input signal sequence based on a predetermined rule. The method is known.
- FIG. 27 shows an encoding apparatus according to the encoding method
- FIG. 28 shows the decoding apparatus, and the operation will be briefly described.
- the encoding apparatus includes a normalization value generation unit 101, a normalization unit 102, a division unit 103, a vector quantization unit 104, and a normalization value quantization unit 105.
- the input signal series X may be a time-domain sample string, or may be a sample string based on a spectrum obtained by converting such a sample string into the frequency domain for each frame.
- the normalized value generation unit 101 outputs the square root of the average value of the power of the samples in the input signal series X as the normalized value G.
- Normalization value quantizer 105 a normalization value G obtained in the normalization value generator 101 quantizes and outputs it as a normalized value quantization index I G.
- the normalization unit 102 receives the input signal sequence X and the normalized value G obtained by the normalized value generation unit 101 or the decoded normalized value G ′ obtained by the normalized value quantization unit 105 as an input signal.
- the vector quantization unit 104 performs vector quantization for each of the divided input signal sequences that are the output of the dividing unit 103, and outputs a vector quantization index k i .
- the vector quantization unit 104 has a vector codebook 104T in which, for example, a predetermined representative vector value is associated with a finite number of indexes of 2 or more, and the vector quantization unit 104 receives the given divided input signal.
- An index k i corresponding to the representative vector value having the smallest distance scale with the sequence u i is output as a vector quantization index.
- the decoding apparatus includes a vector decoding unit 111, a reconstruction unit 112, a normalized value decoding unit 113, and a denormalization unit 114.
- the vector decoding unit 111 has the same vector codebook 111T as the vector quantization unit 104 of the encoding device, and the representative vector value corresponding to each vector quantization index k i given from the encoding device is represented by the vector codebook. By reading from 111T, k i is decoded and output as a divided output signal sequence v i .
- the reconstructing unit 112 uses the divided output signal sequence v i for one frame from the vector decoding unit 111, and the relationship between x and u i and the relationship between y and v i in the dividing unit 103 of the encoding device. Are reconstructed based on a predetermined rule for making the same.
- Normalization value decoding unit 113 outputs the differentially normalized value quantization index I G decoding to decode normalization value G 'sent from the encoding device.
- the denormalization unit 114 receives the reconstructed normalized value output signal sequence y and the decoded normalized value G ′ as input and multiplies the output signal sequence y by the decoded normalized value G ′ to denormalize. , And output as an output signal series Y.
- Patent Document 1 discloses that the frequency domain signal for each frame is normalized and the result is vector quantized.
- Non-Patent Document 1 shows that in CELP encoding, each frame of a time-domain input signal is divided into subframes, and a series of powers of those subframes are vector quantized.
- the decoding apparatus uses a vector quantization index obtained by vector quantization of a divided input signal sequence obtained by dividing the normalized input signal sequence in such an encoding device, and a normalized value obtained only from the input signal.
- SNR Signal-to-noise ratio
- the present invention has been made in view of these points, and an object thereof is to provide an encoding method, a decoding method, an apparatus thereof, a program thereof, and a recording medium with little error.
- the encoding method includes: A normalization for obtaining a normalized input signal sequence that is a sequence of samples obtained by normalizing the input signal in the frame with a normalized value for the input signal obtained from the input signal in the frame for each frame composed of a plurality of input signal samples. Conversion stage, A signal quantization stage for quantizing the normalized input signal sequence to generate a signal quantization index; A correction coefficient that generates a correction coefficient that minimizes the distance measure between the input signal series and the signal series obtained by denormalizing the signal series corresponding to the signal quantization index with the normalized value corrected by the correction coefficient.
- the encoding method is: A normalization for obtaining a normalized input signal sequence that is a sequence of samples obtained by normalizing the input signal in the frame with a normalized value for the input signal obtained from the input signal in the frame for each frame composed of a plurality of input signal samples.
- the first correction coefficient A correction coefficient generation stage for generating a value obtained by dividing the above by the second correction coefficient as the correction coefficient;
- the decoding method comprises: A normalized value decoding process for decoding the input normalized value quantization index to generate a normalized value for each frame; A vector decoding process for decoding the input signal quantization index to generate a normalized output signal sequence for each frame; A correction coefficient decoding process for decoding the input correction coefficient quantization index to generate a correction coefficient for each frame; A normalized value correction process for correcting the normalized value with the correction coefficient and generating a corrected normalized value; A denormalization process in which the normalized output signal sequence is denormalized with the corrected normalized value to generate an output signal for each frame; including.
- the input signal obtained for each frame is minimized so as to minimize the encoding error from the input / output relationship when generating a vector quantization index in encoding.
- movement flow of the encoding apparatus of FIG. The figure which shows the operation example of the division part 103 in FIG.
- movement flow of the encoding apparatus of FIG. The figure which shows the function structural example of 3rd Example of the encoding apparatus by this invention.
- the figure which shows the function structure of the decoding apparatus corresponding to the encoding apparatus of FIG. The figure which shows the function structural example of 6th Example of the encoding apparatus by this invention.
- the figure which shows the function structure of the decoding apparatus corresponding to the encoding apparatus of FIG. The figure which shows the function structural example of 10th Example of the encoding apparatus by this invention.
- the figure which shows the function structure of the decoding apparatus corresponding to the encoding apparatus of FIG. The figure which shows the function structural example of 11th Example of the encoding apparatus by this invention.
- the figure which shows the function structure of the decoding apparatus corresponding to the encoding apparatus of FIG. The figure which shows the function structural example of 14th Example of the encoding apparatus by this invention.
- the figure which shows the function structure of the decoding apparatus corresponding to the encoding apparatus of FIG. The figure which shows schematic functional structure of the experimental apparatus which performed simulation.
- the normalization value G generated by the normalization value generation unit 101 is defined, for example, as in Expression (1).
- the normalization value G defined by equation (1) is the square root of the average value of the power of the input signal for each frame. Further, for the purpose of simplifying the description, a case where there is no division step will be described here.
- the idea is to obtain an output signal sequence in which the error is minimized by giving or multiplying the normalized output signal sequence y by the correction coefficient ⁇ . Therefore, the error d ′ representing the distance measure between the input signal sequence of the encoding device and the output signal sequence of the decoding device in each frame of the encoding method according to the present invention can be expressed by the following equation (3).
- ⁇ that minimizes the error d ′ can be obtained by obtaining ⁇ that satisfies the following equation (4).
- the encoding error d ′ according to the present invention is smaller than or equal to the encoding error d according to the conventional method. Therefore, encoding and decoding with less error and improved SNR can be realized by performing encoding and decoding in consideration of the corrected normalized value G * .
- FIG. 1 shows an example of a functional configuration of a first embodiment of an encoding apparatus 10 using the encoding method according to the present invention
- FIG. 2 shows an operation flow thereof.
- Each part of the apparatus operates for each frame of a plurality of samples.
- the input signal for each frame may be a time domain signal or a frequency domain signal.
- the input signal is a frequency domain signal.
- the encoding device includes a normalization value generation unit 101, a normalization unit 102, a division unit 103, a vector quantization unit 104, a normalization value quantization unit 105, and a normalization value correction unit 20.
- the encoding apparatus has a configuration in which a normalized value correction unit 20 is added to the configuration of the encoding apparatus according to the conventional method shown in FIG.
- the processing units given the same symbols as in FIG. 27 perform the same operations as in FIG.
- the encoding apparatus according to this embodiment may be realized by reading a predetermined program into a computer including, for example, a ROM, a RAM, and a CPU and executing the program by the CPU.
- the normalized value generation unit 101 outputs the square root of the average value of the power of each input signal sample in the input signal series X as the normalized value G.
- the average value of the power of each input signal sample for example, an average of absolute values may be used as the normalized value G.
- the standard deviation of the input signal per frame may be used as the normalized value.
- the normalized input signal sequence x is a signal in the frequency domain, and is composed of 16 spectral samples x 0 ,..., X 15 in which the normalized input signal sequence x of one frame is arranged, for example, in ascending order of frequency.
- h (i) represents the number of samples of the i-th divided input signal sequence u i .
- One frame of normalized input signal sequence x composed of 16 frequency spectra is divided into four divided input signal sequences every four samples.
- the vector quantization unit 104 has a vector code book 104T in which a predetermined representative vector is associated with each of two or more finite indexes, and a divided input signal sequence u that is an input vector to be quantized.
- the index k i corresponding to the representative vector that minimizes the distance measure to i is output as a vector quantization index.
- divided input signal sequence u each sample u i of i, j and the representative vector w i of each sample w i for example, the following equation as the distance measure between j of
- the vector quantization index k i associated with w i that minimizes the representative vector w i corresponding to k i is the divided output signal sequence v i.
- vector quantization is described as an example.
- the signal quantization processing generates an index for obtaining a divided output signal sequence v i corresponding to the divided input signal sequence u i
- the divided input signal The sequence u i may perform processing other than vector quantization such as scalar quantization of each sample and output of a corresponding index. That is, it can be said that the vector quantization unit 104 is a signal quantization unit when generalized.
- the normalized value correcting unit 20 receives the divided input signal sequence u i output from the dividing unit 103 and the vector quantization index k i output from the vector quantizing unit 104 as an input so that the error is minimized. Is corrected (step S20).
- the correction coefficient ⁇ for correcting the normalization value G is the total division output obtained by decoding the vector quantization index k i for the total division input signal sequence u i in the frame multiplied by the normalization value G.
- the signal sequence v i is generated based on the idea shown in the above equation (5) so as to minimize the error of the product of the corrected normalized value G * .
- the normalized value G output from the normalized value generation unit 101 is corrected with the correction coefficient ⁇ and output.
- the normalized value quantization unit 105 quantizes the corrected normalized value G * and outputs it as a normalized quantization index I G ′ (step S105).
- the normalized value correction unit 20 includes a vector decoding unit 21, a correction coefficient generation unit 22, and a correction calculation unit 23.
- the vector decoding unit 21 has the same vector codebook 111T as the vector codebook 104T in the encoding device, and refers to the vector codebook 111T to generate a divided output signal sequence (representative vector) corresponding to the vector quantization index k i. ) Decode and output v i .
- a divided output signal sequence representedative vector
- k i the vector quantization index
- v i the vector quantization indexes k 0 ⁇ k 3 as illustrated in FIG. 3
- the correction coefficient generation unit 22 calculates the normalized value correction coefficient ⁇ using the calculation formula shown in Expression (10) with the divided input signal series u i and the divided output signal series v i as inputs (step S22).
- the correction calculator 23 multiplies the normalized value G output from the normalized value generator 101 by the normalized value correction coefficient ⁇ output from the correction coefficient generator 22 as a corrected normalized value G *.
- the normalized value is corrected by (step S23).
- the signal corresponding to the normalized input signal in this embodiment, the signal corresponding to the divided input signal sequence u i and vector quantization, in this embodiment, the vector quantization result
- the normalized value that minimizes the distance measure between the input signal to be encoded and the decoded output signal as represented by Equation (10) based on the divided output signal sequence v i obtained by decoding A correction coefficient ⁇ is calculated.
- the normalized value G * is corrected by correcting the normalized value G using the normalized value correction coefficient ⁇ , and the quantized index I G ′ is output together with the vector quantized index k i , or described later.
- the normalized value G and the correction coefficient ⁇ that are not corrected are quantized, and the quantization indexes I G and I ⁇ are output together with the vector quantization index k i . Therefore, encoding with fewer errors than before can be performed.
- the configuration of the decoding apparatus corresponding to the encoding apparatus in FIG. 1 may be exactly the same as that of the decoding apparatus 110 in the conventional method shown in FIG.
- the normalized value quantization index I G ′ is decoded by the normalized value decoding unit 113 instead of the decoded normalized value G ′ corresponding to the uncorrected normalized value G in the conventional method.
- the decoding apparatus 110 can perform decoding with good SNR.
- the correction coefficient generation unit 22 obtains the normalized value correction coefficient ⁇ from the divided divided input signal series u i and the divided output signal series v i decoded as it is. Therefore, it is not necessary to reconfigure the divided input signal sequence to the signal sequence before division. Therefore, the amount of calculation can be reduced as compared with the reconfiguration method.
- the vector quantization unit 104 in the encoding apparatus described in FIG. The following formula (11) is obtained by expanding the formula (9).
- w i, j is the sum of squares (ie, the sum of the squares of the vectors), regardless of the divided input signal sequence u i , all representative vectors stored in the vector codebook 104T. Can be calculated in advance and stored in the vector codebook 104T.
- the square sum of the divided input signal series u i of the first item is a given fixed value.
- the vector quantization unit 104 searches for the representative vector w i that minimizes the error d i with respect to the divided input signal sequence u i, the second term representing the inner product of the vectors of the equation (11) and the square sum of the vectors What is necessary is just to search for the representative vector w i that minimizes the sum with the third term representing.
- the second term when w i the error d i is minimized is determined as divided output signal sequence v i (factor -2 excluded) and the third term the following equation
- FIG. 4 shows a functional configuration of the encoding apparatus
- FIG. 5 shows an operation flow thereof.
- the configuration of this embodiment is the same as that of the embodiment of FIG. 1 except that the vector decoding unit 21 is removed and the second term and the second term of the equation (11) when the vector quantization index k i is determined from the vector quantization unit 104 are
- the three terms are given to the correction coefficient generator 22 as M P i and M Q i expressed by the equations (12a) and (12b), respectively.
- the correction coefficient generation unit 22 calculates the correction coefficient ⁇ as
- the operation flow of FIG. 5 eliminates the vector decoding process S21 in the normalized value correction step S20 in FIG. 2, and is an equation when the vector quantization index for the divided input signal sequence u i is determined in the vector quantization step S104.
- Other operations are the same as those in FIG.
- the configuration for generating the correction coefficient without performing the vector decoding described in the second embodiment can be applied to other embodiments described below.
- the conventional decoding apparatus shown in FIG. 28 can be used as it is.
- FIG. 6 shows a functional configuration of a third embodiment of the encoding apparatus using the encoding method according to the present invention
- FIG. 7 shows an operation flow thereof.
- the correction coefficient generator 22 in the first embodiment of FIG. 1 is composed of a first correction coefficient generator 22a and a second correction coefficient generator 22b.
- the correction calculation unit 23 in the first embodiment of FIG. 1 includes a first correction unit 23a and a second correction unit 23b.
- Other configurations are the same as those of the encoding apparatus of FIG.
- Formula (14) is the same as the numerator of formula (10) described above.
- Equation (15) is the same as the denominator of Equation (10) above.
- the first correction unit 23a multiplies the normalized value G obtained from the input signal output from the normalized value generation unit 101 by the first correction coefficient ⁇ 1 and outputs ⁇ 1 G (step S23a).
- the second correction unit 23b divides the normalized value ⁇ 1 G obtained by multiplying the first correction coefficient by the first correction unit 23a by the second correction coefficient ⁇ 2 and outputs the result as a corrected normalized value G * (step).
- the second correction unit 23b is a normalized value after dividing the second correction coefficient beta 2, the first correction portion 23a may be multiplied by the first correction coefficient beta 1 to the value. That is, the order of the processes in step S23a and step S23b may be reversed.
- FIG. 6 can perform encoding with a good SNR in the same manner as the encoding apparatus shown in FIGS. 1 and 4.
- the conventional decoding apparatus shown in FIG. 28 can be used as it is.
- FIG. 8 shows a functional configuration of a fourth embodiment of an encoding apparatus using the encoding method according to the present invention
- FIG. 9 shows an operation flow thereof.
- the normalization value correction unit 20 of the encoding device includes a vector decoding unit 21, a reconstruction unit 61, a correction coefficient generation unit 62, and a correction calculation unit 23.
- Other configurations are the same as those of the encoding apparatus of FIG. 1, FIG. 4 and FIG.
- the divided output signal sequences v 0 to v 3 divided every four shown in FIG. 3 are rearranged in the same order as the normalized input signal sequences x 0 to x 15 . That is, the operation reverse to that of the dividing unit 103 is performed to reconstruct the normalized output signal sequence y.
- the correction coefficient generation unit 62 receives the normalized input signal series x and the normalized output signal series y as inputs, and generates a correction coefficient ⁇ using the calculation formula shown in Expression (16) (Step S62).
- N-1 ⁇ is a normalized input signal sequence
- subscript n is normal
- the sample number N in the normalized signal sequence represents the number of samples included in the normalized signal sequence, that is, the frame length.
- Such an encoding apparatus having the configuration shown in FIG. 8 can perform encoding with a good SNR as in the encoding apparatuses shown in FIGS.
- the conventional decoding apparatus shown in FIG. 28 can be used as it is.
- the normalized index G G ′ of the normalized value G * corrected by correcting the normalized value G with the correction coefficient ⁇ or the first correction coefficient ⁇ 1 and the second correction coefficient ⁇ 2 is obtained.
- the normalized value G is quantized as it is without being corrected, and the quantization index is output as well as the quantization index of the correction coefficient ⁇ , which is decoded on the decoding side.
- the normalized value G ′ may be corrected with the correction coefficient ⁇ ′.
- FIG. 10 shows a functional configuration of an embodiment of such an encoding apparatus.
- This fifth embodiment is the same as the embodiment of FIG. 1 except that the correction calculation unit 23 is removed and the normalized value G is directly supplied to the normalized value quantization unit 105, and a correction coefficient quantization unit 106 is provided to generate a correction coefficient.
- the correction coefficient ⁇ generated by the unit 22 is quantized by the correction coefficient quantization unit 106.
- the normalized value quantization unit 105 quantizes the given normalized value G and outputs a normalized value quantization index IG.
- the correction coefficient quantization unit 106 quantizes the given correction coefficient ⁇ and outputs a correction coefficient quantization index I ⁇ .
- the indexes k, I G and I ⁇ generated by this encoding device are given to the decoding device.
- FIG. 11 shows a functional configuration of a decoding apparatus corresponding to the encoding apparatus of FIG.
- This decoding apparatus is obtained by adding a correction coefficient decoding unit 115 and a normalized value correction unit 116 to the conventional configuration of FIG.
- the correction coefficient decoding unit 115 decodes the given correction coefficient quantization index I ⁇ to generate a decoding correction coefficient ⁇ ′, and provides it to the normalized value correction unit 116.
- the normalized value correction unit 116 generates a decoded normalized value G * ′ by multiplying the decoded normalized value G ′ by the decoding correction coefficient ⁇ ′, and supplies the decoded normalized value G * ′ to the inverse normalizing unit 114.
- the denormalization unit 114 generates the output signal Y by multiplying the normalized output signal sequence y reconstructed by the reconstruction unit 112 by the corrected normalized normalization value G * ′.
- Other operations are the same as those in FIG.
- a corrected decoded normalized value G * ′ ⁇ ′G ′, which is the product of the decoded decoded normalized value G ′ and the decoded decoded correction coefficient ⁇ ′, is reconstructed.
- the corrected decoding normalization value G * ′ is multiplied by the output v i of the vector decoding unit 111, and the multiplication result is input to the reconstruction unit 112. Also good.
- one of the decoding correction coefficient ⁇ ′ and the decoded normalized value G ′ may be multiplied by v i and supplied to the reconstruction unit 112, and the other may be multiplied by the output of the reconstruction unit 112 to be the output signal Y. This is also true for other decoding devices.
- FIG. 12 shows a functional configuration of the sixth embodiment of the encoding apparatus.
- the correction coefficient calculation method in FIG. 6 is applied to the embodiment in FIG. 10, and instead of the correction coefficient generator 22 in FIG. 10, a first correction coefficient generator 22A and a second correction coefficient generator are generated.
- a unit 22B and a division unit 22C are provided.
- the first and second correction coefficient generation units 22A and 22B calculate the first correction coefficient ⁇ 1 and the second correction coefficient ⁇ 2 according to the equation (14), respectively.
- the division unit 22C generates the correction coefficient ⁇ by dividing the first correction coefficient ⁇ 1 by the second correction coefficient ⁇ 2 and supplies the correction coefficient ⁇ to the correction coefficient quantization unit 106.
- Other operations are the same as those in FIG. Further, the indexes k, I G , and I ⁇ output by the encoding device in FIG. 12 can be decoded by the decoding device in FIG.
- FIG. 13 shows a functional configuration of the seventh embodiment of the encoding apparatus.
- the correction coefficient ⁇ is obtained from the first and second correction coefficients ⁇ 1 and ⁇ 2 , the correction coefficient ⁇ is quantized, and the correction coefficient quantization index I ⁇ is output.
- the correction coefficient quantization index I ⁇ is obtained directly from the first and second correction coefficients ⁇ 1 and ⁇ 2 .
- the quantization correction coefficient ⁇ ′ corresponding to the correction coefficient quantization index I ⁇ is determined so that the error d ′ shown in the expression (3) is minimized, but the expression (3) can be rewritten as the following expression. it can.
- ⁇ ′ can be determined so that the error d ′ in Equation (17) is minimized.
- Equation (18) can be rewritten as the following equation using Equation (14) and (15).
- the configuration of the encoding device in FIG. 13 is the same as the configuration in FIG. 12 except that the division unit 22C is removed and a correction coefficient quantization unit 106a is provided instead of the correction coefficient quantization unit 106.
- the correction coefficient quantization unit 106a is provided with a correction coefficient codebook 106Ta that holds a plurality of predetermined quantization correction coefficients ⁇ ′ in association with indexes I ⁇ .
- the first and second correction coefficients ⁇ 1 and ⁇ 2 from the first and second correction coefficient generation units 22A and 22B are given to the correction coefficient quantization unit 106a.
- the correction coefficient quantization unit 106a searches the correction coefficient codebook 106Ta for a quantization correction coefficient ⁇ ′ that minimizes the error d ′′ in Expression (19), determines the quantization coefficient, and outputs the correction coefficient quantization index I ⁇ .
- the operation of is the same as that of Fig. 12.
- the decoding device of Fig. 11 can be used as a decoding device corresponding to the encoding device of Fig. 13, the decoding device of Fig. 11 can be used.
- FIG. 14 shows a modified embodiment of the encoding apparatus of FIG. 13, and the correction coefficient generation unit 22 described in FIG. 4 instead of the vector decoding unit 21 and the first and second correction coefficient generation units 22A and 22B in FIG. And the values P i and Q i of equations (12a) and (12b) calculated when the vector quantization unit 104 vector quantizes the divided input signal sequence u i are given to the correction coefficient generation unit 22.
- the correction coefficient generation unit 22 calculates the first and second correction coefficients ⁇ 1 and ⁇ 2 from P i and Q i by the following equation
- FIG. 15 shows a functional configuration of the ninth embodiment of the encoding apparatus.
- first and second correction coefficient quantization units 106A and 106B are provided instead of the division unit 22C and the correction coefficient quantization unit 106 in FIG.
- the first and second correction coefficient quantizing units 106A and 106B quantize the first and second correction coefficients ⁇ 1 and ⁇ 2 given from the first and second correction coefficient generating units 22A and 22B, respectively.
- 1 and second correction coefficient quantization indexes I ⁇ 1 and I ⁇ 2 are output.
- Other operations are the same as those in FIG.
- FIG. 16 shows a functional configuration of a decoding apparatus corresponding to the encoding apparatus of FIG.
- a first correction coefficient decoding unit 115A, a second correction coefficient decoding unit 115B, and a division unit 115C are provided instead of the correction coefficient decoding unit 115 in the decoding apparatus of FIG.
- the first and second correction coefficient quantization indexes I ⁇ 1 and I ⁇ 2 given from the encoding device are decoded by the first and second correction coefficient decoding units 115A and 115B, respectively, and the first and second decoding corrections are performed.
- Coefficients ⁇ ′ 1 and ⁇ ′ 2 are output.
- the division unit 115 ⁇ / b> C generates a decoding correction coefficient ⁇ ′ by dividing ⁇ ′ 1 by ⁇ ′ 2 and supplies the decoded correction coefficient ⁇ ′ to the normalized value correction unit 116.
- the calculation of the normalized value correcting unit 116 and the denormalizing unit 114 as a whole is obtained by multiplying each sample of v i by ⁇ ′ 1 G ′ / ⁇ ′ 2 as each sample y n of Y. In this way, any calculation order may be used. Other operations are the same as those in FIG.
- the correction coefficient ⁇ or the first and second correction coefficients ⁇ 1 and ⁇ 2 are determined for each frame in the encoding device. Therefore, the decoding device corrects the normalized value for each frame. An example to do is shown.
- this coding apparatus of the tenth embodiment for each divided input signal sequence u i decided correction coefficient gamma i, corresponding correction to the divided output signal sequence v i for each divided output signal sequence v i in the decoding device The error due to normalization is reduced by multiplying the coefficient ⁇ i .
- FIG. 17 shows the functional configuration of the tenth embodiment of the encoding apparatus, which is the correction coefficient generation unit 22 and the correction coefficient quantization unit 106 shown in FIG. 10, respectively, and the correction coefficient sequence generation unit 22a and the correction coefficient sequence quantization unit 106b. It has been replaced with.
- the correction coefficient sequence quantization unit 106b has a correction coefficient sequence codebook 106Tb in which the correction coefficient representative vector ⁇ ′ is associated with the index I ⁇ .
- the correction coefficient representative vector ⁇ ′ i that minimizes the error ⁇ expressed by the above is searched from the correction coefficient string codebook 106Tb, and the corresponding index I ⁇ is output.
- FIG. 18 shows a functional configuration of a decoding apparatus corresponding to the encoding apparatus of FIG.
- This decoding apparatus has a configuration in which a correction coefficient sequence decoding unit 115b is added to the decoding apparatus in FIG. 28, and a divided output sequence correction unit 117 is further provided between the vector decoding unit 111 and the reconstruction unit 112. Yes.
- the correction coefficient sequence decoding unit 115b decodes the input correction coefficient sequence quantization index I ⁇ i or the correction coefficient vector quantization index I ⁇ to generate a sequence of decoded correction coefficient ⁇ ′ i , and the divided output sequence correction unit 117 To give.
- the correction coefficient sequence codebook 115Tb that is the same as the correction coefficient sequence codebook 106Tb used for the quantization is the correction coefficient sequence decoding unit 115a. Is provided.
- the divided output sequence correction unit 117 multiplies each divided output signal sequence v i by the corresponding decoding correction coefficient ⁇ i and gives the multiplication result to the reconfiguration unit 112. Other operations are the same as those in FIG.
- FIG. 19 shows a functional configuration of an eleventh embodiment of the encoding apparatus.
- the correction coefficient sequence generation unit 22a in the embodiment of FIG. 17 is removed, and a divided input signal sequence x i, j is used by using the correction coefficient sequence quantization unit 106c instead of the correction coefficient sequence quantization unit 106b.
- the correction coefficient vector quantization index I ⁇ corresponding to the column of the quantization correction coefficient ⁇ ′ i that directly minimizes the error is obtained with reference to the correction coefficient string codebook 106Tc from the corresponding divided output signal sequence v i .
- the correction coefficient sequence codebook 106Tc of the correction coefficient sequence quantizer 106c are held by a plurality of correction coefficient representative vectors corresponding to the correction coefficient vector quantization index I gamma, respectively, the correction coefficient sequence quantizer 106c Is the following equation (22) from the given divided input signal sequence u i and the corresponding divided output signal sequence v i
- the correction coefficient vector quantization index I ⁇ corresponding to the column of the quantization correction coefficient ⁇ ′ i that minimizes the error ⁇ represented by is determined with reference to the correction coefficient string codebook 106Tc.
- the correction coefficient string codebook 106Tc instead of the equation (22),
- the correction coefficient vector quantization index I ⁇ corresponding to the column of ⁇ ′ i that minimizes may be searched from the correction coefficient sequence codebook 106Tc. Other operations are the same as those in FIG. As the decoding apparatus corresponding to the encoding apparatus in FIG. 19, the decoding apparatus in FIG. 18 can be used.
- FIG. 20 shows a functional configuration of the twelfth embodiment of the encoding apparatus.
- the embodiment of FIG. 20 is a modified embodiment of the encoding apparatus of FIG. If equation (23) is transformed using equations (12a) and (12b),
- P i and Q i are calculated at the time of vector quantization of the divided input signal sequence u i in the vector quantization unit 104. Therefore, in this embodiment, the vector decoding unit 21 is not used, and P i and Q i are supplied from the vector quantization unit 104 to the correction coefficient sequence quantization unit 106d, and the same correction coefficient sequence codebook 106Tc as in FIG. Is used to output the correction coefficient vector quantization index I ⁇ corresponding to the column of the quantization correction coefficient ⁇ ′ i that minimizes the error ⁇ ′ in Expression (24).
- Other operations are the same as those in FIG. 19, and the same decoding apparatus as the correction coefficient string codebook 106Tc in FIG. 20 is used as the correction coefficient string codebook 115Tb in the decoding apparatus in FIG. Good.
- [Thirteenth embodiment] 17, 19, and 20 obtain a correction coefficient ⁇ i for each divided signal sequence and output the quantization index.
- the correction coefficient for each divided output signal sequence v i The case where the correction by is performed is shown.
- to generate a divided input signal sequence u i divided input normalization value g i for each, thereby respectively normalizing the divided input signal sequence u i, to minimize errors due to the normalization A column of correction coefficients is generated as follows.
- FIG. 21 shows the functional configuration of the thirteenth embodiment of the encoding apparatus. Similar to the encoding apparatus of each of the above-described embodiments, a normalization value generation unit 101, a normalization unit 102, a division unit 103, and a vector quantization unit 104 A normalized value quantization unit 105 is provided. In this embodiment, a split input normalization unit 109 is further provided between the split unit 103 and the vector quantization unit 104, and a split input for generating a split input normalization value g i to be given to the split input normalization unit 109.
- a normalization value generation unit 108, a correction coefficient sequence generation unit 22a that generates a sequence of correction coefficients ⁇ i , a normalization value correction unit 23 ′, and a correction normalization value quantization unit 105 ′ are provided.
- the divided input normalized value generation unit 108 calculates, for example, the following equation from the divided input signal sequence u i
- the divided input normalized value g i is calculated by The divided input normalization unit 109 outputs a sequence of samples obtained by dividing each sample of the divided input signal system u i by the divided input normalized value g i as a normalized divided input signal sequence u ′.
- the correction coefficient sequence generation unit 22 a generates a sequence of correction coefficients ⁇ i from the normalized divided input signal sequence u ′ i and the normalized divided output signal sequence v ′ i from the vector decoding unit 21.
- the column calculation method of the correction coefficient ⁇ i for example, the calculation method according to the equation (20) described in FIG. 17 can be used. However, to use the u i, v u instead of the i 'i, v' i.
- the normalized value correction unit 23 ′ generates a corrected divided input normalized value g * i by multiplying the divided input normalized value g i by each column of the correction coefficients ⁇ i .
- the correction normalization value quantization unit 105 ′ quantizes the correction division input normalization value g * i in the same manner as the correction coefficient string quantization unit 106b in FIG. 17, and the corresponding correction normalization value quantization index I g *. i is output.
- FIG. 22 shows a functional configuration of a decoding apparatus corresponding to the encoding apparatus of FIG.
- a divided output sequence inverse normalization unit 119 is provided between the vector decoding unit 111 and the reconstruction unit 112, and a corrected normalized value decoding unit 113 ′ is further added. It has a configuration.
- the corrected normalized value decoding unit 113 ′ decodes the input corrected normalized value quantization index I g * i and outputs a decoded corrected normalized value g * ′ i .
- the corrected normalized value decoding unit 119 multiplies the divided output signal sequence v i by the decoded corrected normalized value g * ′ i and gives the result to the reconstructing unit 112.
- Other operations are the same as those in FIG.
- FIG. 23 shows a functional configuration of the fourteenth embodiment of the encoding apparatus.
- the division input normalized value g i is not corrected in the embodiment of FIG. 21, but the columns of the divided input normalized value g i and the correction coefficient ⁇ i are respectively quantized and their indexes are output. It is a thing. Therefore, a divided input normalized value quantization unit 105 "and a correction coefficient sequence quantization unit 106 'are provided instead of the normalized value correction unit 23' and the corrected normalized value quantization unit 105 'in FIG. divided input normalization value quantizer 105 "quantizes the divided input normalization value g i, and outputs the quantization index I gi.
- the correction coefficient sequence quantization unit 106 ′ has a correction coefficient sequence code book 106Tc ′ that holds a plurality of correction coefficient representative vectors in association with the correction coefficient quantization indexes, respectively. Quantize the sequence of coefficients ⁇ i and output the corresponding index I ⁇ i . Other operations are the same as those in FIG.
- FIG. 24 shows a functional configuration of a decoding apparatus corresponding to the encoding apparatus of FIG. In this configuration, instead of the corrected normalized value decoding unit 113 ′ in the configuration of FIG. 22, a divided input normalized value decoding unit 113 ′′, a correction coefficient sequence decoding unit 115 ′, and a divided normalized value correction unit 116 ′.
- the divided input normalized value decoding unit 113 ′′ decodes the input index I gi and outputs a decoded divided input normalized value g ′ i .
- the correction coefficient sequence decoding unit 115 ′ has the same correction coefficient sequence codebook 115Tc ′ as the correction coefficient sequence codebook 106Tc ′ in the encoding device, and refers to the codebook 115Tc ′ for the input index I ⁇ i. Decode and output a sequence of decoding correction coefficients ⁇ ′ i . Divided normalization value corrector 116 'is decoded correction coefficient gamma' outputs 'correction divided input normalization value by multiplying the i g *' column decode divided input normalization value g a i i, divided output sequence inverse normalizer To the conversion unit 119. Other operations are the same as those in FIG.
- the present invention is characterized in that the normalized value is corrected by the encoding device or the decoding device.
- the encoding apparatus in the system that corrects the normalization value by the encoding apparatus employs a configuration in which the corrected normalization value is quantized by the normalization value quantization unit.
- the normalized value is quantized by the normalized value quantizing unit or the normalized value quantizing unit and the divided input normalized value quantizing unit, and the correction coefficient is corrected by the correction coefficient.
- a configuration in which quantization is performed by the quantization unit or the correction coefficient sequence quantization unit is employed.
- the normalized value quantization unit, the divided input normalized value quantization unit, the correction coefficient quantization unit, and the correction coefficient sequence quantization unit can be collectively referred to as a normalized information quantization unit.
- the divided input normalization value and the correction coefficient can be referred to as normalization information.
- FIG. 25 shows a schematic configuration of the experimental apparatus 80 used for the simulation.
- the experimental device 80 includes an encoding device 81 and a decoding device 82.
- the encoding device 81 divides the input signal into a low-frequency signal and a high-frequency signal by the band divider 81a, and then encodes the low-frequency signal in the time domain (low-frequency encoder 81b), Encoding is performed in the region (high frequency encoder 81c).
- the encoding method according to the present invention shown in FIG. 1 is applied to the high-pass encoder 81c.
- the decoding device 82 decodes the low frequency code and the high frequency code by the conventional method shown in FIG.
- This experimental device 80 was given an input signal in which a 57-second audio signal was converted to a discrete value by 16 kHz sampling, and the SNR after decoding was compared with the conventional method. The result is shown in FIG.
- the vertical axis represents SNR in (dB), and the horizontal axis represents the encoding method. It was confirmed that the SNR of the encoding method of the present invention was 11.9 dB, which is 0.6 dB higher than the conventional method of 11.3 dB.
- the band was divided into two due to experimental reasons.
- the encoding method of the present invention there is no need to limit the bandwidth.
- the encoding apparatus according to the present invention has been described as an example of operating in the frequency domain.
- the present invention is not limited to this example.
- the encoding method and apparatus of the present invention can also be applied to signal encoding in the time domain.
- the dividing unit 103 is configured by a filter bank, for example, but the technical idea of the present invention can be applied as it is.
- Each component in each of the embodiments of the encoding device and the decoding device described above may perform the processing by a dedicated processor such as a DSP (digital signal processor). Further, when the processing of the components in the above apparatus is realized by a computer, the processing contents of the functions that each apparatus should have are described by a program. Then, by executing this program on the computer, the processing means in each apparatus is realized on the computer.
- a dedicated processor such as a DSP (digital signal processor).
- the program describing the processing contents can be recorded on a computer-readable recording medium.
- a computer-readable recording medium for example, any recording medium such as a magnetic recording device, an optical disk, a magneto-optical recording medium, and a semiconductor memory may be used.
- a magnetic recording device a hard disk device, a flexible disk, a magnetic tape or the like, and as an optical disk, a DVD (Digital Versatile Disc), a DVD-RAM (Random Access Memory), a CD-ROM (Compact Disc Read Only) Memory), CD-R (Recordable) / RW (ReWritable), etc., magneto-optical recording medium, MO (Magneto Optical disc) etc., semiconductor memory, EEP-ROM (Electronically Erasable and Programmable-Read Only Memory) etc. Can be used.
- this program is distributed by selling, transferring, or lending a portable recording medium such as a DVD or CD-ROM in which the program is recorded. Further, the program may be distributed by storing the program in a recording device of a server computer and transferring the program from the server computer to another computer via a network.
- each means may be configured by executing a predetermined program on a computer, or at least a part of these processing contents may be realized by hardware.
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Theoretical Computer Science (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
Description
また、非特許文献1には、CELP符号化において時間領域入力信号の各フレームをサブフレームに分割し、それらのサブフレームの一連のパワーをベクトル量子化することが示されている。
複数の入力信号サンプルから成るフレーム毎に、フレーム内の入力信号から求めた上記入力信号に対する正規化値により上記フレーム内の入力信号を正規化したサンプルの系列である正規化入力信号系列を求める正規化段階と、
上記正規化入力信号系列を量子化して信号量子化インデックスを生成する信号量子化段階と、
入力信号の系列と、信号量子化インデックスに対応する信号系列を、補正係数で補正した正規化値で逆正規化して得られる信号系列と、の距離尺度が最小となる補正係数を生成する補正係数生成段階と、
上記補正係数と正規化値を、又は上記補正係数で補正された正規化値を、量子化して正規化情報量子化インデックスを生成する正規化情報量子化段階と、
少なくとも上記信号量子化インデックスと上記正規化情報量子化インデックスとを含む符号を出力する符号出力段階と、
を含む。
複数の入力信号サンプルから成るフレーム毎に、フレーム内の入力信号から求めた上記入力信号に対する正規化値により上記フレーム内の入力信号を正規化したサンプルの系列である正規化入力信号系列を求める正規化段階と、
上記正規化入力信号系列をフレーム毎に予め定められた規則に基づいて分割して分割入力信号系列を生成する分割段階と、
上記分割入力信号系列をベクトル量子化してベクトル量子化インデックスを生成するベクトル量子化段階と、
上記ベクトル量子化インデックスに対応した信号系列を分割出力信号系列として生成する復号化段階と、
上記分割入力信号系列と上記分割出力信号系列の内積値の総和を第1補正係数とし、上記分割出力信号系列のベクトルの2乗和の総和を第2補正係数としたとき、上記第1補正係数を上記第2補正係数で除算したものを上記補正係数として生成する補正係数生成段階と、
上記補正係数と正規化値を、又は上記補正係数で補正された正規化値を、量子化して正規化情報量子化インデックスを生成する正規化情報量子化段階と、
少なくとも上記ベクトル量子化インデックスと上記正規化情報量子化インデックスとを含む符号を出力する符号出力段階と、
を含む符号化方法。
入力された正規化値量子化インデックスを復号化してフレーム毎の正規化値を生成する正規化値復号過程と、
入力された信号量子化インデックスを復号化してフレーム毎の正規化出力信号系列を生成するベクトル復号化過程と、
入力された補正係数量子化インデックスを復号化してフレーム毎の補正係数を生成する補正係数復号化過程と、
上記正規化値を上記補正係数で補正し、補正された正規化値を生成する正規化値補正過程と、
上記正規化出力信号系列を上記補正された正規化値で逆正規化してフレーム毎の出力信号を生成する逆正規化過程と、
を含む。
〔発明の基本的な考え〕
実施例の説明の前に、この発明の符号化方法の基本的な考えを説明する。上記した正規化値生成部101で生成する正規化値Gを、例えば式(1)のように定義する。
ここでXはフレーム毎の入力信号サンプルX0, …, XN-1の列であり、Nはフレーム当たりのサンプル数である。よって式(1)で定義される正規化値Gは、フレーム毎の入力信号のパワーの平均値の平方根である。また、説明を簡単にする目的でここでは分割段階がない場合で説明する。正規化値Gで除算されて正規化された正規化入力信号系列をx={xn; n=0, …, N-1}、復号化装置における復号化後の正規化出力信号系列をy={yn; n=0, …, N-1}と表わす。
各フレームにおける正規化前の入力信号系列X=Gxと、復号化装置における逆正規化後の出力信号系列Y=Gyとの間の距離尺度を表す誤差dを求めると式(2)となる。
一方、この発明の符号化方法は、この誤差dが最小となるように復号化装置に与える正規化値Gを補正係数γで補正し、補正した正規化値G*=γGを復号化装置に与える、あるいは補正係数γを正規化出力信号系列yに乗ずることで誤差を最小化した出力信号系列を得る考えである。従って、この発明による符号化方法の各フレームでの符号化装置の入力信号系列と復号化装置の出力信号系列との間の距離尺度を表す誤差d'は次式(3)で表わせる。
図1にこの発明による符号化方法を用いた符号化装置10の第1実施例の機能構成例を示し、その動作フローを図2に示す。本装置の各部は複数サンプルによるフレーム毎に動作する。フレーム毎の入力信号は時間領域の信号であっても、周波数領域の信号であってもよいが、以降の説明では入力信号が周波数領域信号の場合で説明する。符号化装置は、正規化値生成部101と、正規化部102と、分割部103と、ベクトル量子化部104と、正規化値量子化部105と、正規化値補正部20とを備える。符号化装置は、図27に示した従来法による符号化装置の構成に正規化値補正部20を追加した構成である。図27と同じ記号を付してある処理部は、図27と同じ動作をする。この実施例の符号化装置は、例えばROM、RAM、CPU等で構成されるコンピュータに所定のプログラムが読み込まれて、CPUがそのプログラムを実行することで実現してもよい。
正規化値補正部20を更に詳しく説明する。正規化値補正部20は、ベクトル復号化部21と、補正係数生成部22と、補正演算部23とで構成される。
補正係数生成部22は、分割入力信号系列uiと、分割出力信号系列viを入力として式(10)に示す計算式で正規化値補正係数γを計算する(ステップS22)。
式(10)は、上記した入力信号系列X=Gxと出力信号系列Y=Gyの各フレームにおける誤差dを最小化する正規化値補正係数を求める式(5)と同じ意味である。
〔第2実施例〕
図1で説明した符号化装置におけるベクトル量子化部104は、
前記式(9)を展開した次式(11)
第2実施例はこのような考えに基づくものであり、その符号化装置の機能構成を図4に示し、その動作フローを図5に示す。この実施例の構成は図1の実施例において、ベクトル復号化部21を除去し、ベクトル量子化部104からベクトル量子化インデックスkiが決定された時の式(11)の第2項と第3項をそれぞれ式(12a),(12b)で表されるM個のPiとM個のQiとして補正係数生成部22に与える。補正係数生成部22は補正係数γを次式
図5の動作フローは図2における正規化値補正ステップS20内のベクトル復号化過程S21を除去し、ベクトル量子化段階S104で分割入力信号系列uiに対するベクトル量子化インデックスが決定されたときの式(11)の第2項と第3項をそれぞれ上記Pi, Qi (i=0,…, M-1)として保持し、補正係数生成過程S22で式(10)の代わりに式(13)により補正係数γを計算する。その他の動作は図2と同様である。
図6にこの発明による符号化方法を用いた符号化装置の第3実施例の機能構成を示し、その動作フローを図7に示す。この実施例においては図1の第1実施例における補正係数生成部22は、第1補正係数生成部22aと第2補正係数生成部22bとで構成される。また、この第3実施例においては、図1の第1実施例における補正演算部23が第1補正部23aと第2補正部23bとで構成される。その他の構成は図1の符号化装置と同様である。
第2補正係数生成部22bは、ベクトル復号化部21の出力する分割出力信号系列viを入力として、i=0,…,M-1のそれぞれの分割出力信号系列の全サンプルの2乗和の総和を式(15)に示す第2補正係数β2として生成する(ステップS22b)。
第1補正部23aは、正規化値生成部101が出力する入力信号から求めた正規化値Gに第1補正係数β1を乗算し、β1Gを出力する(ステップS23a)。第2補正部23bは、第1補正部23aで第1補正係数を乗算した正規化値β1Gを、第2補正係数β2で除算し、補正した正規化値G*として出力する(ステップS23b)。なお、第2補正部23bが、正規化値を第2補正係数β2で除算した後に、第1補正部23aがその値に第1補正係数β1を乗算してもよい。つまり、ステップS23aとステップS23bの処理は順序を逆転してもよい。
図8にこの発明による符号化方法を用いた符号化装置の第4実施例の機能構成を示し、その動作フローを図9に示す。符号化装置の正規化値補正部20は、ベクトル復号化部21と、再構成部61と、補正係数生成部62と、補正演算部23とで構成される。他の構成は上記した図1、図4及び図6の符号化装置と同じである。 再構成部61は、ベクトル復号化部21からの複数の分割出力信号系列viを入力として、予め定められた規則に基づいて正規化出力信号系列y={yn; n=0,…,N-1}を再構成する(ステップS61)。例えば、図3に示した4個置きに分割された分割出力信号系列v0~v3を、正規化入力信号系列x0~x15と同じ順番に並べ替える。つまり分割部103と逆の動作を行って正規化出力信号系列yを再構成する。
上述の各符号化装置では正規化値Gを補正係数γ、又は第1補正係数β1と第2補正係数β2で補正して補正された正規化値G*の量子化インデックスIG'を出力したが、以下の各実施例で示すように正規化値Gを補正せずそのまま量子化して量子化インデックスを出力すると共に、補正係数γの量子化インデックスも出力し、復号化側において復号化正規化値G'を補正係数γ'で補正するようにしてもよい。
図12は符号化装置の第6実施例の機能構成を示す。この実施例は図10の実施例に図6における補正係数の計算手法を適用したものであり、図10における補正係数生成部22の代わりに第1補正係数生成部22Aと、第2補正係数生成部22Bと、除算部22Cとが設けられている。第1及び第2補正係数生成部22A,22Bは図6の場合と同様に、式(14)により第1補正係数β1及び第2補正係数β2をそれぞれ計算する。除算部22Cは第1補正係数β1を第2補正係数β2で除算することにより補正係数γを生成し、補正係数量子化部106に与える。その他の動作は図10の場合と同様である。また、図12の符号化装置により出力されたインデックスk、IG, Iγは図11の復号化装置によって復号化することができる。
図13は符号化装置の第7実施例の機能構成を示す。図12の実施例では第1及び第2補正係数β1,β2から補正係数γを求め、その補正係数γを量子化し、その補正係数量子化インデックスIγを出力する場合を示したが、図13の実施例では第1及び第2補正係数γ1, γ2から直接補正係数量子化インデックスIγを求める。
ところで、補正係数量子化インデックスIγに対応する量子化補正係数γ'は式(3)で示す誤差d'が最小となるように決めるが、式(3)を次式のように書き直すことができる。
図14は図13の符号化装置の変形実施例であり、図13におけるベクトル復号化部21、第1及び第2補正係数生成部22A,22Bの代わりに図4で説明した補正係数生成部22が設けられ、ベクトル量子化部104において分割入力信号系列uiをベクトル量子化する際に計算された式(12a), (12b)の値Pi, Qiを補正係数生成部22に与える。補正係数生成部22は、Pi, Qiから第1及び第2補正係数β1, β2を次式
図15は符号化装置の第9実施例の機能構成を示す。この実施例は図12において除算部22Cと補正係数量子化部106の代わりに第1及び第2補正係数量子化部106A,106Bが設けられている。第1及び第2補正係数量子化部106A,106Bはそれぞれ第1及び第2補正係数生成部22A,22Bから与えられた第1及び第2補正係数β1, β2を量子化し、それらの第1及び第2補正係数量子化インデックスIβ1, Iβ2を出力する。その他の動作は図12の場合と同様である。
前述の各実施例では符号化装置に置いてフレーム毎に補正係数γ又は第1及び第2補正係数β1, β2を決め、従って、復号化装置においてはフレーム毎に正規化値の補正を行う例を示した。この第10実施例の符号化装置では、分割入力信号系列ui毎に補正係数γiを決め、復号化装置において分割出力信号系列vi毎にその分割出力信号系列viに対し対応する補正係数γiを乗算することにより正規化による誤差を小さくする。
図19は符号化装置の第11実施例の機能構成を示す。この実施例では、図17の実施例における補正係数列生成部22aを除去し、補正係数列量子化部106bの代わりに補正係数列量子化部106cを使って分割入力信号系列xi,jとそれに対応する分割出力信号系列viから直接誤差を最小にする量子化補正係数γ'iの列に対応する補正係数ベクトル量子化インデックスIγを補正係数列符号帳106Tcを参照して求める。即ち、補正係数列量子化部106cの補正係数列符号帳106Tcには複数の補正係数代表ベクトルがそれぞれ補正係数ベクトル量子化インデックスIγに対応して保持されており、補正係数列量子化部106cは与えられた分割入力信号系列uiとそれに対応する分割出力信号系列viから次式(22)
図20は符号化装置の第12実施例の機能構成を示す。図20の実施例は図19の符号化装置の変形実施例である。式(12a), (12b)を使って式(23)を変形すると次式
図17,19,20の符号化装置では分割信号系列毎に補正係数γiを求め、その量子化インデックスを出力し、図18の復号化装置では、分割出力信号系列viそれぞれに対し補正係数による補正を行う場合を示した。それに対し、この第13実施例では、分割入力信号系列ui毎に分割入力正規化値giを生成し、それによって分割入力信号系列uiをそれぞれ正規化し、その正規化による誤差を最小にするように補正係数の列を生成する。
分割入力正規化値生成部108は分割入力信号系列uiから例えば次式
図23は符号化装置の第14実施例の機能構成を示す。この実施例は図21の実施例において、分割入力正規化値giの補正を行わず、分割入力正規化値giと補正係数γiの列をそれぞれ量子化してそれらのインデックスを出力するようにしたものである。従って、図21における正規化値補正部23'と補正正規化値量子化部105'の代わりに分割入力正規化値量子化部105"と補正係数列量子化部106'が設けられている。分割入力正規化値量子化部105"は分割入力正規化値giを量子化し、その量子化インデックスIgiを出力する。補正係数列量子化部106'は複数の補正係数代表ベクトルを、それぞれ補正係数量子化インデックスに対応させて保持する補正係数列符号帳106Tc’を有しており、その符号帳を参照して補正係数γiの列を量子化し、対応するインデックスIγiを出力する。その他の動作は図21の場合と同様である。
この発明の符号化方法によるSNRを評価した。シミュレーションに用いた実験装置80の概略的な構成を図25に示す。実験装置80は符号化装置81と復号化装置82とで構成される。
Claims (24)
- 複数の入力信号サンプルから成るフレーム毎に、フレーム内の入力信号から求めた上記入力信号に対する正規化値により上記フレーム内の入力信号を正規化したサンプルの系列である正規化入力信号系列を求める正規化段階と、
上記正規化入力信号系列を量子化して信号量子化インデックスを生成する信号量子化段階と、
入力信号の系列と、信号量子化インデックスに対応する信号系列を、補正係数で補正した正規化値で逆正規化して得られる信号系列と、の距離尺度が最小となる補正係数を生成する補正係数生成段階と、
上記補正係数と正規化値を、又は上記補正係数で補正された正規化値を、量子化して正規化情報量子化インデックスを生成する正規化情報量子化段階と、
少なくとも上記信号量子化インデックスと上記正規化情報量子化インデックスとを含む符号を出力する符号出力段階と、
を含む符号化方法。 - 請求項1に記載の符号化方法において、
更に、上記信号量子化インデックスに対応した信号系列を正規化出力信号系列として生成する復号化段階を含み、
上記補正係数生成段階は、上記正規化入力信号系列と上記正規化出力信号系列の内積値を第1補正係数とし、上記正規化出力信号系列のベクトルの2乗和を第2補正係数としたとき、上記第1補正係数を上記第2補正係数で除算したものを上記補正係数として生成するものである。 - 請求項1に記載の符号化方法において、
更に、上記信号量子化インデックスに対応した信号系列を正規化出力信号系列として生成する復号化段階を含み、
上記補正係数生成段階は、上記正規化入力信号系列と上記正規化出力信号系列の内積値を第1の上記補正係数として生成する第1補正係数生成過程と、上記正規化出力信号系列のベクトルの2乗和を第2の上記補正係数として生成する第2補正係数生成過程とを含み、
更に、上記正規化値に上記第1の補正係数を乗じ上記第2の補正係数で除したものを上記補正された正規化値として生成する補正演算段階を含む。 - 請求項1に記載の符号化方法において、
上記補正係数生成段階は、上記正規化入力信号系列と上記正規化出力信号系列の内積値γ1と、上記正規化出力信号系列のベクトルの2乗和γ2とを上記補正係数として生成する過程を含み、
上記正規化情報量子化段階は、予め決めた複数の代表補正係数をそれぞれ補正係数量子化インデックスに対応させて保持する補正係数符号帳を参照し、次式
d"=-2γ'γ1+γ'2γ2
で表される距離尺度d"が最小となる量子化補正係数γ'に対応する補正係数量子化インデックスを探索し出力する補正係数量子化過程を含む。 - 請求項1記載の符号化方法において、
上記信号量子化段階は、予め作成された複数の代表ベクトルをそれぞれベクトル量子化インデックスに対応させて保持するベクトル符号帳を参照して、上記正規化入力信号系列と上記ベクトル符号帳の代表ベクトルとの内積と、上記ベクトル符号帳の代表ベクトルの2乗和とを使って求める距離尺度が最小となる代表ベクトルを探索する過程を含み、
上記補正係数生成段階は、上記信号量子化段階で上記距離尺度を最小としたときの上記内積を上記2乗和で除算したものを上記補正係数として生成するものである。 - 請求項1記載の符号化方法において、
上記信号量子化段階は、予め作成された複数の代表ベクトルをそれぞれベクトル量子化インデックスに対応させて保持するベクトル符号帳を参照して、上記正規化入力信号系列と上記ベクトル符号帳の代表ベクトルとの内積と、上記ベクトル符号帳の代表ベクトルの2乗和とを使って求める距離尺度が最小となる代表ベクトルを探索する過程を含み、
上記補正係数生成段階は、上記信号量子化段階で上記距離尺度を最小としたときの上記内積を第1の上記補正係数、上記2乗和を第2の上記補正係数として得るものであり、
更に、上記正規化値に上記第1の補正係数を乗じ上記第2の補正係数で除したものを上記補正された正規化値として生成する補正演算段階を含む。 - 請求項1記載の符号化方法において、上記信号量子化段階は、予め作成された複数の代表ベクトルをそれぞれベクトル量子化インデックスに対応させて保持するベクトル符号帳を参照して、上記正規化入力信号系列と上記ベクトル符号帳の代表ベクトルとの内積と、上記ベクトル符号帳の代表ベクトルの2乗和とを使って求める距離尺度が最小となる代表ベクトルを探索する過程を含み、
上記補正係数量子化段階は、上記ベクトル量子化段階で上記距離尺度を最小としたときの上記内積を第1係数γ1、上記2乗和を第2係数γ2とする組を上記補正係数として得るものであり、
上記正規化情報量子化段階は、予め決めた複数の量子化補正係数をそれぞれ補正係数量子化インデックスに対応させて保持する補正係数符号帳を参照し、次式
d"=-2γ'γ1+γ'2γ2
で表される距離尺度d"が最小となる量子化補正係数γ'に対応する補正係数量子化インデックスを探索し出力する補正係数量子化過程を含む。 - 請求項1乃至7のいずれか記載の符号化方法において、更に、上記正規化入力信号系列をフレーム毎に予め定められた規則に基づいて分割した分割入力信号系列を生成する分割段階を含み、生成した分割入力信号系列を上記正規化入力信号系列として信号量子化段階以降を行う。
- 複数の入力信号サンプルから成るフレーム毎に、フレーム内の入力信号から求めた上記入力信号に対する正規化値により上記フレーム内の入力信号を正規化したサンプルの系列である正規化入力信号系列を求める正規化段階と、
上記正規化入力信号系列をフレーム毎に予め定められた規則に基づいて分割して分割入力信号系列を生成する分割段階と、
上記分割入力信号系列をベクトル量子化してベクトル量子化インデックスを生成するベクトル量子化段階と、
上記ベクトル量子化インデックスに対応した信号系列を分割出力信号系列として生成する復号化段階と、
F上記分割入力信号系列と上記分割出力信号系列の内積値の総和を第1補正係数とし、上記分割出力信号系列のベクトルの2乗和の総和を第2補正係数としたとき、上記第1補正係数を上記第2補正係数で除算したものを上記補正係数として生成する補正係数生成段階と、
上記補正係数と正規化値を、又は上記補正係数で補正された正規化値を、量子化して正規化情報量子化インデックスを生成する正規化情報量子化段階と、
少なくとも上記ベクトル量子化インデックスと上記正規化情報量子化インデックスとを含む符号を出力する符号出力段階と、
を含む符号化方法。 - 複数の入力信号サンプルから成るフレーム毎に、フレーム内の入力信号から求めた上記入力信号に対する正規化値により上記フレーム内の入力信号を正規化したサンプルの系列である正規化入力信号系列を求める正規化部と、
上記正規化入力信号系列を量子化して信号量子化インデックスを生成する信号量子化部と、
入力信号の系列と、信号量子化インデックスに対応する信号系列を、補正係数で補正した正規化値で逆正規化して得られる信号系列と、の距離尺度が最小となる補正係数を生成する補正係数生成部と、
上記補正係数と正規化値を、又は上記補正係数で補正された正規化値を、量子化して正規化情報量子化インデックスを生成する正規化情報量子化部と、
少なくとも上記信号量子化インデックスと上記正規化情報量子化インデックスとを含む符号を出力する符号出力部と、
を含む符号化装置。 - 請求項10に記載の符号化装置において、
更に、上記信号量子化インデックスに対応した信号系列を正規化出力信号系列として生成する復号化部を含み、
上記補正係数生成部は、上記正規化入力信号系列と上記正規化出力信号系列の内積値を第1補正係数とし、上記正規化出力信号系列のベクトルの2乗和を第2補正係数としたとき、上記第1補正係数を上記第2補正係数で除算したものを上記補正係数として生成するように構成されている。 - 請求項10に記載の符号化装置において、
更に、上記信号量子化インデックスに対応した信号系列を正規化出力信号系列として生成する復号化部を含み、
上記補正係数生成部は、上記正規化入力信号系列と上記正規化出力信号系列の内積値を第1の上記補正係数として生成する第1補正係数生成部と、上記正規化出力信号系列のベクトルの2乗和を第2の上記補正係数として生成する第2補正係数生成部とを含み、
更に、上記正規化値に上記第1の補正係数を乗じ上記第2の補正係数で除したものを上記補正された正規化値として生成する補正演算部を含む。 - 請求項10に記載の符号化装置において、
上記補正係数生成部は、上記正規化入力信号系列と上記正規化出力信号系列の内積値γ1と、上記正規化出力信号系列のベクトルの2乗和の総和γ2とを上記補正係数として生成するように構成されており、
上記正規化情報量子化部は、予め決めた複数の代表補正係数をそれぞれ補正係数量子化インデックスに対応させて保持する補正係数符号帳を参照し、次式
d"=-2γ'γ1+γ'2γ2
で表される距離尺度d"が最小となる量子化補正係数γ'に対応する補正係数量子化インデックスを探索し出力する補正係数量子化部を含む。 - 請求項10記載の符号化装置において、
上記信号量子化部は、予め作成された複数の代表ベクトルをそれぞれベクトル量子化インデックスに対応させて保持するベクトル符号帳を参照して、上記正規化入力信号系列と上記ベクトル符号帳の代表ベクトルとの内積と、上記ベクトル符号帳の代表ベクトルの2乗和とを使って求める距離尺度が最小となる代表ベクトルを探索するように構成されており、
上記補正係数生成部は、上記信号量子化段階で上記距離尺度を最小としたときの上記内積を上記2乗和で除算したものを上記補正係数として生成するように構成されている。 - 請求項10記載の符号化装置において、
上記信号量子化部は、予め作成された複数の代表ベクトルをそれぞれベクトル量子化インデックスに対応させて保持するベクトル符号帳を参照して、上記正規化入力信号系列と上記ベクトル符号帳の代表ベクトルとの内積と、上記ベクトル符号帳の代表ベクトルの2乗和とを使って求める距離尺度が最小となる代表ベクトルを探索するように構成されており、
上記補正係数生成部は、上記信号量子化部で上記距離尺度を最小としたときの上記内積を第1の上記補正係数、上記2乗和を第2の上記補正係数として得るように構成されており、
更に、上記正規化値に上記第1の補正係数を乗じ上記第2の補正係数で除したものを上記補正された正規化値として生成する補正演算部が設けられている。 - 請求項10記載の符号化装置において、上記信号量子化部は、予め作成された複数の代表ベクトルをそれぞれベクトル量子化インデックスに対応させて保持するベクトル符号帳を参照して、上記正規化入力信号系列と上記ベクトル符号帳の代表ベクトルとの内積と、上記ベクトル符号帳の代表ベクトルの2乗和とを使って求める距離尺度が最小となる代表ベクトルを探索するように構成されており、
上記補正係数量子化部は、上記ベクトル量子化段階で上記距離尺度を最小としたときの上記内積を第1係数γ1、上記2乗和を第2係数γ2とする組を上記補正係数として得るように構成されており、
上記正規化情報量子化部は、予め決めた複数の量子化補正係数をそれぞれ補正係数量子化インデックスに対応させて保持する補正係数符号帳を参照し、次式
d"=-2γ'γ1+γ'2γ2
で表される距離尺度d"が最小となる量子化補正係数γ'に対応する補正係数量子化インデックスを探索し出力する補正係数量子化部を含む。 - 請求項10乃至16のいずれか記載の符号化装置において、更に、上記正規化入力信号系列をフレーム毎に予め定められた規則に基づいて分割した分割入力信号系列を生成する分割部を含み、生成した分割入力信号系列を上記正規化入力信号系列として出力する。
- 複数の入力信号サンプルから成るフレーム毎に、フレーム内の入力信号から求めた上記入力信号に対する正規化値により上記フレーム内の入力信号を正規化したサンプルの系列である正規化入力信号系列を求める正規化部と、
上記正規化入力信号系列をフレーム毎に予め定められた規則に基づいて分割して分割入力信号系列を生成する分割部と、
上記分割入力信号系列をベクトル量子化してベクトル量子化インデックスを生成するベクトル量子化部と、
上記ベクトル量子化インデックスに対応した信号系列を分割出力信号系列として生成する復号化部と、
上記分割入力信号系列と上記分割出力信号系列の内積値の総和を第1補正係数とし、上記分割出力信号系列のベクトルの2乗和の総和を第2補正係数としたとき、上記第1補正係数を上記第2補正係数で除算したものを上記補正係数として生成する補正係数生成部と、
上記補正係数と正規化値を、又は上記補正係数で補正された正規化値を、量子化して正規化情報量子化インデックスを生成する正規化情報量子化部と、
少なくとも上記ベクトル量子化インデックスと上記正規化情報量子化インデックスとを含む符号を出力する符号出力部と、
を含む符号化装置。 - 入力された正規化値量子化インデックスを復号化してフレーム毎の正規化値を生成する正規化値復号過程と、
入力された信号量子化インデックスを復号化してフレーム毎の正規化出力信号系列を生成するベクトル復号化過程と、
入力された補正係数量子化インデックスを復号化してフレーム毎の補正係数を生成する補正係数復号化過程と、
上記正規化値を上記補正係数で補正し、補正された正規化値を生成する正規化値補正過程と、
上記正規化出力信号系列を上記補正された正規化値で逆正規化してフレーム毎の出力信号を生成する逆正規化過程と、
を含む復号化方法。 - 請求項19の復号化方法において、
上記信号復号化過程は、
入力された信号量子化インデックスを復号化してフレーム毎に複数の分割出力信号系列を生成するものであり、
更に、上記複数の分割出力信号系列中のサンプルを予め定めた規則に基づいて並べたものを上記正規化出力信号系列とする再構成過程を含む。 - 入力された正規化値量子化インデックスを復号化してフレーム毎の正規化値を生成する正規化値復号部と、
入力された信号量子化インデックスを復号化してフレーム毎の正規化出力信号系列を生成するベクトル復号化部と、
入力された補正係数量子化インデックスを復号化してフレーム毎の補正係数を生成する補正係数復号化部と、
上記正規化値を上記補正係数で補正し、補正された正規化値を生成する正規化値補正部と、
上記正規化出力信号系列を上記補正された正規化値で逆正規化してフレーム毎の出力信号を生成する逆正規化部と、
を含む復号化装置。 - 請求項21の復号化装置において、更に、出力信号系列を予め決めた規則に基づいて上記正規化出力信号系列に再構成する再構成部を含む。
- 請求項10、18,21の何れかに記載した装置としてコンピュータを機能させるための装置プログラム。
- 請求項23に記載した何れかの装置プログラムを記録したコンピュータで読み取り可能な記録媒体。
Priority Applications (6)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
ES09703930T ES2409843T3 (es) | 2008-01-24 | 2009-01-23 | Método de codificación, método de decodificación, y dispositivo y programa para los mismos, y medio o soporte de registro o grabación |
EP20090703930 EP2234273B8 (en) | 2008-01-24 | 2009-01-23 | Coding method, decoding method, apparatuses thereof, programs thereof, and recording medium |
US12/812,549 US8724734B2 (en) | 2008-01-24 | 2009-01-23 | Coding method, decoding method, apparatuses thereof, programs thereof, and recording medium |
CN2009801023883A CN101911501B (zh) | 2008-01-24 | 2009-01-23 | 编码方法、解码方法及其装置、以及其程序和记录介质 |
JP2009550585A JP5097217B2 (ja) | 2008-01-24 | 2009-01-23 | 符号化方法、符号化装置、及びそのプログラムと記録媒体 |
CA 2711539 CA2711539C (en) | 2008-01-24 | 2009-01-23 | Coding method, decoding method, apparatuses thereof, programs thereof, and recording medium |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2008-013868 | 2008-01-24 | ||
JP2008013868 | 2008-01-24 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2009093714A1 true WO2009093714A1 (ja) | 2009-07-30 |
Family
ID=40901217
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2009/051123 WO2009093714A1 (ja) | 2008-01-24 | 2009-01-23 | 符号化方法、復号化方法、それらの装置、及びそれらのプログラムと記録媒体 |
Country Status (7)
Country | Link |
---|---|
US (1) | US8724734B2 (ja) |
EP (1) | EP2234273B8 (ja) |
JP (2) | JP5097217B2 (ja) |
CN (1) | CN101911501B (ja) |
CA (1) | CA2711539C (ja) |
ES (1) | ES2409843T3 (ja) |
WO (1) | WO2009093714A1 (ja) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2012005209A1 (ja) * | 2010-07-05 | 2012-01-12 | 日本電信電話株式会社 | 符号化方法、復号方法、装置、プログラム及び記録媒体 |
WO2012005211A1 (ja) * | 2010-07-05 | 2012-01-12 | 日本電信電話株式会社 | 符号化方法、復号方法、符号化装置、復号装置、プログラム、及び記録媒体 |
WO2012005210A1 (ja) * | 2010-07-05 | 2012-01-12 | 日本電信電話株式会社 | 符号化方法、復号方法、装置、プログラムおよび記録媒体 |
WO2012005212A1 (ja) * | 2010-07-05 | 2012-01-12 | 日本電信電話株式会社 | 符号化方法、復号方法、符号化装置、復号装置、プログラム、及び記録媒体 |
CN102812642A (zh) * | 2010-03-09 | 2012-12-05 | 日本电信电话株式会社 | 编码方法、解码方法、装置、程序及记录介质 |
JP2017138605A (ja) * | 2017-02-16 | 2017-08-10 | ノキア テクノロジーズ オーユー | ベクトル量子化 |
US10665247B2 (en) | 2012-07-12 | 2020-05-26 | Nokia Technologies Oy | Vector quantization |
Families Citing this family (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9202473B2 (en) * | 2011-07-01 | 2015-12-01 | Nokia Technologies Oy | Multiple scale codebook search |
US9437213B2 (en) | 2012-03-05 | 2016-09-06 | Malaspina Labs (Barbados) Inc. | Voice signal enhancement |
US9015044B2 (en) | 2012-03-05 | 2015-04-21 | Malaspina Labs (Barbados) Inc. | Formant based speech reconstruction from noisy signals |
US9384759B2 (en) | 2012-03-05 | 2016-07-05 | Malaspina Labs (Barbados) Inc. | Voice activity detection and pitch estimation |
EP3046105B1 (en) | 2013-09-13 | 2020-01-15 | Samsung Electronics Co., Ltd. | Lossless coding method |
KR102270106B1 (ko) | 2013-09-13 | 2021-06-28 | 삼성전자주식회사 | 에너지 무손실 부호화방법 및 장치, 신호 부호화방법 및 장치, 에너지 무손실 복호화방법 및 장치, 및 신호 복호화방법 및 장치 |
CN103729191B (zh) * | 2014-01-06 | 2015-08-19 | 中国南方电网有限责任公司 | 一种可灵活调节的编码生成方法及系统 |
EP3648103B1 (en) * | 2014-04-24 | 2021-10-20 | Nippon Telegraph And Telephone Corporation | Decoding method, decoding apparatus, corresponding program and recording medium |
US10418042B2 (en) * | 2014-05-01 | 2019-09-17 | Nippon Telegraph And Telephone Corporation | Coding device, decoding device, method, program and recording medium thereof |
US9838700B2 (en) * | 2014-11-27 | 2017-12-05 | Nippon Telegraph And Telephone Corporation | Encoding apparatus, decoding apparatus, and method and program for the same |
CA2991341A1 (en) | 2015-07-06 | 2017-01-12 | Nokia Technologies Oy | Bit error detector for an audio signal decoder |
EP3306817B8 (en) * | 2016-10-07 | 2021-04-21 | Rohde & Schwarz GmbH & Co. KG | Predistortion system and method |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH07261800A (ja) | 1994-03-17 | 1995-10-13 | Nippon Telegr & Teleph Corp <Ntt> | 変換符号化方法、復号化方法 |
JPH11177434A (ja) * | 1997-10-07 | 1999-07-02 | Yamaha Corp | 音声符号化復号方式 |
JP2001298367A (ja) * | 2000-04-12 | 2001-10-26 | Nippon Telegr & Teleph Corp <Ntt> | オーディオ信号符号化方法、オーディオ信号復号化方法およびオーディオ信号符号化/復号化装置と前記方法を実施するプログラムを記録した記録媒体 |
Family Cites Families (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH07118658B2 (ja) * | 1986-03-26 | 1995-12-18 | 日本電信電話株式会社 | 信号符号化方法 |
US5684920A (en) * | 1994-03-17 | 1997-11-04 | Nippon Telegraph And Telephone | Acoustic signal transform coding method and decoding method having a high efficiency envelope flattening method therein |
JP3087814B2 (ja) * | 1994-03-17 | 2000-09-11 | 日本電信電話株式会社 | 音響信号変換符号化装置および復号化装置 |
KR100208710B1 (ko) * | 1995-12-27 | 1999-07-15 | 윤종용 | 비디오 신호처리시스템에서 비표준의 동기신호를 처리하기 위한 장치 및 그 방법 |
US6904404B1 (en) * | 1996-07-01 | 2005-06-07 | Matsushita Electric Industrial Co., Ltd. | Multistage inverse quantization having the plurality of frequency bands |
JP3800704B2 (ja) * | 1997-02-13 | 2006-07-26 | ソニー株式会社 | 映像信号処理装置及び方法 |
JP3616307B2 (ja) * | 2000-05-22 | 2005-02-02 | 日本電信電話株式会社 | 音声・楽音信号符号化方法及びこの方法を実行するプログラムを記録した記録媒体 |
CN1288626C (zh) * | 2001-10-25 | 2006-12-06 | 皇家飞利浦电子股份有限公司 | 利用减少的带宽在传输通道上传输宽带音频信号的方法 |
JP4168976B2 (ja) * | 2004-05-28 | 2008-10-22 | ソニー株式会社 | オーディオ信号符号化装置及び方法 |
-
2009
- 2009-01-23 US US12/812,549 patent/US8724734B2/en active Active
- 2009-01-23 EP EP20090703930 patent/EP2234273B8/en active Active
- 2009-01-23 ES ES09703930T patent/ES2409843T3/es active Active
- 2009-01-23 CN CN2009801023883A patent/CN101911501B/zh active Active
- 2009-01-23 WO PCT/JP2009/051123 patent/WO2009093714A1/ja active Application Filing
- 2009-01-23 JP JP2009550585A patent/JP5097217B2/ja active Active
- 2009-01-23 CA CA 2711539 patent/CA2711539C/en active Active
-
2012
- 2012-08-16 JP JP2012180362A patent/JP5285798B2/ja active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH07261800A (ja) | 1994-03-17 | 1995-10-13 | Nippon Telegr & Teleph Corp <Ntt> | 変換符号化方法、復号化方法 |
JPH11177434A (ja) * | 1997-10-07 | 1999-07-02 | Yamaha Corp | 音声符号化復号方式 |
JP2001298367A (ja) * | 2000-04-12 | 2001-10-26 | Nippon Telegr & Teleph Corp <Ntt> | オーディオ信号符号化方法、オーディオ信号復号化方法およびオーディオ信号符号化/復号化装置と前記方法を実施するプログラムを記録した記録媒体 |
Non-Patent Citations (4)
Title |
---|
"Proceedings of the 5th International Conference on Signal Processing, 2000 (WCCC-ICSP 2000)", vol. 2, August 2000, article JIANG WEI ET AL.: "An Efficient Coding Scheme Based on Sub-Frame Vector Quantization", pages: 1073 - 1076, XP008137442 * |
"Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing, 2007 (ICASSP 2007)", vol. 4, April 2007, article A. MOUCHTARIS ET AL. ET AL.: "Conditional Vector Quantization for Voice Conversion", pages: IV-505 - IV-508, XP031463897 * |
See also references of EP2234273A4 |
TOSHIO MIKI ET AL.: "Pitch Synchronous Innovation CELP (PSI-CELP)", IEICE TRANSACTIONS, vol. J77-A, no. 3, March 1994 (1994-03-01), pages 314 - 324 |
Cited By (23)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102812642A (zh) * | 2010-03-09 | 2012-12-05 | 日本电信电话株式会社 | 编码方法、解码方法、装置、程序及记录介质 |
EP2573941A1 (en) * | 2010-07-05 | 2013-03-27 | Nippon Telegraph and Telephone Corporation | Encoding method, decoding method, device, program, and recording medium |
WO2012005212A1 (ja) * | 2010-07-05 | 2012-01-12 | 日本電信電話株式会社 | 符号化方法、復号方法、符号化装置、復号装置、プログラム、及び記録媒体 |
JP5325340B2 (ja) * | 2010-07-05 | 2013-10-23 | 日本電信電話株式会社 | 符号化方法、復号方法、符号化装置、復号装置、プログラム、及び記録媒体 |
WO2012005211A1 (ja) * | 2010-07-05 | 2012-01-12 | 日本電信電話株式会社 | 符号化方法、復号方法、符号化装置、復号装置、プログラム、及び記録媒体 |
CN102959623A (zh) * | 2010-07-05 | 2013-03-06 | 日本电信电话株式会社 | 编码方法、解码方法、编码装置、解码装置、程序及记录介质 |
CN102959871A (zh) * | 2010-07-05 | 2013-03-06 | 日本电信电话株式会社 | 编码方法、解码方法、编码装置、解码装置、程序及记录介质 |
CN102959873A (zh) * | 2010-07-05 | 2013-03-06 | 日本电信电话株式会社 | 编码方法、解码方法、装置、程序及记录介质 |
CN102959872A (zh) * | 2010-07-05 | 2013-03-06 | 日本电信电话株式会社 | 编码方法、解码方法、装置、程序及记录介质 |
JP5331248B2 (ja) * | 2010-07-05 | 2013-10-30 | 日本電信電話株式会社 | 符号化方法、復号方法、装置、プログラム及び記録媒体 |
WO2012005209A1 (ja) * | 2010-07-05 | 2012-01-12 | 日本電信電話株式会社 | 符号化方法、復号方法、装置、プログラム及び記録媒体 |
EP2573941A4 (en) * | 2010-07-05 | 2013-06-26 | Nippon Telegraph & Telephone | ENCODING METHOD, DECODING METHOD, DEVICE, PROGRAM, AND RECORDING MEDIUM |
WO2012005210A1 (ja) * | 2010-07-05 | 2012-01-12 | 日本電信電話株式会社 | 符号化方法、復号方法、装置、プログラムおよび記録媒体 |
EP2573942A1 (en) * | 2010-07-05 | 2013-03-27 | Nippon Telegraph And Telephone Corporation | Encoding method, decoding method, device, program, and recording medium |
JP5331249B2 (ja) * | 2010-07-05 | 2013-10-30 | 日本電信電話株式会社 | 符号化方法、復号方法、装置、プログラムおよび記録媒体 |
JP5337305B2 (ja) * | 2010-07-05 | 2013-11-06 | 日本電信電話株式会社 | 符号化方法、復号方法、符号化装置、復号装置、プログラム、及び記録媒体 |
US8711012B2 (en) | 2010-07-05 | 2014-04-29 | Nippon Telegraph And Telephone Corporation | Encoding method, decoding method, encoding device, decoding device, program, and recording medium |
EP2573942A4 (en) * | 2010-07-05 | 2014-08-13 | Nippon Telegraph & Telephone | CODING METHOD, DECODING METHOD AND DEVICE THEREFOR, PROGRAM THEREFOR AND RECORDING MEDIUM |
CN102959623B (zh) * | 2010-07-05 | 2014-08-13 | 日本电信电话株式会社 | 编码方法、解码方法、编码装置、解码装置、程序及记录介质 |
US9319645B2 (en) | 2010-07-05 | 2016-04-19 | Nippon Telegraph And Telephone Corporation | Encoding method, decoding method, encoding device, decoding device, and recording medium for a plurality of samples |
CN102959871B (zh) * | 2010-07-05 | 2016-09-21 | 日本电信电话株式会社 | 编码方法、解码方法、编码装置、解码装置、程序及记录介质 |
US10665247B2 (en) | 2012-07-12 | 2020-05-26 | Nokia Technologies Oy | Vector quantization |
JP2017138605A (ja) * | 2017-02-16 | 2017-08-10 | ノキア テクノロジーズ オーユー | ベクトル量子化 |
Also Published As
Publication number | Publication date |
---|---|
EP2234273B1 (en) | 2013-03-20 |
EP2234273A4 (en) | 2011-12-28 |
ES2409843T3 (es) | 2013-06-28 |
JP5097217B2 (ja) | 2012-12-12 |
EP2234273A1 (en) | 2010-09-29 |
EP2234273B8 (en) | 2013-08-07 |
US8724734B2 (en) | 2014-05-13 |
CA2711539A1 (en) | 2009-07-30 |
US20110044405A1 (en) | 2011-02-24 |
CN101911501B (zh) | 2013-07-10 |
ES2409843T8 (es) | 2013-11-11 |
CN101911501A (zh) | 2010-12-08 |
CA2711539C (en) | 2013-10-08 |
JPWO2009093714A1 (ja) | 2011-05-26 |
JP5285798B2 (ja) | 2013-09-11 |
JP2013008040A (ja) | 2013-01-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP5285798B2 (ja) | 符号化方法、復号化方法、それらの装置、及びそれらのプログラムと記録媒体 | |
JP5603484B2 (ja) | 符号化方法、復号方法、符号化装置、復号装置、プログラム、記録媒体 | |
US10083703B2 (en) | Frequency domain pitch period based encoding and decoding in accordance with magnitude and amplitude criteria | |
RU2643646C2 (ru) | Кодер для кодирования аудиосигнала, система передачи аудио и способ определения значений коррекции | |
US7599833B2 (en) | Apparatus and method for coding residual signals of audio signals into a frequency domain and apparatus and method for decoding the same | |
EP2650878B1 (en) | Encoding method, encoder, periodic feature amount determination method, periodic feature amount determination apparatus, program and recording medium | |
US7310598B1 (en) | Energy based split vector quantizer employing signal representation in multiple transform domains | |
CN103918028B (zh) | 基于自回归系数的有效表示的音频编码/解码 | |
US7027980B2 (en) | Method for modeling speech harmonic magnitudes | |
EP2571170B1 (en) | Encoding method, decoding method, encoding device, decoding device, program, and recording medium | |
RU2673691C1 (ru) | Устройство преобразования коэффициентов линейного предсказания и способ преобразования коэффициентов линейного предсказания | |
JP5336943B2 (ja) | 符号化方法、復号方法、符号化器、復号器、プログラム | |
US11621010B2 (en) | Coding apparatus, coding method, program, and recording medium | |
JP7326285B2 (ja) | 音声音響統合復号および符号化のqmfに基づく高調波トランスポーザーの改良のための方法、機器、およびシステム | |
JP6780108B2 (ja) | 符号化装置、復号装置、平滑化装置、逆平滑化装置、それらの方法、およびプログラム | |
Ramabadran et al. | An iterative interpolative transform method for modeling harmonic magnitudes | |
WO2012053149A1 (ja) | 音声分析装置、量子化装置、逆量子化装置、及びこれらの方法 | |
Teague et al. | Enhanced spectral modeling for MBE speech coders | |
Mikhael et al. | A new linear predictor employing vector quantization in nonorthogonal domains for high quality speech coding |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 200980102388.3 Country of ref document: CN |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 09703930 Country of ref document: EP Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2009550585 Country of ref document: JP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2711539 Country of ref document: CA |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2009703930 Country of ref document: EP |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
WWE | Wipo information: entry into national phase |
Ref document number: 12812549 Country of ref document: US |