JPWO2010073977A1 - Encoding method, decoding method, apparatus thereof, program, and recording medium - Google Patents

Abstract

By performing a linear predictive analysis on the input time series signal, at least a first-order PARCOR coefficient and a second-order PARCOR coefficient are calculated, respectively, and a value determined according to the calculated first-order PARCOR coefficient and a second-order PARCOR coefficient. A parameter determined according to a relational expression established between values determined according to the PARCOR coefficient is calculated, and a code corresponding to information including the parameter and the primary PARCOR coefficient or the secondary PARCOR coefficient is generated. . At the time of decoding, the PARCOR coefficient is decoded using the correlation between the primary PARCOR coefficient and the secondary PARCOR coefficient.

Description

  The present invention relates to a technique for encoding a time-series signal by linear prediction analysis, and more particularly to a method for encoding a PARCOR coefficient obtained by linear prediction analysis, a decoding method, a device thereof, a program, and a recording medium.

  When transmitting time-series signals such as audio signals and video information through a communication channel or recording them on an information recording medium, the method of transmitting and recording after converting the time-series signals into compressed codes is the transmission efficiency and recording efficiency. This is effective. In addition, with the spread of broadband in recent years and the increase in storage capacity, lossless compression coding methods that require complete reproduction of the original signal are more important than lossy compression coding methods that prioritize high compression rates. (See Non-Patent Document 1, for example). Under such circumstances, technology for lossless compression coding of acoustic signals using elemental technology such as linear prediction analysis has been approved as the MPEG (Moving Picture Expert Group) international standard “MPEG-4 ALS” (for example, Non-patent document 2).

  FIG. 1 is a block diagram for explaining a functional configuration of a coding apparatus 1010 of a conventional lossless compression coding method. FIG. 2 is a block diagram for explaining a functional configuration of a decoding apparatus 1020 that decodes a code generated by the encoding apparatus 1010 of FIG. FIG. 3 is a diagram for explaining how the first-order PARCOR coefficient and the second-order PARCOR are encoded in the encoding apparatus 1010 of FIG. First, a conventional lossless compression coding method will be described with reference to these drawings.

[Encoding method]
A sampled and quantized PCM (pulse code modulation) time series signal x (n) (n is an index indicating discrete time) is input to the frame buffer 1011 of the encoding apparatus 1010. The frame buffer 1011 buffers time series signals x (n) (n = 1,..., N) (N is a positive integer) for a predetermined time interval (hereinafter referred to as “frame”). The encoding device 1010 encodes the time series signal x (n) (n = 1,..., N) for each frame.

  First, a time-series signal x (n) (n = 1,..., N) for one frame is sent to the linear prediction analysis unit 1012. The linear prediction analysis unit 1012 performs linear to M-th order by linear prediction analysis. PARCOR coefficients k (m) up to (m = 1, 2,..., M) are calculated. M is a positive integer indicating the predicted order. The m-th order PARCOR coefficient means a PARCOR coefficient of a linear prediction model of the prediction order m. In the linear prediction analysis, a time series signal x (n) at a certain time point n and M time points n-1, n-2,..., NM time series signals x ( n-1), x (n-2), ..., x (nM) are weighted by coefficients α (m) (m = 1, ..., M) (referred to as "linear prediction coefficients") It is assumed that a linear linear combination is established between the object and the prediction residual e (n). A linear prediction model based on this assumption is as follows. In linear prediction analysis, the following linear prediction coefficient α (m) (m = 1,2, ..., M) is applied to the input time series signal x (n) (n = 1, ..., N). ) Or a PARCOR coefficient k (m) (m = 1, 2,..., M) or the like that can be converted thereto. Β · γ represents a product β × γ of β and γ.

e (n) = x (n) + α (1) ・ x (n-1) + α (2) ・ x (n-2) + ... + α (M) ・ x (nM)
Further, a time series signal y (n) at a certain time point n is converted into M time points n-1, n-2,..., NM time series signals x (n-1), A linear FIR (Finite Impulse Response) filter of the following equation that is estimated using x (n−2),..., x (nM) is called a “linear prediction filter”.

y (n) =-{α (1) ・ x (n-1) + α (2) ・ x (n-2) + ... + α (M) ・ x (nM)}
The calculated PARCOR coefficients k (m) (m = 1, 2,..., M) from the first order to the Mth order are sent to the nonlinear quantization unit 1013 and quantized to obtain the first order to the Mth order. Quantized PARCOR coefficients i (m) (m = 1, 2,..., M) are generated. The “quantized PARCOR coefficient” may be the quantized value of the PARCOR coefficient itself or an index attached to the quantized value of the PARCOR coefficient. The quantized PARCOR coefficients i (m) (m = 1, 2,..., M) from the 1st order to the Mth order are sent to the coefficient coding unit 1014, where they are entropy coded to obtain the coefficient code C _k. Generated. This encoding is performed independently for each PARCOR coefficient k (m) (m = 1, 2,..., M) from the first order to the Mth order. For example, as shown in FIG. 3, the first-order quantized PARCOR coefficient i (1) and the second-order quantized PARCOR coefficient i (2) are entropy-coded independently of each other. Further, the quantized PARCOR coefficients i (m) (m = 1, 2,..., M) from the first order to the Mth order are also sent to the linear prediction coefficient conversion unit 1015. The linear prediction coefficient conversion unit 1015 calculates linear prediction coefficients α (m) (m = 1, 2,..., M) of the prediction order M linear prediction filter using these. The linear prediction unit 1016 includes a time series signal x (n) (n = 1,..., N) for one frame and each linear prediction coefficient α (m) (m = 1, 2,..., M). Are used to generate linear prediction values y (n) (n = 1,..., N) by linear prediction. The subtraction unit 1017 calculates a prediction residual (also referred to as “prediction error”) e (n) obtained by subtracting the linear prediction value y (n) from the time-series signal x (n) (prediction filter processing). Calculated prediction residuals e (n) is sent to residual coding unit 1018, where entropy-coded by residual code C _e is generated. A coefficient code C _k generated by the coefficient coding section 1014, residual code C _e generated by the residual encoding unit 1018 is sent to the synthesis unit 1019. Coefficient code C _k and the residual code C _e is combined by the combining unit 1019, the code C _g is generated.

[Decryption method]
Code C _g input to the decoding device 1020 is separated into a coefficient code C _k and the residual code C _e by the demultiplexer 1021. The coefficient code C _k and the residual code C _e, is respectively decoded by the coefficient decoding unit 1022 and the residual decoder 1023, a quantization from the primary to M order PARCOR coefficients i (m) (m = 1 , ... , M) and prediction residuals e (n) (n = 1,..., N) are generated. The quantized PARCOR coefficients i (m) (m = 1,..., M) from the first order to the Mth order are sent to the linear prediction coefficient conversion unit 1024. The linear prediction coefficient conversion unit 1024 uses these to calculate each linear prediction coefficient α (m) (m = 1,..., M) of the linear prediction filter of the prediction order M. The linear prediction unit 1025 uses each calculated linear prediction coefficient α (m) (m = 1,..., M) and the time-series signal x (n) output from the addition unit 1026 in the past to perform linear operation. A linear prediction value y (n) is generated by the prediction. The adder 1026 adds the linear prediction value y (n) and the prediction residual e (n) to generate a time series signal x (n) (inverse prediction filter processing).

MatHans, “Lossless Compression of Digital Audio”, IEEE SIGNAL PROCESSING MAGAZINE, July 2001, pp.21-32. ISO / IEC 14496-3 AMENDMENT 2: Audio Lossless Cording (ALS), new audio profiles and BSAC extensions.

  An object of this invention is to provide the technique which improves the encoding compression rate of the PARCOR coefficient obtained by the linear prediction analysis of the time series signal.

  In the present invention, a code corresponding to the PARCOR coefficient is generated using the correlation between the primary PARCOR coefficient and the secondary PARCOR coefficient. At the time of decoding, the PARCOR coefficient is decoded using the correlation between the primary PARCOR coefficient and the secondary PARCOR coefficient.

  In the first aspect of the present invention, at the time of encoding, at least a first-order PARCOR coefficient and a second-order PARCOR coefficient are calculated by performing linear prediction analysis on an input time-series signal, and the calculated first-order PARCOR coefficient is calculated. The parameter determined according to the relational expression established between the value determined according to the PARCOR coefficient and the value determined according to the second order PARCOR coefficient is calculated, and the parameter and the first order PARCOR coefficient or the second order PARCOR coefficient are calculated. The code | cord | chord corresponding to the information containing any one is produced | generated. At the time of decoding, a code corresponding to information including the parameter and either the primary PARCOR coefficient or the secondary PARCOR coefficient is decoded, and at least the parameter and the primary PARCOR are decoded. A decoded value corresponding to the coefficient or a decoded value corresponding to the second-order PARCOR coefficient is generated, and a restoration value of the second-order PARCOR coefficient is calculated using the parameter and the decoded value corresponding to the first-order PARCOR coefficient. Or using the parameter and the decoded value corresponding to the second order PARCOR coefficient, the restoration value of the first order PARCOR coefficient is calculated.

  Further, in the second aspect of the present invention, at the time of encoding, at least a first-order PARCOR coefficient and a second-order PARCOR coefficient are respectively calculated by performing linear prediction analysis on the input time-series signal. When the absolute value of the PARCOR coefficient is equal to or greater than a predetermined threshold, the first variable length encoding method is selected as an encoding method for generating a code corresponding to the secondary PARCOR coefficient, and the primary PARCOR When the absolute value of the coefficient is less than the threshold, a second variable length encoding method different from the first variable length encoding method is selected as an encoding method for generating a code corresponding to the secondary PARCOR coefficient, Using the selected encoding method, a second-order quantized PARCOR coefficient obtained by quantizing the second-order PARCOR coefficient is encoded, and a code corresponding to the second-order PARCOR is generated.Then, at the time of decoding, the absolute value of the decoded value of the code corresponding to the primary PARCOR coefficient is compared with a predetermined threshold value, and the secondary value is determined by the decoding method corresponding to the predetermined first variable length encoding method. Whether the code corresponding to the PARCOR coefficient is decoded, or the code corresponding to the secondary PARCOR coefficient is decoded by a decoding method corresponding to a predetermined second variable length encoding method different from the first variable length encoding method Determine.

  Accordingly, the total code amount of codes corresponding to the primary PARCOR coefficient and the secondary PARCOR coefficient is reduced as compared with the case where the primary PARCOR coefficient and the secondary PARCOR coefficient are encoded independently of each other. Can do.

  As described above, in the present invention, since the PARCOR coefficient is encoded using the correlation between the first-order PARCOR coefficient and the second-order PARCOR coefficient, the PARCOR obtained by the linear prediction analysis of the time-series signal is performed. The coding compression rate of the coefficient can be improved.

The block diagram for demonstrating the function structure of the encoding apparatus of the conventional lossless compression encoding system. The block diagram for demonstrating the function structure of the decoding apparatus which decodes the code | symbol produced | generated with the encoding apparatus of FIG. The figure for demonstrating a mode that a primary PARCOR coefficient and a secondary PARCOR are encoded in the encoding apparatus of FIG. The graph which plotted the relationship between the 1st-order PARCOR coefficient K (1) and the 2nd-order PARCOR coefficient K (2) obtained by carrying out the linear prediction analysis of the acoustic signal. FIG. 5A shows the ratio k (2) / k (1) between the first-order PARCOR coefficient k (1) and the second-order PARCOR coefficient k (2) obtained by linear prediction analysis of the acoustic signal. It is a graph which illustrates frequency. FIG. 5B is a graph for explaining that the appearance probability distribution of the secondary PARCOR coefficient k (2) changes according to the magnitude of the primary PARCOR coefficient k (1). The graph which illustrated the frequency of the secondary PARCOR coefficient k (2), and the frequency of -b which satisfy | fills relational expression k (2) = a * k (1) + b (a = -0.75). The block diagram for demonstrating the function structure of the encoding apparatus of 1st Embodiment. FIG. 8A is a block diagram for explaining the details of the nonlinear quantization unit and the parameter calculation unit shown in FIG. 7, and FIG. 8B is the details of the coefficient encoding unit shown in FIG. It is a block diagram for demonstrating. The block diagram for demonstrating the function structure of the decoding apparatus of 1st Embodiment. FIG. 10A is a block diagram for explaining details of the coefficient decoding unit shown in FIG. 9, and FIG. 10B is a diagram for explaining details of the PARCOR coefficient calculation unit shown in FIG. It is a block diagram. The flowchart for demonstrating the encoding method of 1st Embodiment. The flowchart for demonstrating the decoding method of 1st Embodiment. The graph for demonstrating and explaining the nonlinear quantization method of a primary PARCOR coefficient. The graph for demonstrating and explaining the nonlinear quantization method of a secondary PARCOR coefficient. FIG. 15A is a block diagram for explaining details of the nonlinear quantization unit and the parameter calculation unit of the second embodiment, and FIG. 15B shows details of the PARCOR coefficient calculation unit of the second embodiment. It is a block diagram for demonstrating. The flowchart for demonstrating the encoding method of 2nd Embodiment. The flowchart for demonstrating the decoding method of 2nd Embodiment. The block diagram for demonstrating the function structure of the encoding apparatus of 3rd Embodiment. The block diagram for demonstrating the function structure of the decoding apparatus of 3rd Embodiment. 20A is a block diagram for explaining details of the nonlinear quantization unit and the parameter calculation unit shown in FIG. 18, and FIG. 20B shows details of the PARCOR coefficient calculation unit shown in FIG. It is a block diagram for demonstrating. The flowchart for demonstrating the encoding method of 3rd Embodiment. The flowchart for demonstrating the decoding method of 3rd Embodiment. The block diagram for demonstrating the function structure of the encoding apparatus in the modification 2 of 3rd Embodiment. The block diagram for demonstrating the function structure of the encoding apparatus of 4th Embodiment. The block diagram for demonstrating the detail of the nonlinear quantization part, parameter calculation part, and selection part which were shown in FIG. The block diagram for demonstrating the function structure of the decoding apparatus of 4th Embodiment. The block diagram for demonstrating the detail of the PARCOR coefficient calculation part shown in FIG. The flowchart for demonstrating the encoding method of 4th Embodiment. The flowchart for demonstrating the decoding method of 4th Embodiment. The block diagram for demonstrating the detail of the nonlinear quantization part of the encoding apparatus in the modification 1 of 4th Embodiment, a selection part, and a parameter calculation part. The block diagram for demonstrating the detail of the PARCOR coefficient calculation part of the decoding apparatus in the modification 1 of 4th Embodiment. The flowchart for demonstrating the encoding method of the modification 1 of 4th Embodiment. The flowchart for demonstrating the decoding method of the modification 1 of 4th Embodiment. The block diagram for demonstrating the function structure of the encoding apparatus of 5th Embodiment. The block diagram for demonstrating the detail of the quantization method selection part shown in FIG. 34, a quantization part, and an encoding method selection part. The block diagram for demonstrating the detail of the coefficient encoding part shown in FIG. The block diagram for demonstrating the function structure of the decoding apparatus of 5th Embodiment. The block diagram for demonstrating the detail of the coefficient decoding part shown in FIG. The flowchart for demonstrating the encoding method of 5th Embodiment. The flowchart for demonstrating the encoding method of 5th Embodiment.

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

  In the following, after describing the principle of the present invention, each embodiment will be described.

〔principle〕
<Correlation of PARCOR coefficients>
When the PARCOR coefficient is viewed as a parameter for specifying the linear prediction model, the primary PARCOR coefficient and the secondary PARCOR coefficient have no correlation. However, when performing linear predictive analysis of time series signals such as acoustic signals, video signals, biological signals, seismic signals, sensor array signals, etc., from the characteristics of these signals, the first and second order PARCOR coefficients and There is often a correlation between the two (unpublished).

  For example, when an acoustic signal is subjected to linear prediction analysis, the first-order PARCOR coefficient is often in the vicinity of 1, and in this case, the second-order PARCOR coefficient is often in the vicinity of -1. Further, when the acoustic signal is close to white noise and the first-order value (= first-order PARCOR coefficient) of the autocorrelation function is close to 0, the second-order PARCOR coefficient is often close to 0.

  FIG. 4 is a graph plotting the relationship between the first-order PARCOR coefficient k (1) and the second-order PARCOR coefficient k (2) obtained by linear prediction analysis of the acoustic signal. Here, the horizontal axis represents the primary PARCOR coefficient k (1), and the vertical axis represents the secondary PARCOR coefficient k (2). As shown in this graph, the first-order PARCOR coefficient k (1) is about 1 and the second-order PARCOR coefficient k (2) is often about -1. In particular, in a region where the absolute values of the primary PARCOR coefficient k (1) and the secondary PARCOR coefficient k (2) are close to 1, the primary PARCOR coefficient k (1) and the secondary PARCOR coefficient k (2) There is a strong correlation between

  FIG. 5A shows the ratio k (2) / k (1) between the first-order PARCOR coefficient k (1) and the second-order PARCOR coefficient k (2) obtained by linear prediction analysis of the acoustic signal. ). Here, the horizontal axis indicates the ratio k (2) / k (1), and the vertical axis indicates the frequency. Also from this graph, the sign of the first-order PARCOR coefficient k (1) and the second-order PARCOR coefficient k (2) is likely to be reversed, and the first-order PARCOR coefficient k (1) and the second-order PARCOR coefficient k (1) It can be seen that there is a correlation with the PARCOR coefficient k (2). This graph also shows that the absolute value of the second-order PARCOR coefficient k (2) tends to be smaller than the absolute value of the first-order PARCOR coefficient k (1).

  FIG. 5B is a graph for explaining that the appearance probability distribution of the secondary PARCOR coefficient k (2) changes according to the magnitude of the primary PARCOR coefficient k (1).

  Here, the horizontal axis indicates a secondary index, and the vertical axis indicates the appearance probability. The secondary index means an index attached to the quantized value of the secondary PARCOR coefficient k (2). In FIG. 5B, the graph shown in black shows the probability distribution of the secondary index when the primary index is 0, 1, 2, and the graph shown in white shows the primary index. The appearance probability distribution of the secondary index in the case of 10, 11, 12 is shown. The primary index means an index attached to the quantized value of the primary PARCOR coefficient k (1). The index in this example is a value that decreases monotonously in a broad sense with respect to an increase in the absolute value of the PARCOR coefficient. In addition, the primary index in FIG. 5B corresponds to a primary PARCOR coefficient k (1) that is positive, and the secondary index corresponds to a secondary PARCOR coefficient k (2) that is negative. . For example, the primary indexes 0, 1, 2 are indexes assigned to the quantized values of the primary PARCOR coefficient k (1) near 0.99. The primary indexes 10, 11, and 12 are indexes assigned to the quantized values of the primary PARCOR coefficient k (1) near 0.9. The secondary indexes 10, 11, 12, and 13 are indexes attached to the quantized values of the secondary PARCOR coefficient k (2) in the vicinity of -0.5 to -0.4. The secondary indexes 1, 2, 3, and 4 are indexes attached to the quantized values of the secondary PARCOR coefficient k (2) in the vicinity of −0.9 to −0.8.

  As illustrated in FIG. 5B, when the primary PARCOR coefficient k (1) is large (for example, around 0.99), the secondary PARCOR coefficient k (2) (for example, having a large absolute value) The probability of obtaining −0.9 to −0.8) is increased. On the other hand, when the primary PARCOR coefficient k (1) is small (for example, around 0.9), the secondary PARCOR coefficient k (2) having a small absolute value (for example, −0.5 to −0.4). ) Is likely to be obtained. That is, the frequency distribution of the secondary PARCOR coefficient k (2) varies depending on the magnitude of the primary PARCOR coefficient k (1). As the first order PARCOR coefficient k (1) is larger and closer to 1, the second order PARCOR coefficient k (2) is smaller and close to −1 (the absolute value is large and close to 1).

<Encoding of PARCOR coefficient>
As described above, there is a correlation between the first-order PARCOR coefficient and the second-order PARCOR coefficient (unpublished). In the present invention, this correlation is used to generate a code corresponding to the PARCOR coefficient. At the time of decoding, the PARCOR coefficient is decoded using the correlation between the primary PARCOR coefficient and the secondary PARCOR coefficient.

[First aspect]
The encoding device according to the first aspect calculates (II) at least a first-order PARCOR coefficient and a second-order PARCOR coefficient by performing linear prediction analysis on the input time-series signal. A parameter determined according to a relational expression established between a value determined according to the first-order PARCOR coefficient and a value determined according to the second-order PARCOR coefficient, and (III) the parameter and the first-order PARCOR coefficient Alternatively, a code corresponding to information including any one of the secondary PARCOR coefficients is generated. The decoding apparatus in this case includes (IV) a parameter determined according to a relational expression established between a value determined according to the first-order PARCOR coefficient and a value determined according to the second-order PARCOR coefficient, and the first-order PARCOR A code corresponding to information including either a coefficient or the second-order PARCOR coefficient, and at least the decoded value corresponding to the first-order PARCOR coefficient or the second-order PARCOR coefficient A corresponding decoded value is generated, and a restored value of the second-order PARCOR coefficient is calculated using the parameter and the decoded value corresponding to the first-order PARCOR coefficient, or the parameter and the second-order PARCOR coefficient are calculated. The restoration value of the first-order PARCOR coefficient is calculated using the decoded value corresponding to.

  Thereby, the total code amount of codes corresponding to the primary PARCOR coefficient and the secondary PARCOR coefficient can be reduced as compared with the case where the primary PARCOR coefficient and the secondary PARCOR coefficient are encoded independently of each other.

  Further, in the method of this aspect, as the correlation between the first-order PARCOR coefficient and the second-order PARCOR coefficient increases, the absolute value of the parameter approaches a specific value, and the variance of the parameter is the variance of the first-order PARCOR coefficient or 2 In the case where it becomes smaller than the variance of the next PARCOR coefficient, a particularly remarkable effect is obtained. In this case, it is desirable that a code corresponding to the parameter is generated using a variable length coding method. In addition, this variable length encoding method preferably satisfies the following conditions.

  << Condition >> When the absolute value of the first encoding target is closer to a specific value than the absolute value of the second encoding target, the first frequency is higher than the second frequency.

  First frequency: The frequency at which a code having a shorter code length than the second encoding target code is assigned to the first encoding target. That is, the frequency at which the code length of the code assigned to the first encoding target is shorter than the code length of the code assigned to the second encoding target.

  Second frequency: The frequency with which a code having a longer code length than the second encoding target code is assigned to the first encoding target. That is, the frequency with which the code length of the code assigned to the first encoding target is longer than the code length of the code assigned to the second encoding target.

  Further, a relational expression established between a value determined according to the first-order PARCOR coefficient and a value determined according to the second-order PARCOR coefficient is, for example, the following relational expression (A) or (B).

<< Relational Expression (A) >> Depending on the secondary PARCOR coefficient by the sum of a first multiplication value obtained by multiplying a predetermined weighting coefficient by a value determined according to the primary PARCOR coefficient and the first variable value An equation that can represent a fixed value. For example,
First multiplication value + first variable value = a value determined according to the second-order PARCOR coefficient,
An equation that becomes However, the first multiplication value = predetermined weighting factor × value determined according to the first-order PARCOR coefficient.

<< Relational Expression (B) >> Depending on the primary PARCOR coefficient by the sum of a second multiplication value obtained by multiplying a predetermined weighting coefficient by a value determined according to the secondary PARCOR coefficient and the second variable value An equation that can represent a fixed value. For example,
Second multiplied value + second variable value = value determined according to the second-order PARCOR coefficient,
An equation that becomes However, the second multiplication value = a predetermined weighting factor × a value determined according to the second-order PARCOR coefficient.

  Moreover, the above-mentioned parameters are, for example, the following parameters (A) or (B).

  << Parameter (A) >> A value determined according to the first variable value satisfying the relational expression (A).

  << Parameter (B) >> A value determined according to the second variable value satisfying the relational expression (B).

  Furthermore, the above relational expressions (A) and (B) and the parameters (A) and (B) satisfy the following relation (A) or (B), for example.

  << Relationship (A) >> A value determined according to the first-order PARCOR coefficient is a value that monotonously increases (monotonically non-decrease) at least with respect to an increase in the positive first-order PARCOR coefficient. In addition, the value determined according to the second-order PARCOR coefficient is a value that monotonically increases in a broad sense with respect to an increase in at least the negative second-order PARCOR coefficient. In addition, the absolute value of the parameter increases monotonously in a broad sense with respect to the first variable value that satisfies the above-described relational expression (A).

  << Relationship (B) >> A value determined according to the first-order PARCOR coefficient is a value that monotonously decreases (monotonically non-increases) at least with respect to an increase in the positive first-order PARCOR coefficient. The value determined according to the second-order PARCOR coefficient is a value that monotonously decreases in a broad sense with respect to an increase in at least a negative second-order PARCOR coefficient. In addition, the absolute value of the parameter increases monotonously in a broad sense with respect to an increase in the second variable value that satisfies the above-described relational expression (B).

When a value that increases monotonously in a broad sense with respect to an increase in value γ in a certain section is expressed as f (γ), f (γ ₁ ) ≦ f for any γ ₁ ≦ γ ₂ belonging to the section. The relationship (γ ₂ ) is established. In addition, when a value that decreases monotonously in a broad sense with respect to an increase in the value γ in a certain section is expressed as g (γ), g (γ ₁ ) ≧ g for any γ ₁ ≦ γ ₂ belonging to the section. The relationship (γ ₂ ) is established. As described above, the first-order PARCOR coefficient and the second-order PARCOR coefficient tend to be opposite to each other (see FIG. 5A). Therefore, when the above relation (A) or (B) is satisfied and the predetermined weighting coefficient in the above relational expression (A) or (B) is negative, the first-order PARCOR coefficient and the second-order PARCOR coefficient The larger the correlation with the PARCOR coefficient is, the closer the absolute value of the parameter is to 0, and the variance of the parameter is smaller than the variance of the primary PARCOR coefficient or the variance of the secondary PARCOR coefficient. In this case, for example, it is desirable to encode such parameters by a variable length encoding method that satisfies the following conditions.

  << Condition >> When the absolute value of the first encoding target is closer to 0 than the absolute value of the second encoding target, the first frequency is higher than the second frequency.

  First frequency: The frequency at which a code having a shorter code length than the second encoding target code is assigned to the first encoding target.

  Second frequency: The frequency with which a code having a longer code length than the second encoding target code is assigned to the first encoding target.

  Note that such variable length coding methods include Rice code (sometimes referred to as “Golomb-Rice code”), Golomb code, unary code (Unary). code) (sometimes referred to as “alpha code”), Huffman code, and the like.

  Further, as described above, the PARCOR coefficient takes a value between −1 and +1. Further, the correlation between the first-order PARCOR coefficient and the second-order PARCOR coefficient is larger as the absolute values of the first-order PARCOR coefficient and the second-order PARCOR coefficient are closer to 1, and smaller as they are closer to 0. Therefore, when the absolute value of the first-order PARCOR coefficient is equal to or greater than a predetermined threshold (threshold is a value between 0 and +1), the correlation between the first-order PARCOR coefficient and the second-order PARCOR coefficient is used. The first-order PARCOR coefficient and the second-order PARCOR coefficient may be independently encoded when the PARCOR coefficient is less than a predetermined threshold value. Further, it is determined whether or not the absolute value of the first-order PARCOR coefficient is equal to or larger than a predetermined threshold, and whether or not the absolute value of the second-order PARCOR coefficient is equal to or larger than a predetermined threshold. The method of conversion may be determined. That is, when the absolute value of the second-order PARCOR coefficient is equal to or greater than a predetermined threshold value (threshold value is 0 or more and +1 or less), the correlation between the first-order PARCOR coefficient and the second-order PARCOR coefficient is used. The first-order PARCOR coefficient and the second-order PARCOR coefficient may be independently encoded when the PARCOR coefficient is less than a predetermined threshold value.

  Note that the threshold determination may be performed in a region before quantization or may be performed in a region after quantization. Further, the region after quantization means a region of quantized values or an index region attached to the quantized values. In addition, when the magnitude relationship is reversed between the pre-quantization region and the post-quantization region, the magnitude relationship in threshold determination is reversed between the pre-quantization region and the post-quantization region. When the magnitude relation in the threshold determination is reversed before and after quantization, the process of performing “Process 1” when A ≧ T and performing “Process 2” when A <T is performed when A ′ ≦ T ′. “Process 1” is performed on the screen, and “Process 2” is performed when A ′> T ′. A and T are values of the region before quantization, and A 'and T' are values of the region after quantization with respect to it. The same applies to threshold determination at the time of decoding (the same applies hereinafter).

  For example, when the encoding apparatus performs the above steps (II) and (III) when the absolute value of the primary PARCOR coefficient is greater than or equal to a predetermined first threshold, the absolute value of the primary PARCOR coefficient May be less than the first threshold, a code corresponding to information including the primary PARCOR coefficient and the secondary PARCOR coefficient may be generated. In addition, when the absolute value of the second-order PARCOR coefficient is equal to or greater than a predetermined first threshold, the encoding apparatus executes the above steps (II) and (III), and the absolute value of the second-order PARCOR coefficient May be less than the first threshold, a code corresponding to information including the primary PARCOR coefficient and the secondary PARCOR coefficient may be generated.

  Further, the decoding apparatus may specify the decoding method using the magnitude of the absolute value of the primary or secondary PARCOR coefficient as an index. That is, the step (V) executed by the decoding device may include the following steps (V-a) or (V-b).

  << Step (V-a) >> The absolute value of the decoded value corresponding to the primary PARCOR coefficient is compared with a predetermined second threshold value. From the comparison result, it is determined whether or not the restoration value of the second-order PARCOR coefficient is calculated using the parameter and the decoded value corresponding to the first-order PARCOR coefficient.

  << Step (V-b) >> The absolute value of the decoded value corresponding to the second-order PARCOR coefficient is compared with a predetermined second threshold value or more. From the comparison result, it is determined whether or not the restoration value of the primary PARCOR coefficient is calculated using the parameter and the decoded value corresponding to the secondary PARCOR coefficient.

  In such a case, it is not necessary to include additional bits for specifying the decoding method in the code. Therefore, the code length of the code can be shortened.

[Second embodiment]
The encoding device according to the second aspect calculates (VI) at least a first-order PARCOR coefficient and a second-order PARCOR coefficient by performing linear prediction analysis on the input time-series signal, and (VII) a first-order PARCOR coefficient, respectively. When the absolute value of the PARCOR coefficient is equal to or greater than a predetermined threshold, the first variable length coding method is selected as the coding method for generating a code corresponding to the second order PARCOR coefficient, When the absolute value of the PARCOR coefficient is less than the threshold, a second variable length encoding method different from the first variable length encoding method is selected as an encoding method for generating a code corresponding to the second order PARCOR coefficient. (VIII) Using the encoding method selected in step (VII), the second-order quantized PARCOR coefficient obtained by quantizing the second-order PARCOR coefficient is encoded, and a code corresponding to the second-order PARCOR is obtained. It is formed. Note that the predetermined threshold value in step (VII) is a value between −1 and 1, but the threshold determination may be performed in the region before quantization or in the region after quantization. May be.

  As described with reference to FIG. 5B, the frequency distribution of the secondary PARCOR coefficient varies depending on the magnitude of the primary PARCOR coefficient. Therefore, by selecting a variable-length coding method for encoding the second-order quantized PARCOR coefficient according to the magnitude of the first-order PARCOR coefficient (step (VII)), it corresponds to the second-order PARCOR coefficient. The code amount of the code can be reduced.

  For example, in the example of FIG. 5B, when the primary index is 0, 1, 2, the secondary index is encoded using the following first variable length encoding method (A). To do. On the other hand, if the primary index is 10, 11, 12, the secondary index is encoded using the following second variable length encoding method (B).

  << First Variable Length Encoding Method (A) >> A variable length encoding method in which the third frequency is higher than the fourth frequency when the first encoding target is closer to 5 or 6 than the second encoding target. That is, when the distance between the first encoding target value and 5 or 6 is shorter than the distance between the second encoding target value and 5 or 6, the third frequency is a variable length higher than the fourth frequency. Encoding method.

  Third frequency: The frequency at which a code having a shorter code length than the second encoding target code is assigned to the first encoding target.

  Fourth frequency: The frequency at which a code having a longer code length than the second encoding target code is assigned to the first encoding target.

  << First Variable Length Encoding Method (B) >> A variable length encoding method in which the third frequency is higher than the fourth frequency when the first encoding target is closer to 7 or 8 than the second encoding target. That is, when the distance between the first encoding target value and 7 or 8 is shorter than the distance between the second encoding target value and 7 or 8, the third frequency is a variable length higher than the fourth frequency. Encoding method.

  An example of such a variable length encoding method is a Rice encoding method or a Huffman encoding method. Thereby, the amount of codes can be reduced as compared with the case where the secondary index is always encoded using the following variable length encoding method (C).

  << Variable-length encoding method (C) >> A variable-length encoding method in which the third frequency is higher than the fourth frequency when the first encoding target is closer to 6 or 7 than the second encoding target. That is, when the distance between the first encoding target value and 6 or 7 is shorter than the distance between the second encoding target value and 6 or 7, the third frequency is a variable length higher than the fourth frequency. Encoding method.

  When this is generalized, the first variable length encoding method selected in step (VII) is such that the absolute value of the first encoding target is set to a first value that is predetermined in advance than the absolute value of the second encoding target. When close, the third frequency is an encoding method higher than the fourth frequency. That is, the first variable length encoding method uses the third frequency when the distance between the first encoding target value and the first value is shorter than the distance between the second encoding target value and the first value. Is an encoding method higher than the fourth frequency.

  On the other hand, in the second variable length encoding method, the third frequency is greater than the fourth frequency when the absolute value of the first encoding target is closer to the predetermined second value than the absolute value of the second encoding target. Is a high encoding method. That is, the second variable length encoding method uses the third frequency when the distance between the first encoding target value and the second value is shorter than the distance between the second encoding target value and the second value. Is an encoding method higher than the fourth frequency.

  When the second-order PARCOR coefficient corresponding to the second-order PARCOR coefficient is larger as the absolute value of the second-order PARCOR coefficient is closer to 1, the first value is larger than the second value. When the second-order quantized PARCOR coefficient corresponding to the second-order PARCOR coefficient is smaller as the absolute value of the second-order PARCOR coefficient is closer to 1, the first value is smaller than the second value.

  Further, as described above, when the absolute value of the primary PARCOR coefficient is large and close to 1, the absolute value of the secondary PARCOR coefficient tends to be large. Furthermore, as the absolute value of the PARCOR coefficient is closer to 1, the influence of the quantization error on the linear prediction result is larger. Therefore, preferably, the quantization step size (sometimes referred to as “interval size”) when quantizing the second-order PARCOR coefficient is changed according to the absolute value of the first-order PARCOR coefficient. Let For example, a predetermined first quantization method is selected between the steps (VI) and (VII) when the absolute value of the first-order PARCOR coefficient is greater than or equal to a predetermined second threshold. When the absolute value of the next PARCOR coefficient is less than a predetermined second threshold, a predetermined second quantization method having a quantization step size larger than that of the first quantization method is selected, and the selected quantization is selected The method performs a step of quantizing the second-order PARCOR coefficient and generating a quantized value of the second-order PARCOR coefficient. Thereby, the energy of a prediction residual can be made small efficiently. That is, the amount of decrease in the energy of the prediction residual can be increased with respect to the amount of increase in the code amount of the code corresponding to the PARCOR coefficient. As a result, encoding efficiency can be improved.

[First Embodiment]
Next, a first embodiment of the present invention will be described.

  In the first embodiment, the following configuration will be described in the framework described in [First aspect] of [Principle] described above. However, this does not limit the present invention.

  Relational expression: A value determined according to the second-order PARCOR coefficient by the sum of the first multiplication value obtained by multiplying a predetermined weighting coefficient by a value determined according to the first-order PARCOR coefficient and the first variable value. An expressible equation.

  A value determined according to the first order PARCOR coefficient: the first order PARCOR coefficient. This is a “value that increases monotonously in a broad sense with respect to an increase in the first-order PARCOR coefficient”.

  A value determined according to the second order PARCOR coefficient: the second order PARCOR coefficient. This is “a value that increases monotonically in a broad sense with respect to an increase in the second-order PARCOR coefficient”.

  Parameter: a first quantized variable value obtained by quantizing a first variable value that satisfies the above relational expression. This is a “value determined according to the first variable value”, and the absolute value of the parameter increases monotonously in a broad sense with respect to the increase in the first variable value. The “first quantized variable value” may be the quantized value itself of the first variable value, or may be an index attached to the quantized value of the first variable value (the same applies hereinafter). ).

  -Code of step (III): Code corresponding to information including parameters and first-order PARCOR coefficients.

<Configuration>
FIG. 7 is a block diagram for explaining a functional configuration of the encoding device 10 according to the first embodiment. 8A is a block diagram for explaining details of the nonlinear quantization unit 11 and the parameter calculation unit 12 shown in FIG. 7, and FIG. 8B is a coefficient encoding unit shown in FIG. 13 is a block diagram for explaining the details of FIG. FIG. 9 is a block diagram for explaining a functional configuration of the decoding device 20 according to the first embodiment. FIG. 10A is a block diagram for explaining details of the coefficient decoding unit 21 shown in FIG. 9, and FIG. 10B shows details of the PARCOR coefficient calculating unit 22 shown in FIG. It is a block diagram for. In these drawings, the same reference numerals as those in FIGS. 1 and 2 are used for the same components as those in FIGS.

  As illustrated in FIG. 7, the encoding apparatus 10 according to the present embodiment includes a frame buffer 1011, a linear prediction analysis unit 1012, a nonlinear quantization unit 11, a parameter calculation unit 12, a coefficient encoding unit 13, a linear prediction coefficient conversion unit 1015, A linear prediction unit 1016, a subtraction unit 1017, a residual encoding unit 1018, and a synthesis unit 1019 are included. As shown in FIG. 8A, the parameter calculation unit 12 of this embodiment includes an inverse quantization unit 12a, a weight coefficient multiplication unit 12b, a subtraction unit 12c, and a parameter quantization unit 12d. As shown in FIG. 8B, the coefficient encoding unit 13 of this embodiment includes a PARCOR coefficient encoding unit 13a and a parameter encoding unit 13b. As shown in FIG. 9, the decoding device 20 according to the present embodiment includes a separation unit 1021, a coefficient decoding unit 21, a PARCOR coefficient calculation unit 22, a linear prediction coefficient conversion unit 23, a residual decoding unit 1023, a linear prediction unit 1025, And an adder 1026. Also, as shown in FIG. 10A, the coefficient decoding unit 21 of this embodiment includes a PARCOR coefficient decoding unit 21a and a parameter decoding unit 21b. Also, as shown in FIG. 10B, the PARCOR coefficient calculation unit 22 of this embodiment includes inverse quantization units 22a and 22c, a weight coefficient multiplication unit 22b, and an addition unit 22d.

  Note that the encoding device 10 and the decoding device 20 of this embodiment are, for example, a known computer or a dedicated computer that includes a CPU (central processing unit), a RAM (random-access memory), a ROM (read-only memory), and the like. This is a special device configured by reading a predetermined program and executing it by the CPU. That is, the frame buffer 1011 of the encoding device 10 is a memory such as a RAM, a cache memory, and a register, for example, and includes a linear prediction analysis unit 1012, a nonlinear quantization unit 11, a parameter calculation unit 12, a coefficient encoding unit 13, a linear encoding unit, and the like. The prediction coefficient conversion unit 1015, the linear prediction unit 1016, the subtraction unit 1017, the residual encoding unit 1018, and the synthesis unit 1019 are, for example, processing units that are constructed when the CPU executes a predetermined program. Further, the separation unit 1021, the coefficient decoding unit 21, the PARCOR coefficient calculation unit 22, the linear prediction coefficient conversion unit 23, the residual decoding unit 1023, the linear prediction unit 1025, and the addition unit 1026 of the decoding device 20 are, for example, predetermined by the CPU. It is a processing part constructed by executing this program. Further, at least a part of these processing units may be configured by an electronic circuit such as an integrated circuit. Furthermore, if necessary, the encoding device 10 or the decoding device 20 may be provided with a temporary memory that stores data output by the processing of each processing unit and reads the data during another processing of each processing unit. The method for realizing each processing unit is the same in the following embodiments and modifications thereof.

<Encoding method>
FIG. 11 is a flowchart for explaining the encoding method of the first embodiment. Hereinafter, the encoding method of this embodiment will be described with reference to FIG. In the following, only the processing for one frame will be described, but actually the same processing is executed for each frame.

  A sampled and quantized PCM-format time-series signal x (n) is input to the frame buffer 1011 of the encoding device 10 (FIG. 7). These time-series signals x (n) may be linearly quantized (sometimes referred to as “uniform quantization”), or companded (eg, ITU-T Recommendation G. 711, “Pulse Code Modulation (PCM) of Voice Frequencies”) may be used for nonlinear quantization (sometimes referred to as “non-uniform quantization”). Further, the time series signal x (n) may not be a signal in the PCM format but a signal that is not quantized. The frame buffer 1011 buffers time-series signals x (n) (n = 1,..., N) for one frame, and the encoding device 10 performs time-series signals x (n) (n = 1, ..., N).

  A time series signal x (n) (n = 1,..., N) for one frame is sent to the linear prediction analysis unit 1012. The linear prediction analysis unit 1012 performs linear prediction analysis from the first order to the Mth order. PARCOR coefficient k (m) (m = 1, 2,..., M) is calculated (step S10). The linear prediction analysis unit 1012 may be configured to perform linear prediction analysis of the time series signal x (n) (n = 1,..., N) as it is, or when it is input after nonlinear quantization. The configuration may be such that linear prediction analysis is performed after the sequence signal x (n) is mapped to linear quantization or other nonlinear quantization. The predicted order M is a positive integer, but the value may be a fixed value, or an optimal value may be determined based on the MDL principle (Minimum Description Length Principle) or the like. However, since the method of the present invention can be used when the prediction order M is 2 or more, only the case where the prediction order M is 2 or more will be described in this embodiment. That is, step S10 performs at least the first-order PARCOR coefficient k (1) and the second-order PARCOR coefficient k by performing linear prediction analysis on the input time-series signal x (n) (n = 1,..., N). This is a step of calculating a PARCOR coefficient k (2). If the encoding apparatus 10 sometimes selects the prediction order M = 1, the method of this embodiment may be applied only when the prediction order M ≧ 2. The PARCOR coefficient may be calculated by a sequential method such as the Levinson-Durbin method or the Burg method, or for each prediction order M such as an autocorrelation method or a covariance method. Alternatively, simultaneous equations (simultaneous equations having a linear prediction coefficient that minimizes the prediction residual) as a solution may be performed. In addition, the PARCOR coefficient k (m) takes a value between −1 and 1 inclusive, but an infinite digit operation cannot be performed by an arithmetic unit using a computer or the like, so it is actually expressed by an integer from −32768 to +32767, for example. Processing may be performed using a PARCOR coefficient k (m) mapped to a possible range.

  The calculated PARCOR coefficients k (m) (m = 1, 2,..., M) from the first order to the Mth order are sent to the nonlinear quantization unit 11 and quantized to obtain the first order to the Mth order. Quantized PARCOR coefficients i (m) (m = 1, 2,..., M) are generated (step S20). The “quantized PARCOR coefficient” may be the quantized value of the PARCOR coefficient itself or an index attached to the quantized value of the PARCOR coefficient. Further, the “quantized PARCOR coefficient” may be monotonically increasing in a broad sense with respect to an increase in the value of the PARCOR coefficient, or may be monotonically decreasing in a broad sense with respect to an increase in the value of the PARCOR coefficient. . In addition, the “quantized PARCOR coefficient” may be monotonically increasing with an increase in the absolute value of the PARCOR coefficient, or monotonically decreasing with an increase in the absolute value of the PARCOR coefficient. Also good. Also, the closer the absolute value of the PARCOR coefficient is to 1, the greater the influence that the quantization error of the PARCOR coefficient has on the linear prediction result. For this reason, it is desirable to perform nonlinear quantization in which the quantization step size is smaller as the absolute value of the PARCOR coefficient is closer to 1.

FIG. 13 is a graph for illustrating and explaining a nonlinear quantization method for a first-order PARCOR coefficient, and FIG. 14 is a graph for illustrating and explaining a nonlinear quantization method for a second-order PARCOR coefficient. is there. Here, the horizontal axis of these graphs indicates the PARCOR coefficient generated by the linear prediction analysis unit 1012, and the vertical axis indicates the index assigned to the quantized value of the PARCOR coefficient. For example, since the first-order PARCOR coefficient k (1) often takes a value near 1, the quantization step size is increased as the first-order PARCOR coefficient k (1) is closer to 1, as illustrated in FIG. Perform nonlinear quantization to make it smaller. In the example of FIG. 13, the possible range of PARCOR coefficients k (1) (-1 1) has three regions (from -1 _{p 2} region, the region from the _{p 2 p 6,} from _{p 6} 1 region) The quantization step size is different for each region. Further, since the second-order PARCOR coefficient k (2) often takes a value in the vicinity of −1, as illustrated in FIG. 14, the quantization step is increased as the second-order PARCOR coefficient k (2) is closer to −1. Perform nonlinear quantization to reduce size. The graph illustrating the non-linear quantization illustrated in FIG. 14 is obtained by inverting the sign of the horizontal axis of the graph illustrating the non-linear quantization illustrated in FIG.

  Although the case where the first-order and second-order PARCOR coefficients are nonlinearly quantized is illustrated here, a configuration in which these are linearly quantized may be used. The quantization of the third or higher order PARCOR coefficient may be linear quantization or non-linear quantization. Further, the quantization value of the PARCOR coefficient may increase monotonously in a broad sense or increase monotonously in a broad sense with respect to the increase of the PARCOR coefficient. Further, it is not always necessary to quantize the PARCOR coefficient of each order with the same quantization method, and the configuration may be such that the quantization method is different between the orders, and the number of quantization bits may be different between the orders.

  Next, between the value determined according to the first-order PARCOR coefficient k (1) calculated at step 10 and the value determined according to the second-order PARCOR coefficient k (2). A parameter determined according to the relational expression that holds is calculated (step S30).

[Details of Step S30]
In step S30, first, the inverse quantization unit 12a of the parameter calculation unit 12 (FIG. 8A) inversely quantizes the first-order quantized PARCOR coefficient i (1) output from the nonlinear quantization unit 11, A primary PARCOR coefficient k ′ (1) (corresponding to “a value determined according to the primary PARCOR coefficient”) is generated (step S31). Note that the process of dequantizing the quantized PARCOR coefficient i (m) is performed by any one of predetermined values k ′ (within the range of the PARCOR coefficient k (m) corresponding to the quantized PARCOR coefficient i (m). This is a process for obtaining m). For example, when the quantized PARCOR coefficient corresponding to the PARCOR coefficient k (m) of η1 ≦ k (m) <η2 is i (m), the PARCOR coefficient k ′ obtained by dequantizing the quantized PARCOR coefficient i (m) An example of (m) is the average value of η1 and η2.

  Next, a second-order PARCOR coefficient k (2) (corresponding to “a value determined according to the second-order PARCOR coefficient”) output from the linear prediction analysis unit 1012 by the weight coefficient multiplication unit 12b and the subtraction unit 12c. The weighted difference value k (2) −a · k ′ (1) is calculated using a predetermined weighting coefficient a and the first-order PARCOR coefficient k ′ (1) obtained by inverse quantization. (Step S32). In this example, the weighting coefficient multiplication unit 12b calculates a first multiplication value a · k ′ (1) obtained by multiplying a predetermined weighting coefficient a by a primary PARCOR coefficient k ′ (1), and the subtraction unit 12c. Subtracts the first multiplication value a · k ′ (1) from the second-order PARCOR coefficient k (2) to calculate the weighted difference value k (2) −a · k ′ (1). However, the process of subtracting k ′ (1) with k (2) as the initial value is repeated a times to calculate the weighted difference value k (2) −a · k ′ (1). It is also possible to obtain the mitsu difference value k (2) −a · k ′ (1). The weighting factor a may be positive or negative. However, when there is a correlation between the primary PARCOR coefficient and the secondary PARCOR coefficient, the positive and negative tend to be reversed. (See FIG. 5A.) The weighting factor a is preferably a negative value. As a result, the weighted difference value k (2) −a · k ′ (1) approaches 0 as the primary PARCOR coefficient and the secondary PARCOR coefficient are correlated. Moreover, -0.8 can be illustrated as a specific example of the weight coefficient a.

  Next, the parameter quantization unit 12d quantizes the weighted difference value k (2) −a · k ′ (1) calculated in step S32 to generate the parameter b, and outputs the parameter b (step b). S33). Note that the quantization performed by the parameter quantization unit 12d may be linear quantization or non-linear quantization. Further, the parameter b in this example may be the quantized value itself of the weighted difference value k (2) −a · k ′ (1), or may be an index attached to the quantized value. Good. Further, the absolute value of the parameter b increases monotonously in a broad sense with respect to an increase in the corresponding weighted difference value k (2) −a · k ′ (1). The parameter b may be monotonically increasing in a broad sense with respect to the corresponding weighted difference value k (2) −a · k ′ (1), or may be monotonously decreasing in a broad sense. The increase in absolute value of the corresponding weighted difference value k (2) −a · k ′ (1) may be monotonically increasing in a broad sense or may be monotonously decreasing in a broad sense ([[ End of description of details of step S30].

Next, the coefficient encoding unit 13 (FIG. 8B) performs the parameter code C _b corresponding to the parameter _b and the quantized PARCOR coefficients i (m ′) from the first order to the Mth order (excluding the second order). A coefficient code C _k corresponding to (m ′ = 1, 3,..., M) is generated, and information b, k (m ′) (including parameter b and first-order PARCOR coefficient k (1) ( A parameter code C _b and a coefficient code C _k corresponding to m ′ = 1, 3,..., M) are generated (step S40). In this embodiment, the parameter encoding unit 13 _b generates a parameter code C _b corresponding to the parameter b output from the parameter calculation unit 12, and the PARCOR coefficient encoding unit 13 a is output from the nonlinear quantization unit 11. A coefficient code C _k corresponding to each quantized PARCOR coefficient i (m ′) (m ′ = 1, 3,..., M) from the next to the Mth order (excluding the second order) is generated. Here, the encoding of the parameter b is performed by, for example, Rice encoding or entropy encoding. In particular, when the parameter b is biased to near 0, it is desirable from the viewpoint of improving the compression ratio to use a variable length coding method (Rice coding or the like) that assigns a code with a shorter code length as the encoding target is closer to 0. . Further, the encoding of each quantized PARCOR coefficient i (m ′) (m ′ = 1, 3,..., M) is performed by, for example, entropy encoding. Each quantized PARCOR coefficient i (m ′) (m ′ = 1, 3,..., M) is encoded using each quantized PARCOR coefficient i (m ′) as a separate encoding target. Alternatively, all the quantized PARCOR coefficients i (m ′) (m ′ = 1, 3,..., M) may be performed as one encoding target.

  The quantized PARCOR coefficients i (m) (m = 1, 2,..., M) from the first order to the Mth order generated in step S20 are also sent to the linear prediction coefficient conversion unit 1015 (FIG. 7). . Using these, the linear prediction coefficient conversion unit 1015 calculates each linear prediction coefficient α (m) (m = 1, 2,..., M) of the linear prediction filter of the prediction order M (step S50). This calculation is performed by applying a part of the Levinson algorithm after dequantizing the quantized PARCOR coefficient i (m), for example.

  The linear prediction unit 1016 and the subtraction unit 1017 include time series signals x (n) (n = 1,..., N) for one frame and linear prediction coefficients α (m) (m = 1, 2,... ., M) and the prediction residual e (n) (n = 1,..., N) is calculated by the prediction filter process (step S60). Note that the linear prediction unit 1016 and the subtraction unit 1017 perform prediction filter processing using, for example, the time series signal x (n) (n = 1,..., N) as it is. However, when the time series signal x (n) (n = 1,..., N) is a nonlinearly quantized signal, the linear prediction unit 1016 and the subtraction unit 1017 perform the time series signal x (n). (N = 1, ..., N) may be mapped to the linear quantization domain or other nonlinear quantum domain before performing prediction filter processing, or the linear prediction value y (n) may be mapped to the nonlinear quantum domain Then, the prediction filter process may be performed.

The calculated prediction residuals e (n) (n = 1,..., N) are sent to the residual encoding unit 1018, where they are entropy encoded, and the prediction residuals e (n) (n = 1, n ..., residual code C _e corresponding to N) is generated (step S70). Note that the residual encoding unit 1018 may perform entropy encoding with each prediction residual e (n) (n = 1,..., N) as separate encoding targets, Entropy encoding may be performed with the prediction residual e (n) (n = 1,..., N) as one encoding target. Coefficient encoding unit 13 parameter codes C _b generated by the coefficient code C _k, the residual coding unit 1018 residual code C _e generated by and sent to the synthesis unit 1019, where it is synthesized code C _g is generated (step S80). Then, the encoding device 10 outputs the generated code _Cg .

<Decoding method>
FIG. 12 is a flowchart for explaining the decoding method according to the first embodiment. Hereinafter, the decoding method of this embodiment will be described with reference to FIG. In the following, only the processing for one frame will be described, but actually the same processing is executed for each frame.

The separation unit 1021 of the decoding device 20 (FIG. 9) separates the code C _g input to the decoding device 20, and the parameter code C _b corresponding to the parameter _b from the first to the Mth order (excluding the second order). quantized PARCOR coefficients i (m ') (m' = 1,3, ..., M) and coefficient code C _k corresponding to the prediction residuals e (n) (n = 1 , ..., n ) corresponding to generate the residual code C _e (step S110). Residual decoder 1023 decodes the residual code C _e outputted from the separation unit 1021, the prediction residuals e (n) (n = 1 , ..., N) to produce a (step S120). Further, the coefficient decoding unit 21 outputs a code C _b corresponding to information including the parameter code C _b and the coefficient code C _k output from the separation unit 1021 (the parameter b and the primary PARCOR coefficient k (1)). , C _k ), and generates parameter b and quantized PARCOR coefficients i (m ′) (m ′ = 1, 3,..., M) (step S130). In this embodiment, parameter decoding section 21b of the coefficient decoding portion 21 (FIG. 10 (A)) performs decoding of parameter codes C _b, PARCOR coefficient decoding portion 21a decodes the coefficient code C _k. The first-order quantized PARCOR coefficient i (1) obtained by decoding by the PARCOR coefficient decoding unit 21a corresponds to “decoded value corresponding to the first-order PARCOR coefficient”.

  Next, the PARCOR coefficient calculation unit 22 uses the parameter b output from the coefficient decoding unit 21 and the first-order quantized PARCOR coefficient i (1), and uses the second-order PARCOR coefficient k ′ (2) (“second order”. (Corresponding to “restored value of PARCOR coefficient of”) (step S140).

[Details of Step S140]
In step S140, the inverse quantization unit 22a inversely quantizes the quantized PARCOR coefficients i (m ′) (m ′ = 1, 3,..., M) output from the coefficient decoding unit 21. PARCOR coefficients k ′ (m ′) (m ′ = 1, 3,..., M) up to the Mth order (excluding the second order) are generated (step S141). Further, the inverse quantization unit 22c inversely quantizes the parameter b output from the coefficient decoding unit 21, and generates an inverse quantization value b ′ of the parameter b (step S142). Then, the weight coefficient multiplication unit 22b and the addition unit 22d perform the first-order PARCOR coefficient k ′ (1) obtained by inverse quantization, the predetermined weight coefficient a, and the inverse quantization value b of the parameter b. Is used to generate a second-order PARCOR coefficient k ′ (2) = a · k ′ (1) + b ′ (step S143). In the present embodiment, the weight coefficient multiplication unit 22b multiplies the first multiplication value a · k ′ (1) obtained by multiplying the predetermined weight coefficient a by the first-order PARCOR coefficient k ′ (1) obtained by inverse quantization. ) And the addition unit 22d adds the first multiplication value a · k ′ (1) and the inverse quantized value b ′ of the parameter b to generate the second-order PARCOR coefficient k ′ (2). To do. However, the process of adding k ′ (1) using the inverse quantized value b ′ as an initial value is repeated a times to calculate the second-order PARCOR coefficient k ′ (2) = a · k ′ (1) + b ′. For example, the second-order PARCOR coefficient k ′ (2) may be obtained by other methods (end of description of [Details of Step S140]).

  The PARCOR coefficient k ′ (m ′) (m ′ = 1, 3,..., M) output from the inverse quantization unit 22a and the PARCOR coefficient k ′ (2) output from the adder 22d are: It is sent to the linear prediction coefficient conversion unit 23 (FIG. 9). The linear prediction coefficient conversion unit 23 calculates each linear prediction coefficient α (m) (m = 1, 2,..., M) of the linear prediction filter of the prediction order M using these. The linear prediction unit 1025 and the addition unit 1026 output the linear prediction coefficients α (m) (m = 1, 2,..., M) output from the linear prediction coefficient conversion unit 23 and the residual decoding unit 1023. Using the prediction residual e (n) (n = 1, 2,..., N) and the time series signal x (n) (n = 1, 2,. ) Is generated (step S160).

[Second Embodiment]
Next, a second embodiment of the present invention will be described.

  In the second embodiment, the following configuration will be described in the framework described in [First Aspect] of [Principle] described above. However, this does not limit the present invention.

  Relational expression: A value determined according to the second-order PARCOR coefficient by the sum of the first multiplication value obtained by multiplying a predetermined weighting coefficient by a value determined according to the first-order PARCOR coefficient and the first variable value. An expressible equation.

  A value determined according to the first-order PARCOR coefficient: a first-order quantized PARCOR coefficient obtained by quantizing the first-order PARCOR coefficient.

  A value determined according to the second-order PARCOR coefficient: a second-order quantized PARCOR coefficient obtained by quantizing the second-order PARCOR coefficient.

  Parameter: A first variable value that satisfies the above relational expression. This is a “value determined according to the first variable value”, and the absolute value of the parameter increases monotonously in a broad sense with respect to the increase in the first variable value.

  -Code of step (III): Code corresponding to information including parameters and first-order PARCOR coefficients.

  Below, it demonstrates centering around difference with 1st Embodiment, and abbreviate | omits description about the matter which is common in 1st Embodiment.

<Configuration>
A difference in configuration between the second embodiment and the first embodiment is a parameter calculation unit of the encoding device 10 and a PARCOR coefficient calculation unit of the decoding device 20.

  FIG. 15A is a block diagram for explaining the details of the nonlinear quantization unit 11 and the parameter calculation unit 112 of the second embodiment, and FIG. 15B is a PARCOR coefficient calculation unit of the second embodiment. 12 is a block diagram for explaining details of 122; FIG. Since the configuration other than the parameter calculation unit 112 and the PARCOR coefficient calculation unit 122 is the same as that of the first embodiment, the description thereof is omitted.

  As shown in FIG. 15A, the parameter calculation unit 112 includes a weight coefficient addition unit 112b and a subtraction unit 112c. As shown in FIG. 15B, the PARCOR coefficient calculation unit 122 includes an inverse quantization unit 122a, a weight coefficient multiplication unit 122b, and an addition unit 122c.

<Encoding method>
FIG. 16 is a flowchart for explaining the encoding method according to the second embodiment. Hereinafter, the encoding method of the second embodiment will be described with reference to FIG.

  The difference between the second embodiment and the first embodiment is that in the second embodiment, the process of step S230 is executed instead of step S30 (FIG. 11). Below, only the process of step S230 is demonstrated.

  In step S230, the parameter calculation unit 112 (FIG. 15A) determines between the value determined according to the first-order PARCOR coefficient k (1) and the value determined according to the second-order PARCOR coefficient k (2). A parameter determined according to the relational expression that holds is calculated. Specifically, the parameter calculation unit 112 calculates the primary and secondary quantization PARCOR coefficients i (1) and i (2) output from the nonlinear quantization unit 11 and a predetermined weighting coefficient a. Then, the parameter b = i (2) −a · i (1) is calculated (step S230). In this embodiment, the first-order quantized PARCOR coefficient i (1) is a value that increases monotonically in a broad sense with respect to an increase in at least a positive first-order PARCOR coefficient k (1), and the second-order quantized PARCOR coefficient. The coefficient i (2) is a value that increases monotonically in a broad sense with respect to an increase in at least a negative second-order PARCOR coefficient k (2). Alternatively, the first-order quantized PARCOR coefficient i (1) is a value that decreases monotonically in a broad sense with respect to an increase in at least a positive first-order PARCOR coefficient k (1), and the second-order quantized PARCOR coefficient i (2 ) Is a value that decreases monotonically in a broad sense with respect to an increase in at least a negative second-order PARCOR coefficient k (2).

  When the number of quantization bits is different between the primary PARCOR coefficient and the secondary PARCOR coefficient, a weighting coefficient a is set so that the difference can be corrected by scaling. For example, when the number of quantization bits of the primary PARCOR coefficient is 5 bits (32 stages) and the number of quantization bits of the secondary PARCOR coefficient is 4 bits (16 stages), the primary quantization PARCOR coefficient The scale of i (1) ½ is equal to the scale of the second-order quantized PARCOR coefficient i (2). In this case, it is desirable that a value obtained by multiplying 1/2 by the original weight is used as the weight coefficient a.

  Further, in this example, the weight coefficient multiplication unit 212b calculates a first multiplication value a · i (1) obtained by multiplying a predetermined weight coefficient a by a first-order quantized PARCOR coefficient i (1), and performs subtraction. The unit 112c calculates the parameter b = i (2) −a · i (1) by subtracting the first multiplication value a · i (1) from the second-order quantized PARCOR coefficient i (2). However, the parameter b may be obtained by other methods such as calculating the parameter b by repeating the process of subtracting i (1) with i (2) as an initial value a times.

<Decoding method>
FIG. 17 is a flowchart for explaining the decoding method according to the second embodiment. Hereinafter, the decoding method according to the second embodiment will be described with reference to FIG.

  The difference between the second embodiment and the first embodiment is that, in the second embodiment, the process of step S340 is executed instead of step S140 (FIG. 12). Below, only the process of step S340 is demonstrated.

  In step S340, the PARCOR coefficient calculation unit 122 (FIG. 15B) uses the parameter b output from the coefficient decoding unit 21 and the first-order quantized PARCOR coefficient i (1) to generate a second-order PARCOR coefficient k. '(2) (corresponding to “restored value of second-order PARCOR coefficient”) is calculated (step S340).

  Specifically, first, the weight coefficient multiplication unit 122b and the addition unit 122c of the PARCOR coefficient calculation unit 122 perform the first-order quantized PARCOR coefficient i (1) (“first-order PARCOR coefficient The second-order quantized PARCOR coefficient i (2) = a · i () using a predetermined weight coefficient a and the parameter b output from the coefficient decoding unit 21. 1) Generate + b (step S341). In the present embodiment, the weight coefficient multiplication unit 122b generates a first multiplication value a · i (1) obtained by multiplying a predetermined weight coefficient a by a first-order quantized PARCOR coefficient i (1), and an addition unit 122d adds the first multiplication value a · i (1) and the parameter b to generate a second-order quantized PARCOR coefficient i (2). However, the second-order quantized PARCOR coefficient i (2) is calculated by other methods, such as calculating the second-order quantized PARCOR coefficient i (2) by repeating the process of adding i (1) a time with the parameter b as an initial value. You may ask for 2).

  Next, the inverse quantization unit 122a outputs the quantized PARCOR coefficient i (m ′) (m ′ = 1, 3,..., M) output from the coefficient decoding unit 21 and the adder 122c. Quantized PARCOR coefficients i (m) (m = 1, 2,..., M) from the first order to the Mth order, which are composed of the quantized PARCOR coefficients i (2), are dequantized, and the first order to the Mth order. PARCOR coefficients k ′ (m) (m = 1, 2,..., M) are generated (step S342).

[Third Embodiment]
Next, a third embodiment of the present invention will be described.

  In the third embodiment, the following configuration will be described in the framework described in [First Mode] of [Principle] described above. However, this does not limit the present invention.

  Relational expression: a value determined according to the primary PARCOR coefficient by the sum of a second multiplication value obtained by multiplying a predetermined weighting coefficient by a value determined according to the secondary PARCOR coefficient and a second variable value. An expressible equation.

  A value determined according to the first order PARCOR coefficient: the first order PARCOR coefficient. This is a “value that increases monotonously in a broad sense with respect to an increase in the first-order PARCOR coefficient”.

  A value determined according to the second order PARCOR coefficient: the second order PARCOR coefficient. This is “a value that increases monotonically in a broad sense with respect to an increase in the second-order PARCOR coefficient”.

Parameter: a second quantized variable value obtained by quantizing a second variable value that satisfies the above relational expression. This is a “value determined according to the second variable value”, and the absolute value of the parameter increases monotonously in a broad sense as the second variable value increases. The “second quantized variable value” may be the quantized value itself of the second variable value, or may be an index attached to the quantized value of the second variable value (the same applies hereinafter). )
Step (III) code: a code corresponding to information including parameters and secondary PARCOR coefficients.

  Below, it demonstrates centering around difference with 1st Embodiment, and abbreviate | omits description about the matter which is common in 1st Embodiment.

<Configuration>
FIG. 18 is a block diagram for explaining a functional configuration of the encoding apparatus 210 according to the third embodiment. FIG. 19 is a block diagram for explaining a functional configuration of the decoding device 320 according to the third embodiment. 20A is a block diagram for explaining the details of the nonlinear quantization unit 11 and the parameter calculation unit 212 shown in FIG. 18, and FIG. 20B shows the PARCOR coefficient shown in FIG. 4 is a block diagram for explaining details of a calculation unit 222. FIG. In these drawings, the same reference numerals as those in the first embodiment are used for the portions already described in the first embodiment, and the description thereof is simplified.

  As shown in these drawings, the difference in configuration between the third embodiment and the first embodiment is a parameter calculation unit 212 of the encoding device 210 and a PARCOR coefficient calculation unit 222 of the decoding device 220. Since the configuration other than these is the same as that of the first embodiment, the description thereof is omitted.

  As illustrated in FIG. 20A, the parameter calculation unit 212 includes an inverse quantization unit 212a, a weight coefficient multiplication unit 212b, a subtraction unit 212c, and a parameter quantization unit 212d. As illustrated in FIG. 20B, the PARCOR coefficient calculation unit 222 includes inverse quantization units 222a and 222c, a weight coefficient multiplication unit 222b, and an addition unit 122c.

<Encoding method>
FIG. 21 is a flowchart for explaining an encoding method according to the third embodiment. Hereinafter, the encoding method of the third embodiment will be described with reference to FIG.

  The difference between the third embodiment and the first embodiment is that, in the third embodiment, the process of step S430 is executed instead of step S30 (FIG. 11), and step S440 instead of step S40. This is the point where the process is executed. Below, only the process of step S430 and S440 is demonstrated.

  In step S430, the parameter calculation unit 212 determines the relationship between the value determined according to the primary PARCOR coefficient k (1) and the value determined according to the secondary PARCOR coefficient k (2). A fixed parameter is calculated (step S430).

[Details of Step S430]
In step S430, first, the inverse quantization unit 212a of the parameter calculation unit 212 inversely quantizes the second-order quantized PARCOR coefficient i (2) output from the nonlinear quantization unit 11, and the second-order PARCOR coefficient k ′. (2) (corresponding to “a value determined according to the second-order PARCOR coefficient”) is generated (step S431).

  Next, the first-order PARCOR coefficient k (1) (corresponding to “a value determined according to the first-order PARCOR coefficient”) output from the linear prediction analysis unit 1012 by the weight coefficient multiplication unit 212b and the subtraction unit 212c. The weighted difference value k (1) −a · k ′ (2) is calculated using a predetermined weighting coefficient a and a second-order PARCOR coefficient k ′ (2) obtained by inverse quantization. (Step S432). In this example, the weight coefficient multiplication unit 312b calculates a second multiplication value a · k ′ (2) obtained by multiplying a predetermined weight coefficient a by a secondary PARCOR coefficient k ′ (2), and the subtraction unit 212c. Subtracts the second multiplication value a · k ′ (1) from the first-order PARCOR coefficient k (1) to calculate the weighted difference value k (1) −a · k ′ (2). However, the process of subtracting k ′ (2) with k (1) as the initial value is repeated a times to calculate the weighted difference value k (1) −a · k ′ (2). It is also possible to obtain the mitsu difference value k (1) −a · k ′ (2).

  Next, the parameter quantization unit 212d quantizes the weighted difference value k (1) −a · k ′ (2) calculated in step S432 to generate and output the parameter b (step S433). Note that the quantization performed by the parameter quantization unit 12d may be linear quantization or non-linear quantization. Further, the parameter b in this example may be the quantized value itself of the weighted difference value k (1) −a · k ′ (2), or may be an index attached to the quantized value. Good. Further, the absolute value of the parameter b increases monotonously in a broad sense with respect to an increase in the corresponding weighted difference value k (1) −a · k ′ (2). The parameter b may be monotonically increasing in a broad sense with respect to the corresponding weighted difference value k (1) −a · k ′ (2), or may be monotonously decreasing in a broad sense. May increase monotonously in a broad sense or increase monotonously in a broad sense with respect to an increase in the absolute value of the weighted difference value k (1) −a · k ′ (2) ([Step End of description of details of S430].

Next, the coefficient encoding unit 13 performs a parameter code C _b corresponding to the parameter _b, and quantized PARCOR coefficients i (m ″) (m ″ = 2, 3,... , M) to generate a coefficient code C _k and information b, k (m ″) (m ″ = 2, 3,... Including parameter b and second-order PARCOR coefficient k (2). ., M), a parameter code C _b and a coefficient code C _k are generated (step S440).

<Decoding method>
FIG. 22 is a flowchart for explaining a decoding method according to the third embodiment. Hereinafter, the decoding method according to the third embodiment will be described with reference to FIG.

  The difference between the third embodiment and the first embodiment is that, in the third embodiment, the process of step S510 is executed instead of step S110 (FIG. 12), and step S540 instead of step S140. This is the point where the process is executed. Hereinafter, only the processes of steps S510 and S540 will be described.

In step S510, the separation unit 1021 of the decoding device 220 separates the code C _g input to the decoding device 220, and the parameter code C _b corresponding to the parameter _b and the quantized PARCOR coefficients i from the second order to the M order. (m '') (m '' = 2,3, ..., M) corresponding to the coefficient code C _k and the prediction residual e (n) (n = 1,2, ..., N) corresponding to generate the residual code C _e (step S510).

  In step S540, the PARCOR coefficient calculation unit 222 uses the parameter b output from the coefficient decoding unit 21 and the second-order quantized PARCOR coefficient i (2), and the first-order PARCOR coefficient k ′ (1) (“1 (Corresponding to “restored value of the next PARCOR coefficient”) (step S540).

[Details of Step S540]
In step S540, the inverse quantization unit 222a of the PARCOR coefficient calculation unit 222 performs the quantization PARCOR coefficient i (m ″) (m ″ = 2, 3,..., M) output from the coefficient decoding unit 21. Are inversely quantized to generate PARCOR coefficients k ′ (m ″) (m ′ = 2, 3,..., M) from the second order to the Mth order (step S541). Further, the inverse quantization unit 222c inversely quantizes the parameter b output from the coefficient decoding unit 21, and generates an inverse quantized value b ′ of the parameter b (step S542). Then, the weight coefficient multiplication unit 222b and the addition unit 222d perform a second-order PARCOR coefficient k ′ (2) obtained by inverse quantization, a predetermined weight coefficient a, and an inverse quantization value b of the parameter b. Is used to generate a first-order PARCOR coefficient k ′ (1) = a · k ′ (2) + b ′ (step S543). In the present embodiment, the weight coefficient multiplication unit 222b uses a second multiplication value a · k ′ (2) obtained by multiplying a predetermined weight coefficient a by a second-order PARCOR coefficient k ′ (2) obtained by inverse quantization. ) And the addition unit 222d generates the first-order PARCOR coefficient k ′ (1) by adding the second multiplication value a · k ′ (2) and the inverse quantized value b ′ of the parameter b. To do. However, the first-order PARCOR coefficient k ′ (1) = a · k ′ (2) + b ′ is calculated by repeating the process of adding k ′ (2) with the inverse quantized value b ′ as an initial value a times. For example, the first-order PARCOR coefficient k ′ (1) may be obtained by other methods (end of description of [Details of Step S540]).

[Modification 1 of Third Embodiment]
The third embodiment reverses the handling of the first-order PARCOR coefficient and the second-order PARCOR coefficient in the first embodiment, and encodes the PARCOR coefficient by a code corresponding to information including the second-order PARCOR coefficient and the parameter. The first-order PARCOR coefficient can be restored from this code. However, the handling of the primary PARCOR coefficient and the secondary PARCOR coefficient in the second embodiment is reversed, and the PARCOR coefficient is encoded by a code corresponding to the information including the secondary PARCOR coefficient and the parameter. The form which can restore | restore a primary PARCOR coefficient may be sufficient.

  That is, the following configurations may be included in the framework described in [First Mode] of [Principle] described above.

  Relational expression: a value determined according to the primary PARCOR coefficient by the sum of a second multiplication value obtained by multiplying a predetermined weighting coefficient by a value determined according to the secondary PARCOR coefficient and a second variable value. An expressible equation.

  A value determined according to the first-order PARCOR coefficient: a first-order quantized PARCOR coefficient obtained by quantizing the first-order PARCOR coefficient.

  A value determined according to the second-order PARCOR coefficient: a second-order quantized PARCOR coefficient obtained by quantizing the second-order PARCOR coefficient.

  Parameter: A second variable value that satisfies the above relational expression. This is a “value determined according to the second variable value”, and the absolute value of the parameter increases monotonously in a broad sense as the second variable value increases.

  Step (III) code: a code corresponding to information including parameters and secondary PARCOR coefficients.

[Modification 2 of the third embodiment]
Compare the amount of code when the PARCOR coefficient is encoded by the method of the first embodiment, when the PARCOR coefficient is encoded by the method of the third embodiment, and when the PARCOR coefficient is encoded by the conventional method. The configuration may be such that an encoding method with the smallest code amount is selected for each frame. FIG. 23 is a block diagram for describing a functional configuration of an encoding apparatus 310 according to Modification 2 of the third embodiment. In this example, the parameter calculation unit 12 generates a parameter (denoted as b ₁ ) as described in the first embodiment, and the parameter calculation unit 212 uses the parameter (b ₂ as described in the third embodiment). Is written). The coefficient coding unit 313 outputs the total code amount of the code of the parameter b _{1 and} the code of the first-order quantized PARCOR coefficient i (1) output from the parameter calculation unit 12 and the parameter calculation unit 212. The code amount of the parameter b _{2 and} the code of the second-order quantized PARCOR coefficient i (2), the code of the first-order quantized PARCOR coefficient i (1) and the second-order quantized PARCOR coefficient i ( The total code amount with the code of 2) is compared, and an encoding method that minimizes the total code amount is selected.

  Also, for other combinations, a configuration in which the encoding amount when the PARCOR coefficient is encoded by the method of any form or modification or the conventional method is compared, and the encoding method having the smallest encoding amount is selected for each frame It may be.

[Fourth Embodiment]
Next, a fourth embodiment of the present invention will be described. In the fourth embodiment, when the primary PARCOR coefficient is equal to or greater than a predetermined threshold, the encoding method using the correlation between the primary PARCOR coefficient and the secondary PARCOR coefficient of the present invention is executed. When the PARCOR coefficient is less than a predetermined threshold, the first-order PARCOR coefficient and the second-order PARCOR coefficient are encoded independently of each other. Below, the following structures are demonstrated among the framework demonstrated by the [1st aspect] of the above-mentioned [principle]. However, this does not limit the present invention.

  Steps (II) and (III) are steps that are executed when the absolute value of the first-order PARCOR coefficient is equal to or greater than a predetermined first threshold value. If the absolute value of the primary PARCOR coefficient is less than the first threshold, a step of generating a code corresponding to information including the primary PARCOR coefficient and the secondary PARCOR coefficient is executed.

  Step (V) uses a parameter and a decoded value corresponding to the first order PARCOR coefficient when the absolute value of the decoded value corresponding to the first order PARCOR coefficient is equal to or larger than a predetermined second threshold value. This is a step of calculating a restoration value of the secondary PARCOR coefficient.

  Relational expression: A value determined according to the second-order PARCOR coefficient by the sum of the first multiplication value obtained by multiplying a predetermined weighting coefficient by a value determined according to the first-order PARCOR coefficient and the first variable value. An expressible equation.

  A value determined according to the first order PARCOR coefficient: the first order PARCOR coefficient. This is a “value that increases monotonously in a broad sense with respect to an increase in the first-order PARCOR coefficient”.

  A value determined according to the second order PARCOR coefficient: the second order PARCOR coefficient. This is “a value that increases monotonically in a broad sense with respect to an increase in the second-order PARCOR coefficient”.

  Parameter: a first quantized variable value obtained by quantizing a first variable value that satisfies the above relational expression. This is a “value determined according to the first variable value”, and the absolute value of the parameter increases monotonously in a broad sense with respect to the increase in the first variable value.

  -Code of step (III): Code corresponding to information including parameters and first-order PARCOR coefficients.

<Configuration>
FIG. 24 is a block diagram for explaining a functional configuration of the encoding device 410 according to the fourth embodiment. FIG. 25 is a block diagram for explaining the details of the nonlinear quantization unit 11, the parameter calculation unit 12, and the selection unit 411 shown in FIG. FIG. 26 is a block diagram for explaining a functional configuration of the decoding apparatus 420 according to the fourth embodiment. FIG. 27 is a block diagram for explaining the details of the PARCOR coefficient calculation unit 422 shown in FIG. In these drawings, the same reference numerals as those in the first embodiment are used for the portions already described in the first embodiment, and the description thereof is simplified.

  As shown in these drawings, the difference in configuration between the fourth embodiment and the first embodiment is that, in the fourth embodiment, the selection unit 411 of the encoding device 410 and the PARCOR coefficient calculation unit of the decoding device 420 are the same. 422. Since the configuration other than these is the same as that of the first embodiment, the description thereof is omitted.

  As illustrated in FIG. 24, the encoding device 410 includes a selection unit 411 in addition to the functional configurations included in the encoding device 10 of the first embodiment. As illustrated in FIG. 25, the selection unit 411 includes a determination unit 411a and a switching unit 411b. As illustrated in FIG. 27, the PARCOR coefficient calculation unit 422 includes inverse quantization units 22a and 22c, a weight coefficient multiplication unit 22b, an addition unit 22d, a determination unit 422a, and a switching unit 422b.

<Encoding method>
FIG. 28 is a flowchart for explaining an encoding method according to the fourth embodiment. Hereinafter, the encoding method of the fourth embodiment will be described using FIG. 28 with a focus on differences from the first embodiment.

  The encoding device 410 first executes the process of step S10 described in the first embodiment. After the processing of step S10, the PARCOR coefficients k (m) (m = 1, 2,..., M) from the first order to the Mth order output from the linear prediction analysis unit 1012 are selected by the selection unit 411 (FIG. 25). Is input. The determination unit 411a of the selection unit 411 determines whether or not the absolute value of the primary PARCOR coefficient k (1) is greater than or equal to a predetermined threshold value (corresponding to a “first threshold value”) (step S630). This threshold value (corresponding to the “first threshold value”) is determined in a range (0 or more and 1 or less) that the absolute value of the PARCOR coefficient can take. Note that this threshold determination may be performed in an area before quantization or in an area after quantization. However, as described later, when the same threshold value determination is performed at the time of decoding and information on the encoding method is shared between the encoding device 410 and the decoding device 420, the threshold value is determined between the encoding time and the decoding time due to the quantization error. It is desirable that the threshold determination in step S630 be performed in the region after quantization or the region inversely quantized therefrom so that the determination results do not differ. For example, when the threshold determination in step S630 is performed in the quantized region, the determination unit 411a converts the input first-order PARCOR coefficient k (1) into the first-order quantized PARCOR coefficient i (1). From this, the threshold value of the first-order quantized PARCOR coefficient i (1) is determined. In this case, the threshold used for threshold determination is the threshold converted into the quantized area. In addition, when the magnitude relationship of values is inverted between the pre-quantization region and the post-quantization region, the size relationship in threshold determination is inverted between the pre-quantization region and the post-quantization region. Thus, the threshold determination performed in the quantized region or the like corresponds to “determining whether or not the absolute value of the primary PARCOR coefficient k (1) is equal to or greater than a predetermined threshold”. In addition, when the information to be determined is a discrete value such as a region after quantization, the determination target is smaller than the threshold value in order to determine whether the determination target is equal to or greater than a predetermined threshold value. It is also possible to perform processing for determining whether or not the discrete value adjacent to the threshold value is exceeded.

  If it is determined in step S630 that the absolute value of the primary PARCOR coefficient k (1) is greater than or equal to a predetermined threshold (corresponding to the “first threshold”), the process branching by the switching unit 411b According to the control, the processing of steps S20 to S80 (FIG. 11) described in the first embodiment is executed.

On the other hand, when it is determined that the absolute value of the primary PARCOR coefficient k (1) is less than a predetermined threshold value (corresponding to the “first threshold value”), the first embodiment is performed according to the processing branch control by the switching unit 411b. After the processing of step S20 described in step S20, the coefficient encoding unit 13 (FIG. 24) performs first-order to M-order quantized PARCOR coefficients i (m) (m = 1, 2,..., M) corresponding to a coefficient code C _k (“a code corresponding to information including a first-order PARCOR coefficient k (1) and a second-order PARCOR coefficient k (2)”) Equivalent) is generated (step S640). After the process of step S640, the processes of steps S50 to S70 described in the first embodiment are executed, and then the synthesis unit 1019 synthesizes the coefficient code C _k and the residual code C _e to obtain a quantized PARCOR coefficient. Generate code C _g corresponding to information including i (m) (m = 1,2, ..., M) and prediction residual e (n) (n = 1,2, ..., N) (Step S650).

<Decoding method>
FIG. 29 is a flowchart for explaining a decoding method according to the fourth embodiment. Hereinafter, the encoding method of the fourth embodiment will be described using FIG. 29 with a focus on differences from the first embodiment.

First, the decoding apparatus 420 executes the processing of steps S110 and S120 described in the first embodiment, and the coefficient decoding unit 21 decodes the coefficient code C _k to obtain the quantized PARCOR coefficient i (m ′) (m ′). = 1, 3, ..., M) is generated (step S730). Thereafter, the determination unit 422a of the PARCOR coefficient calculation unit 422 determines that the first-order quantized PARCOR coefficient i (1) (corresponding to “decoded value corresponding to the first-order PARCOR coefficient”) output from the coefficient decoding unit 21 in advance. It is determined whether or not it is equal to or greater than a predetermined threshold (second threshold) (step S740). This threshold value (second threshold value) is obtained by converting the above-described threshold value (corresponding to “first threshold value”) into an area after quantization.

If it is determined in step S740 that the first-order quantized PARCOR coefficient i (1) is greater than or equal to a predetermined threshold (second threshold), the coefficient decoding unit 21 follows the processing branch control by the switching unit 422b. After decoding the parameter code C _b output from the separation unit 1021 and generating the parameter b (step S760), step S140 described in the first embodiment (“decoding corresponding to parameters and first-order PARCOR coefficients” (Corresponding to the step of calculating the restoration value of the second-order PARCOR coefficient using the value) (FIG. 17) is executed, and the processing of steps S150 and S160 (FIG. 12) is further executed.

  On the other hand, if it is determined in step S740 that the first-order quantized PARCOR coefficient i (1) is less than a predetermined threshold (second threshold), the PARCOR coefficient is determined according to the processing branch control by the switching unit 422b. The inverse quantization unit 22a of the calculation unit 422 calculates the quantized PARCOR coefficients i (m) (m = 1, 2,..., M) from the first order to the Mth order obtained by the decoding of the coefficient decoding unit 21. Inverse quantization is performed to generate PARCOR coefficients k ′ (m) (m = 1, 2,..., M) from the first order to the Mth order (step S750). Note that, when the magnitude relationship between the pre-quantization region and the post-quantization region is reversed, the magnitude relationship in the threshold determination in step S740 is also reversed.

  After step S750, the processes of steps S150 and S160 described in the first embodiment are executed.

[Modification 1 of Fourth Embodiment]
The first modification of the fourth embodiment also uses the correlation between the primary PARCOR coefficient and the secondary PARCOR coefficient according to the present invention when the primary PARCOR coefficient is equal to or greater than a predetermined threshold. Is executed, and when the primary PARCOR coefficient is less than a predetermined threshold, the primary PARCOR coefficient and the secondary PARCOR coefficient are encoded independently of each other. However, in the fourth embodiment, the “encoding method using the correlation between the first-order PARCOR coefficient and the second-order PARCOR coefficient” is realized by the method described in the first embodiment. In the first modification of the embodiment, this is realized by the method described in the second embodiment. That is, the first modification of the fourth embodiment relates to the following configuration in the framework described in [First aspect] of [Principle] described above. However, this does not limit the present invention.

  Steps (II) and (III) are steps that are executed when the absolute value of the first-order PARCOR coefficient is equal to or greater than a predetermined first threshold value. If the absolute value of the primary PARCOR coefficient is less than the first threshold, a step of generating a code corresponding to information including the primary PARCOR coefficient and the secondary PARCOR coefficient is executed.

  Step (V) uses a parameter and a decoded value corresponding to the first order PARCOR coefficient when the absolute value of the decoded value corresponding to the first order PARCOR coefficient is equal to or larger than a predetermined second threshold value. This is a step of calculating a restoration value of the secondary PARCOR coefficient.

  Relational expression: A value determined according to the second-order PARCOR coefficient by the sum of the first multiplication value obtained by multiplying a predetermined weighting coefficient by a value determined according to the first-order PARCOR coefficient and the first variable value. An expressible equation.

  A value determined according to the first-order PARCOR coefficient: a first-order quantized PARCOR coefficient obtained by quantizing the first-order PARCOR coefficient.

  A value determined according to the second-order PARCOR coefficient: a second-order quantized PARCOR coefficient obtained by quantizing the second-order PARCOR coefficient.

  Parameter: A first variable value that satisfies the above relational expression. This is a “value determined according to the first variable value”, and the absolute value of the parameter increases monotonously in a broad sense with respect to the increase in the first variable value.

  -Code of step (III): Code corresponding to information including parameters and first-order PARCOR coefficients.

<Configuration>
The structural differences between the first modification of the fourth embodiment and the fourth embodiment are the selection unit and parameter calculation unit of the encoding device 410 and the PARCOR coefficient calculation unit of the decoding device 420.

  FIG. 30 is a block diagram for explaining details of the nonlinear quantization unit 11, the selection unit 511, and the parameter calculation unit 122 of the encoding device 410 according to Modification 1 of the fourth embodiment. FIG. 31 is a block diagram for explaining details of the PARCOR coefficient calculation unit 522 of the decoding device 420 according to Modification 1 of the fourth embodiment. Since the configuration other than these is the same as that of the fourth embodiment, description thereof is omitted.

  As illustrated in FIG. 30, the selection unit 511 includes a determination unit 511a and a switching unit 511b, and the parameter calculation unit 122 includes a weight coefficient addition unit 112b and a subtraction unit 112c. As illustrated in FIG. 31, the PARCOR coefficient calculation unit 522 includes a determination unit 522a, a switching unit 522b, an inverse quantization unit 122a, a weight coefficient multiplication unit 122b, and an addition unit 122c.

<Encoding method>
FIG. 32 is a flowchart for explaining an encoding method according to the first modification of the fourth embodiment. Hereinafter, the encoding method of the modification 1 of 4th Embodiment is demonstrated using FIG.

  In the first modification of the fourth embodiment, first, the processes of steps S10 and S20 described in the first embodiment are executed. Next, the first to M-order quantized PARCOR coefficients i (m) (m = 1, 2,..., M) output from the nonlinear quantization unit 11 are input to the selection unit 511. The determination unit 511a of the selection unit 511 determines whether or not the absolute value of the first-order quantized PARCOR coefficient i (1) is greater than or equal to a predetermined threshold (step S830). This is equivalent to determining whether or not the absolute value of the primary PARCOR coefficient k (1) is greater than or equal to a predetermined threshold (corresponding to a “first threshold”) in the quantized region. Note that the threshold value used for threshold determination of the first-order quantized PARCOR coefficient i (1) is obtained by converting the first threshold value into a region after quantization.

  In the determination in step S830, the absolute value of the first-order quantized PARCOR coefficient i (1) is equal to or greater than a predetermined threshold (the absolute value of the first-order PARCOR coefficient k (1) is a predetermined threshold (“ Is equal to or greater than “first threshold value”), the processes of steps S230 and S40 to S80 (FIG. 16) described in the second embodiment are executed according to the process branching control by the switching unit 511b. The

  On the other hand, in the determination of step S830, the absolute value of the first-order quantized PARCOR coefficient i (1) is less than a predetermined threshold value (the absolute value of the first-order PARCOR coefficient k (1) is a predetermined threshold value). (Corresponding to “first threshold value”), the processing of steps S640, S50 to S70, and step S650 described in the fourth embodiment is performed according to the process branching control by the switching unit 511b. Executed.

<Decoding method>
FIG. 33 is a flowchart for explaining a decoding method according to the first modification of the fourth embodiment. Hereinafter, the encoding method according to the first modification of the fourth embodiment will be described with reference to FIG. 33 with a focus on differences from the fourth embodiment.

In the first modification of the fourth embodiment, first, after the processing of steps S110 and S120 described in the first embodiment is performed, the coefficient decoding unit 21 decodes the coefficient code C _k and generates a quantized PARCOR coefficient i ( m ′) (m ′ = 1, 3,..., M) is generated (step S730).

  Thereafter, the determination unit 522a of the PARCOR coefficient calculation unit 522 determines that the first-order quantized PARCOR coefficient i (1) (corresponding to “decoded value corresponding to the first-order PARCOR coefficient”) output from the coefficient decoding unit 21 in advance. It is determined whether or not it is equal to or greater than a predetermined threshold (second threshold) (step S740). This threshold value (second threshold value) is obtained by converting the first threshold value into a region after quantization.

  If it is determined in step S740 that the first-order quantized PARCOR coefficient i (1) is greater than or equal to a predetermined threshold value (second threshold value), according to the process branch control by the switching unit 522b, the fourth embodiment Step S760 described in step S760 (FIG. 29) is executed, and then step S340 described in the second embodiment (“reconstruction of the second-order PARCOR coefficient using the parameter and the decoded value corresponding to the first-order PARCOR coefficient is performed. (Corresponding to “step for calculating value”) (FIG. 17) is executed, and further, the processes of steps S150 and S160 (FIG. 12) are executed.

  On the other hand, if it is determined in step S740 that the first-order quantized PARCOR coefficient i (1) is less than a predetermined threshold (second threshold), the PARCOR coefficient is determined according to the processing branch control by the switching unit 522b. The inverse quantization unit 22a of the calculation unit 422 calculates the quantized PARCOR coefficients i (m) (m = 1, 2,..., M) from the first order to the Mth order obtained by the decoding of the coefficient decoding unit 21. Inverse quantization is performed to generate PARCOR coefficients k ′ (m) (m = 1, 2,..., M) from the first order to the Mth order (step S750). After step S750, the processes of steps S150 and S160 described in the first embodiment are executed.

[Modification 2 of the fourth embodiment]
In the fourth embodiment and the first modification thereof, the handling of the primary PARCOR coefficient and the secondary PARCOR coefficient is reversed, and the PARCOR coefficient is encoded by the code corresponding to the information including the secondary PARCOR coefficient and the parameter. The primary PARCOR coefficient can be restored from this code. That is, the following structure is taken out of the framework described in [First aspect] of [Principle] described above. However, this does not limit the present invention.

  Steps (II) and (III) are steps that are executed when the absolute value of the second-order PARCOR coefficient is equal to or greater than a predetermined first threshold value. When the absolute value of the secondary PARCOR coefficient is less than the first threshold, a step of generating a code corresponding to information including the primary PARCOR coefficient and the secondary PARCOR coefficient is executed.

  Step (V) uses a parameter and a decoded value corresponding to the second order PARCOR coefficient when the absolute value of the decoded value corresponding to the second order PARCOR coefficient is equal to or greater than a predetermined second threshold value. This is a step of calculating a restoration value of the primary PARCOR coefficient.

  Relational expression: a value determined according to the primary PARCOR coefficient by the sum of a second multiplication value obtained by multiplying a predetermined weighting coefficient by a value determined according to the secondary PARCOR coefficient and a second variable value. An expressible equation.

  A value determined according to the first order PARCOR coefficient: the first order PARCOR coefficient.

  A value determined according to the second order PARCOR coefficient: the second order PARCOR coefficient.

  Parameter: a second quantized variable value obtained by quantizing a second variable value that satisfies the above relational expression. This is a “value determined according to the second variable value” and a “value that increases monotonously in a broad sense with respect to an increase in the second variable value”.

  Step (III) code: a code corresponding to information including parameters and secondary PARCOR coefficients.

In the above configuration, the value determined according to the first-order PARCOR coefficient is the first-order PARCOR coefficient, the value determined according to the second-order PARCOR coefficient is the second-order PARCOR coefficient, and the parameters satisfy the above relational expression. Instead of the second variable value obtained by quantizing the second variable value,
A value determined according to the first-order PARCOR coefficient: a first-order quantized PARCOR coefficient obtained by quantizing the first-order PARCOR coefficient.

  A value determined according to the second-order PARCOR coefficient: a second-order quantized PARCOR coefficient obtained by quantizing the second-order PARCOR coefficient.

Parameter: A second variable value that satisfies the above relational expression.
It may be configured as follows.

[Fifth Embodiment]
Next, a fifth embodiment of the present invention will be described.

  The fifth embodiment exemplifies the framework described in [Second Aspect] of [Principle] described above. However, this does not limit the present invention.

  Below, it demonstrates centering around difference with 1st Embodiment, and abbreviate | omits description about the matter which is common in 1st Embodiment.

<Configuration>
FIG. 34 is a block diagram for explaining a functional configuration of an encoding apparatus 610 according to the fifth embodiment. FIG. 35 is a block diagram for explaining details of the quantization method selection unit 611, the quantization unit 612, and the encoding method selection unit 613 shown in FIG. 34, and FIG. 36 is the same as FIG. 5 is a block diagram for explaining details of a coefficient encoding unit 614. FIG. FIG. 37 is a block diagram for explaining a functional configuration of the decoding device 620 according to the fifth embodiment. FIG. 38 is a block diagram for explaining details of the coefficient decoding unit 621 shown in FIG.

  As shown in FIG. 34, the structural differences from the encoding device 10 in the first embodiment of the encoding device 610 are the non-linear quantization unit 11, the parameter calculation unit 12, and the coefficient encoding unit of the encoding device 10. 13 is a point replaced with a quantization method selection unit 611, a quantization unit 612, an encoding method selection unit 613, and a coefficient encoding unit 614. As illustrated in FIG. 35, the quantization method selection unit 611 of this embodiment includes a determination unit 611a and a switching unit 611b. The quantization unit 612 includes a quantization unit 612a, a low-precision quantization unit 612b, and a high-precision quantum. A conversion unit 612c. As shown in FIG. 36, the coefficient encoding unit 614 of this embodiment includes a determination unit 614a, a switching unit 614b, and variable length encoding units 614c to 614e.

  In addition, as shown in FIG. 37, the structural difference from the decoding device 20 in the first embodiment of the decoding device 620 is that the coefficient decoding unit 21 and the PARCOR coefficient calculation unit 22 of the decoding device 20 are different from each other in the coefficient decoding unit 621. And the PARCOR coefficient calculation unit 622. As shown in FIG. 38, the coefficient decoding unit 621 of this embodiment includes variable length decoding units 621a, 621d, and 621e, determination units 621b and 621f, switching units 621c and 621g, and inverse quantization units 621h to 621j.

<Encoding method>
FIG. 39 is a flowchart for explaining an encoding method according to the fifth embodiment. Hereinafter, the encoding method of this embodiment will be described with reference to FIG. In the following, only the processing for one frame will be described, but actually the same processing is executed for each frame.

  First, the linear prediction analysis unit 1012 performs the PARCOR coefficient k (m) (m = 1, 2,..., M) from the first order to the Mth order by executing the process of step S10 of the first embodiment. ) Is output (step S10).

  The output primary PARCOR coefficient k (1) is input to the encoding method selection unit 613 (FIG. 35), and the encoding method selection unit 613 determines the absolute value of the primary PARCOR coefficient k (1) in advance. The first variable length encoding method is selected as an encoding method for generating a code corresponding to the second order PARCOR coefficient k (2), and the first order PARCOR coefficient k ( When the absolute value of 1) is less than the threshold value T1, a second variable length code different from the first variable length coding method as a coding method for generating a code corresponding to the second-order PARCOR coefficient k (2) The conversion method is selected (step S910).

[Details of Step S910]
In step S910, first, the encoding method selection unit 613 determines whether or not the absolute value of the primary PARCOR coefficient k (1) is greater than or equal to a predetermined threshold T1 (step S911). This threshold value T1 is determined in advance in a range (0 or more and 1 or less) that the absolute value of the PARCOR coefficient can take. Further, this threshold determination may be performed in a region before quantization or may be performed in a region after quantization. However, as described later, since it is necessary to perform the same threshold determination at the time of decoding and to share information on the encoding method between the encoding device 610 and the decoding device 620, at the time of encoding and decoding due to quantization errors. It is desirable that the threshold determination in step S911 is performed in a region after quantization or a region inversely quantized from the region so that the threshold determination result does not differ between. For example, when the threshold determination in step S911 is performed in the quantized region, the encoding method selection unit 613 converts the input first-order PARCOR coefficient k (1) to the first-order quantized PARCOR coefficient i (1). After the conversion, the threshold value of the first-order quantized PARCOR coefficient i (1) is determined. In this case, the threshold used for threshold determination is the threshold converted into the quantized area. In addition, when the magnitude relationship of values is inverted between the pre-quantization region and the post-quantization region, the size relationship in threshold determination is inverted between the pre-quantization region and the post-quantization region. Thus, the threshold determination performed in the quantized region or the like corresponds to “determining whether or not the absolute value of the primary PARCOR coefficient k (1) is equal to or greater than a predetermined threshold”. In addition, when the information to be determined is a discrete value such as a region after quantization, the determination target is smaller than the threshold value in order to determine whether the determination target is equal to or greater than a predetermined threshold value. It is also possible to perform processing for determining whether or not the discrete value adjacent to the threshold value is exceeded. If the encoding method selection unit 613 determines that the absolute value of the primary PARCOR coefficient k (1) is equal to or greater than a predetermined threshold T1, the encoding method selection unit 613 selects the first variable length encoding method (step S912). If not, the second variable length coding method is selected, and the parameter b indicating the selected content is output (step S913). In this example, the encoding method selection unit 613 outputs b = 0 when the first variable length encoding method is selected, and sets b = 1 when the second variable length encoding method is selected. Output. Further, the first variable length coding method and the second variable length coding method are as described in [Second Aspect] of [Principle] described above, and a specific example of such a coding method is a Rice code. And the Huffman encoding method (end of description of [Details of Step S910]).

  The primary and secondary PARCOR coefficients k (1), k (2) output from the linear prediction analysis unit 1012 are also input to the quantization method selection unit 611 (FIG. 35), and the quantization method selection unit 611 Selects a predetermined first quantization method when the absolute value of the primary PARCOR coefficient k (1) is equal to or greater than a predetermined second threshold T2, and selects the primary PARCOR coefficient k (1) When the absolute value of is less than a predetermined second threshold T2, a predetermined second quantization method having a quantization step size larger than that of the first quantization method is selected, and 2 is selected depending on the selected quantization method. The next PARCOR coefficient k (2) is quantized to generate a second-order quantized PARCOR coefficient (step S920).

[Details of Step S920]
In step S920, first, the determination unit 611a of the quantization method selection unit 611 determines whether or not the input first-order PARCOR coefficient k (1) is greater than or equal to a predetermined second threshold T2 (step S920). S921). This threshold value T2 is determined in advance in a range (0 or more and 1 or less) that the absolute value of the PARCOR coefficient can take. The other details of the threshold determination in step S921 are the same as the threshold determination described in step S911, and thus the description thereof is omitted. When it is determined that the absolute value of the primary PARCOR coefficient k (1) is equal to or greater than a predetermined threshold value T2, the high-precision quantization unit 612c performs the first quantization according to the processing branch control by the switching unit 611b. The second-order PARCOR coefficient k (2) input by the method is quantized to generate and output a second-order quantized PARCOR coefficient i (2) (step S922). On the other hand, when it is determined that the absolute value of the primary PARCOR coefficient k (1) is less than the predetermined threshold T2, the low-precision quantization unit 612b performs the second quantization according to the processing branch control by the switching unit 611b. The second-order PARCOR coefficient k (2) input by the method is quantized to generate and output a second-order quantized PARCOR coefficient i (2) (step S923). The first quantization method and the second quantization method may be a method for linearly quantizing an input signal or a method for nonlinear quantization. However, the quantization step size of the first quantization method for a certain amplitude range of the input signal is smaller than the quantization step size of the second quantization method for the same amplitude range. In other words, the number of quantization steps of the first quantization method for a certain amplitude range of the input signal is larger than the number of quantization steps of the second quantization method for the same amplitude range. That is, the first quantization method is a method of performing quantization of the input signal with finer granularity than the second quantization method (end of description of [Details of Step S920]).

  Further, the first-order, third-order to M-th order PARCOR coefficients k (1), k (3),..., K (M) are input to the quantization unit 612a. The quantization unit 612a quantizes these using a predetermined fixed quantization method, and performs first-order, third-order to M-order quantized PARCOR coefficients i (1), i (3),. M) is generated and output (step S930). Note that the quantization performed by the quantization unit 612a may be linear quantization or non-linear quantization.

Next, the second-order quantized PARCOR coefficient i (2) and the parameter b are input to the coefficient encoding unit 614 (FIG. 36), and the coefficient encoding unit 614 is selected in step S910 indicated by the parameter b. Using the encoding method, the second-order quantized PARCOR coefficient i (2) is encoded to generate a coefficient code C _k (2) corresponding to the second-order PARCOR (step S940).

[Details of Step S940]
In step S940, first, the determination unit 614a (FIG. 36) determines whether or not the encoding method selected in step S910 indicated by the parameter b is the first variable length encoding method (step S931). In this example, it is determined whether b = 0. Here, when it is determined that b = 0, the variable-length encoding unit 614d converts the input second-order quantized PARCOR coefficient i (2) to the first variable length according to the processing branch control by the switching unit 614b. The coefficient code C _k (2) is generated by encoding using the encoding method and is output (step S932). On the other hand, if it is determined that b = 1, the variable-length encoding unit 614e converts the input second-order quantized PARCOR coefficient i (2) into the second variable-length code according to the processing branch control by the switching unit 614b. The coefficient code C _k (2) is generated and output by the encoding method (end of description of step S933 / [detail of step S940]).

Further, the variable length coding unit 614c fixes the input first-order, third-order to M-order quantized PARCOR coefficients i (1), i (3),..., I (M) in a predetermined fixed manner. The coefficient codes C _k (1), C _k (3),..., C _k (M) are generated and output by the variable length encoding method (step S950). An example of variable length encoding performed by the variable length encoding unit 614c is Rice encoding.

After that, after the processing of steps S50 and S60 of the first embodiment is executed, the coefficient code C _k = (C _k (1),..., C _k (M)) generated by the coefficient encoding unit 614. If, residual code C _e generated by the residual encoding unit 1018 is sent to the synthesis unit 1019 (FIG. 34), where combined with code C _g is generated (step S980). Then, the encoding device 610 outputs the generated code _Cg .

<Decoding method>
FIG. 40 is a flowchart for explaining an encoding method according to the fifth embodiment. Hereinafter, the decoding method of this embodiment will be described with reference to FIG. In the following, only the processing for one frame will be described, but actually the same processing is executed for each frame.

The separation unit 1021 (FIG. 37) of the decoding device 620 separates the code C _g input to the decoding device 20, and quantized PARCOR coefficients i (m) (m = 1, 2,. .., M) and coefficient residuals C _k = (C _k (1), ..., C _k (M)) and prediction residuals e (n) (n = 1,2, ..., generating a residual code C _e corresponding to N) (step S1010). Next, the process of step S120 of the first embodiment is executed, and the variable length decoding unit 621a of the coefficient decoding unit 621 (FIG. 38) further receives the input coefficient codes C _k (1), C _k (3). ,..., C _k (M) and generate first-order, third-order to M-order quantized PARCOR coefficients i (1), i (3),. S1030).

Furthermore, the determination unit 621b compares the absolute value of the decoded value i (1) of the coefficient code C _k (1) corresponding to the primary PARCOR coefficient with a predetermined threshold T3, and determines a predetermined first value. The coefficient code C _k (2) corresponding to the second-order PARCOR coefficient is decoded by a decoding method corresponding to the variable length coding method, or a second variable length coding determined in advance different from the first variable length coding method. It is determined whether the coefficient code C _k (2) corresponding to the second-order PARCOR coefficient is decoded by the decoding method corresponding to the method, and the variable length decoding units 621d and 621e decode the coefficient code C _k (2) (step S1040).

[Details of Step S1040]
First, the first-order quantized PARCOR coefficient i (1) that is the decoded value of the coefficient code C _k (1) output from the variable length decoding unit 621a is input to the determination unit 621b. The determination unit 621b determines whether or not the absolute value of the first-order quantized PARCOR coefficient i (1) is equal to or greater than a predetermined threshold T3. The threshold value T3 is a value obtained by quantizing the threshold value T1 in step S910 with the quantization method in step S930.

Here, when it is determined that the absolute value of the first-order quantized PARCOR coefficient i (1) is equal to or greater than a predetermined threshold T3, the variable-length decoding unit 621d inputs the input according to the processing branch control by the switching unit 621c. The obtained coefficient code C _k (2) is decoded by a decoding method corresponding to the first variable length coding method, and a second-order quantized PARCOR coefficient i (2) is generated and output (step S1042). On the other hand, when it is determined that the absolute value of the primary quantized PARCOR coefficient i (1) is less than a predetermined threshold T3, the variable length decoding unit 621e is input in accordance with the processing branch control by the switching unit 621c. The coefficient code C _k (2) is decoded by a decoding method corresponding to the second variable length coding method, and a second-order quantized PARCOR coefficient i (2) is generated and output (step S1043).

  Note that, when the magnitude relationship between values in the pre-quantization region and the post-quantization region is reversed, the magnitude relationship in the threshold determination in step S1040 is reversed (end of description of [details of step S1040]).

Next, the determination unit 621f compares the absolute value of the decoded value of the coefficient code C _k (1) corresponding to the primary PARCOR coefficient with a predetermined second threshold value T4, and determines a predetermined first inverse A quantization method is used to inverse-quantize a decoded value obtained by decoding a code corresponding to a second-order PARCOR coefficient, or a predetermined first quantization step size larger than that of the first inverse quantization method is used. 2. Determine whether to decode the decoded value obtained by decoding the code corresponding to the second-order PARCOR coefficient using the 2 inverse quantization method, and the inverse quantization units 621i and 621j inversely quantize the decoded value. (Step S1050).

[Details of Step S1050]
First, the first-order quantized PARCOR coefficient i (1) that is the decoded value of the coefficient code C _k (1) output from the variable length decoding unit 621a is input to the determination unit 621f. The determination unit 621f determines whether or not the absolute value of the first-order quantized PARCOR coefficient i (1) is greater than or equal to a predetermined threshold T4. The threshold value T4 is a value obtained by quantizing the threshold value T2 in step S920 with the quantization method in step S930.

  Here, when it is determined that the absolute value of the first-order quantized PARCOR coefficient i (1) is greater than or equal to a predetermined threshold T4, the inverse quantization unit 621i is set to 2 according to the processing branch control by the switching unit 621g. The next quantized PARCOR coefficient i (2) is inversely quantized with high accuracy to generate a second-order PARCOR coefficient k ′ (2) (step S1052). This high-precision inverse quantization is the inverse quantization of the high-precision quantization in step S922 and corresponds to the first inverse quantization. On the other hand, when it is determined that the absolute value of the primary quantized PARCOR coefficient i (1) is less than a predetermined threshold T4, the inverse quantization unit 621j performs the secondary quantization according to the processing branch control by the switching unit 621g. The quantized PARCOR coefficient i (2) is subjected to low-precision inverse quantization to generate a second-order PARCOR coefficient k ′ (2) (step S1053). This low-precision inverse quantization is the inverse quantization of the low-precision quantization in step S923 and corresponds to the second inverse quantization.

  In addition, when the magnitude relationship of the value is inverted between the area before quantization and the area after quantization, the magnitude relation in the threshold determination in step S1050 is inverted (end of description of [details of step S1050]).

  Further, the first-order, third-order to M-order quantized PARCOR coefficients i (1), i (3),..., I (M) output from the variable length decoding unit 621a are input to the inverse quantization unit 621h. Then, the inverse quantization unit 621h performs inverse quantization to generate first-order, third-order to M-order PARCOR coefficients k ′ (1), k ′ (3),..., K ′ (M). (Step S1060). This inverse quantization is the inverse quantization of the quantization in step S930.

Thereafter, steps S150 and S160 of the first embodiment are executed.
[Modification 1 of Fifth Embodiment]
The threshold determination process of steps S911 and S921 of the fifth embodiment may be integrated, and the threshold determination process of steps S1041 and S1051 may be integrated.

That is, in the fifth embodiment, in step S911 (FIG. 39), it is determined whether or not the absolute value of the primary PARCOR coefficient k (1) is greater than or equal to a predetermined threshold T1, and the secondary quantized PARCOR. A variable length encoding method for the coefficient i (2) is determined, and in step S920 (FIG. 39), it is determined whether or not the absolute value of the primary PARCOR coefficient k (1) is greater than or equal to a predetermined threshold T2. The quantization method of the PARCOR coefficient k (2) is determined. However, it is determined whether or not the absolute value of the primary PARCOR coefficient k (1) is greater than or equal to a predetermined threshold T1, and the quantization method of the secondary PARCOR coefficient k (2) is determined according to the determination result. And the variable length coding method of the second-order quantized PARCOR coefficient i (2) may be determined. For example, when the absolute value of the primary PARCOR coefficient k (1) is greater than or equal to a predetermined threshold T1, the first variable length encoding method and the first quantization method are selected, and the primary PARCOR coefficient k When the absolute value of (1) is less than a predetermined threshold T1, the second variable length encoding method and the second quantization method may be selected.
In the fifth embodiment, it is determined in step S1041 (FIG. 40) whether or not the absolute value of the first-order quantized PARCOR coefficient i (1) is greater than or equal to a predetermined threshold T3. The decoding method of the coefficient code C _k (2) corresponding to the generalized PARCOR is determined, and whether or not the absolute value of the first-order quantized PARCOR coefficient i (1) is greater than or equal to a predetermined threshold T4 in step S1051. It was determined and the inverse quantization method of the second-order quantized PARCOR coefficient i (2) was determined. However, it is determined whether or not the absolute value of the primary quantized PARCOR coefficient i (1) is greater than or equal to a predetermined threshold T3, and the coefficient code corresponding to the secondary quantized PARCOR is determined according to the determination result. Both the decoding method of C _k (2) and the inverse quantization method of the second-order quantized PARCOR coefficient i (2) may be determined. For example, when the absolute value of the first-order quantized PARCOR coefficient i (1) is equal to or greater than a predetermined threshold T3, the decoding method corresponding to the first variable length coding method and the first dequantization method are selected. When the absolute value of the first-order quantized PARCOR coefficient i (1) is less than a predetermined threshold T3, the decoding method corresponding to the second variable length coding method and the second dequantization method are selected. May be.

[Other variations]
In addition, this invention is not limited to each above-mentioned embodiment. For example, in the first to fourth embodiments described above, the coefficient encoding unit of the encoding device separately generates the coefficient code C _k corresponding to the PARCOR coefficient and the parameter code C _b corresponding to the parameter, A code composed of the coefficient code C _k and the parameter code C _b is a code corresponding to the PARCOR coefficient and the parameter. However, the coefficient encoding unit of the encoding device may generate a code corresponding to the PARCOR coefficient and the parameter, for example, by encoding a bit combination value of the quantized PARCOR coefficient and the parameter.

  In the fifth embodiment, when the absolute value of the primary PARCOR coefficient k (1) is greater than or equal to a predetermined threshold T1, a code corresponding to the secondary PARCOR coefficient k (2) is generated. When the first variable length coding method is selected as the coding method for the first time, and the absolute value of the first order PARCOR coefficient k (1) is less than the threshold value T1, the second order PARCOR coefficient k (2) is supported. A second variable length coding method different from the first variable length coding method is selected as a coding method for generating a code (step S910), and a first order PARCOR coefficient k (1) and a second order PARCOR coefficient k. (2) was encoded separately (steps S940 and S950). However, for example, the bit combination value of the primary PARCOR coefficient k (1) and the secondary PARCOR coefficient k (2) may be one encoding target. That is, when the absolute value of the primary PARCOR coefficient k (1) is equal to or greater than a predetermined threshold T1, the first variable length coding is used as an encoding method for generating a code corresponding to the bit combination value. A first variable length encoding method as an encoding method for generating a code corresponding to the bit combination value when the method is selected and the absolute value of the primary PARCOR coefficient k (1) is less than the threshold T1 A second variable length encoding method different from the above may be selected, and the primary PARCOR coefficient k (1) and the secondary PARCOR coefficient k (2) may be encoded together.

  In addition, as a relational expression established between a value determined according to the first order PARCOR coefficient and a value determined according to the second order PARCOR coefficient, a relational expression other than that described in the first to fourth embodiments is used. The PARCOR coefficient may be encoded using a parameter determined according to the relational expression. For example, an equation that can express a value determined according to the second-order PARCOR coefficient by a sum of a value determined according to the first-order PARCOR coefficient and the first variable value without using a weighting coefficient in the relational expression. And a value determined according to the first variable value that satisfies the condition may be used as a parameter. Such a relational expression is also “depending on the secondary PARCOR coefficient by the sum of the first multiplication value obtained by multiplying a predetermined weighting coefficient by a value determined according to the primary PARCOR coefficient and the first variable value. It is included in the concept of “equations that can express values determined by Further, for example, an equation that can represent a value determined according to the first-order PARCOR coefficient by the sum of the value determined according to the second-order PARCOR coefficient and the second variable value is set as the above relational expression, and A value determined according to the variable value may be used as a parameter. Such a relational expression is also “depending on the primary PARCOR coefficient by the sum of the second multiplication value obtained by multiplying a predetermined weighting coefficient by a value determined according to the secondary PARCOR coefficient and the second variable value. It is included in the concept of “equations that can express values determined by Further, for example, as this relational expression, an equation indicating a ratio between a value determined according to the first-order PARCOR coefficient and a value determined according to the second-order PARCOR coefficient is used, and the value determined according to the ratio is used as a parameter. It may be used. In this case, the absolute value of the parameter approaches 1 as the correlation between the primary PARCOR coefficient and the secondary PARCOR coefficient increases.

  The various processes described above are not only executed in time according to the description, but may also be executed in parallel or individually as required by the processing capability of the apparatus that executes the processes. In addition, it is needless to say that modifications can be appropriately made without departing from the gist of the present invention, such as a combination of the above-described embodiments.

  Further, when the above-described configuration is realized by a computer, processing contents of functions that each device should have are described by a program. The processing functions are realized on the computer by executing the program on the computer.

  The program describing the processing contents can be recorded on a computer-readable recording medium. As the 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.

  The 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. Furthermore, the program may be distributed by storing the program in a storage device of the server computer and transferring the program from the server computer to another computer via a network.

  A computer that executes such a program first stores, for example, a program recorded on a portable recording medium or a program transferred from a server computer in its own storage device. When executing the process, the computer reads a program stored in its own recording medium and executes a process according to the read program. As another execution form of the program, the computer may directly read the program from a portable recording medium and execute processing according to the program, and the program is transferred from the server computer to the computer. Each time, the processing according to the received program may be executed sequentially. Also, the program is not transferred from the server computer to the computer, and the above-described processing is executed by a so-called ASP (Application Service Provider) type service that realizes the processing function only by the execution instruction and result acquisition. It is good. Note that the program in this embodiment includes information that is used for processing by an electronic computer and that conforms to the program (data that is not a direct command to the computer but has a property that defines the processing of the computer).

  As an industrial application field of the present invention, for example, a lossless compression encoding / decoding technique of an acoustic signal can be exemplified. The present invention can also be applied to lossless compression encoding / decoding techniques such as video signals, biological signals, seismic wave signals, and sensor array signals in addition to audio signals.

10, 210, 310, 410, 610 Encoder 20, 220, 420, 620 Decoder

Claims (25)

(A) calculating at least a first-order PARCOR coefficient and a second-order PARCOR coefficient by performing linear prediction analysis on the input time-series signal;
(B) calculating a parameter determined according to a relational expression established between a value determined according to the first-order PARCOR coefficient calculated in step (A) and a value determined according to the second-order PARCOR coefficient; ,
(C) generating a code corresponding to information including the parameter and either the primary PARCOR coefficient or the secondary PARCOR coefficient;
The encoding method characterized by including. The encoding method of claim 1, comprising:
Steps (B) and (C) are steps executed when the absolute value of the first-order PARCOR coefficient is equal to or greater than a predetermined first threshold value, and the absolute value of the first-order PARCOR coefficient is If it is less than the first threshold, a step of generating a code corresponding to information including the primary PARCOR coefficient and the secondary PARCOR coefficient is executed, or the step (B) And (C) are steps executed when the absolute value of the second-order PARCOR coefficient is equal to or greater than a predetermined first threshold value, and the absolute value of the second-order PARCOR coefficient is less than the first threshold value. A code corresponding to information including the first order PARCOR coefficient and the second order PARCOR coefficient is generated.
An encoding method characterized by the above. The encoding method according to claim 1 or 2, comprising:
The code corresponding to the information including the parameter and either the primary PARCOR coefficient or the secondary PARCOR coefficient is:
A code corresponding to the parameter, and a code corresponding to one of the first-order PARCOR coefficient or the second-order PARCOR coefficient,
In the step (C), when the absolute value of the first encoding target is closer to a predetermined value than the absolute value of the second encoding target, the second encoding is performed on the first encoding target. A variable-length encoding method in which the frequency with which a code having a shorter code length than the target code is assigned is higher than the frequency with which a code with a longer code length is assigned to the first encoding target than the second encoding target code Generating a code corresponding to the parameter,
An encoding method characterized by the above. The encoding method according to any one of claims 1 to 3,
The relational expression is determined according to the second order PARCOR coefficient by the sum of a first multiplication value obtained by multiplying a predetermined weighting coefficient by a value determined according to the first order PARCOR coefficient and a first variable value. It is an equation that can express a fixed value, or by a sum of a second multiplication value obtained by multiplying a predetermined weighting factor by a value determined according to the second-order PARCOR coefficient and a second variable value An equation capable of expressing a value determined according to the first order PARCOR coefficient;
The parameter is a value determined according to the first variable value satisfying the relational expression, or a value determined according to the second variable value satisfying the relational expression,
An encoding method characterized by the above. The encoding method according to claim 4, comprising:
The value determined according to the first-order PARCOR coefficient is a value that increases monotonically in a broad sense with respect to at least a positive increase in the first-order PARCOR coefficient, and the value determined according to the second-order PARCOR coefficient is at least negative A value that increases monotonically in a broad sense with respect to an increase in the second-order PARCOR coefficient, or a monotonic sense in a broad sense with respect to an increase in the first-order PARCOR coefficient that has a positive value determined according to the primary PARCOR coefficient A value that decreases in accordance with the second-order PARCOR coefficient, and a value that decreases monotonically in a broad sense with respect to an increase in the second-order PARCOR coefficient that is at least negative,
The absolute value of the parameter increases monotonously in a broad sense with respect to an increase in the one-variable value or the two-variable value that satisfies the relational expression,
The weighting factor is a negative factor.
An encoding method characterized by the above. The encoding method according to claim 5, comprising:
The code corresponding to the information including the parameter and either the primary PARCOR coefficient or the secondary PARCOR coefficient is:
A code corresponding to the parameter, and a code corresponding to one of the first-order PARCOR coefficient or the second-order PARCOR coefficient,
In step (C), when the absolute value of the first encoding target is closer to 0 than the absolute value of the second encoding target, the first encoding target is shorter than the code of the second encoding target. Corresponds to the parameter by a variable length coding method in which the frequency with which a code with a code length is assigned is higher than the frequency with which a code with a code length longer than the code with the second coding target is assigned to the first coding target. Generating a code to
An encoding method characterized by the above. The encoding method according to any one of claims 4 to 6,
The value determined according to the first-order PARCOR coefficient is a first-order PARCOR coefficient, and the value determined according to the second-order PARCOR coefficient is a second-order PARCOR coefficient, and the parameter is the relational expression Whether the first variable value obtained by quantizing the first variable value to be satisfied or the second quantized variable value obtained by quantizing the second variable value,
Alternatively, the value determined according to the first-order PARCOR coefficient is a first-order quantized PARCOR coefficient obtained by quantizing the first-order PARCOR coefficient, and the value determined according to the second-order PARCOR coefficient is 2 A second-order quantized PARCOR coefficient obtained by quantizing the next PARCOR coefficient, wherein the parameter is the first variable value or the second variable value satisfying the relational expression;
An encoding method characterized by the above. (A) calculating at least a first-order PARCOR coefficient and a second-order PARCOR coefficient by performing linear prediction analysis on the input time-series signal;
(B) A first variable length encoding method as an encoding method for generating a code corresponding to the second order PARCOR coefficient when the absolute value of the first order PARCOR coefficient is equal to or greater than a predetermined threshold. And when the absolute value of the first-order PARCOR coefficient is less than the threshold, the first variable-length encoding method as an encoding method for generating a code corresponding to the second-order PARCOR coefficient; Selecting a different second variable length encoding method;
(C) Using the encoding method selected in the step (B), encoding the second-order quantized PARCOR coefficient obtained by quantizing the second-order PARCOR coefficient, and corresponding to the second-order PARCOR Generating a code;
The encoding method characterized by including. The encoding method according to claim 8, comprising:
The first variable length coding method is configured such that when the absolute value of the first encoding target is closer to the predetermined first value than the absolute value of the second encoding target, the first encoding target is the first encoding target. Encoding with a frequency that a code having a shorter code length than the code to be encoded is higher than a frequency with which a code having a longer code length than the code to be encoded is assigned to the first encoding object Is the way
When the absolute value of the first encoding target is closer to a predetermined second value than the absolute value of the second encoding target, the second variable-length encoding method may include the first encoding target as the first encoding target. Encoding with a frequency that a code having a shorter code length than the code to be encoded is higher than a frequency with which a code having a longer code length than the code to be encoded is assigned to the first encoding object Is the way
The closer the absolute value of the second-order PARCOR coefficient is to 1, the larger the second-order quantized PARCOR coefficient corresponding thereto, the first value is greater than the second value,
The closer the absolute value of the second-order PARCOR coefficient is to 1, the smaller the second-order quantized PARCOR coefficient corresponding thereto, the first value is smaller than the second value.
An encoding method characterized by the above. The encoding method according to claim 8 or 9, comprising:
Executed between the steps (A) and (C),
When the absolute value of the primary PARCOR coefficient is equal to or greater than a predetermined second threshold, a predetermined first quantization method is selected, and the absolute value of the primary PARCOR coefficient is predetermined When the second threshold value is less than the second threshold, a second quantization method that is larger than the first quantization method and has a quantization step size that is larger than the first quantization method is selected, and the second-order PARCOR coefficient is quantized by the selected quantization method. Generating the second-order quantized PARCOR coefficient,
An encoding method characterized by the above. (A) a parameter determined according to a relational expression established between a value determined according to the first-order PARCOR coefficient and a value determined according to the second-order PARCOR coefficient, and the first-order PARCOR coefficient or the second-order PARCOR Any one of the coefficients is decoded, and at least the parameter and a decoded value corresponding to the first-order PARCOR coefficient or a decoded value corresponding to the second-order PARCOR coefficient are decoded. Generating step;
(B) Using the parameter and the decoded value corresponding to the first-order PARCOR coefficient, calculate a restored value of the second-order PARCOR coefficient, or decode the value corresponding to the parameter and the second-order PARCOR coefficient Calculating a restored value of the first-order PARCOR coefficient using:
The decoding method characterized by including. The decoding method according to claim 11, comprising:
The step (B) compares the absolute value of the decoded value corresponding to the primary PARCOR coefficient with a predetermined second threshold value, and uses the parameter and the decoded value corresponding to the primary PARCOR coefficient. Determining whether to calculate a restoration value of the second-order PARCOR coefficient, or comparing the absolute value of the decoded value corresponding to the second-order PARCOR coefficient with a predetermined second threshold value. Determining whether to use the parameter and the decoded value corresponding to the second order PARCOR coefficient to calculate a restoration value of the first order PARCOR coefficient,
A decoding method characterized by the above. The decoding method according to claim 11 or 12, comprising:
The relational expression is determined according to the second order PARCOR coefficient by the sum of a first multiplication value obtained by multiplying a predetermined weighting coefficient by a value determined according to the first order PARCOR coefficient and a first variable value. It is an equation that can express a fixed value, or by a sum of a second multiplication value obtained by multiplying a predetermined weighting factor by a value determined according to the second-order PARCOR coefficient and a second variable value An equation capable of expressing a value determined according to the first order PARCOR coefficient;
The parameter is a value determined according to the first variable value satisfying the relational expression, or a value determined according to the second variable value satisfying the relational expression,
A decoding method characterized by the above. The decoding method according to claim 13, comprising:
The value determined according to the first-order PARCOR coefficient is a first-order PARCOR coefficient, and the value determined according to the second-order PARCOR coefficient is a second-order PARCOR coefficient, and the parameter is the relational expression Whether the first variable value obtained by quantizing the first variable value to be satisfied or the second quantized variable value obtained by quantizing the second variable value,
Alternatively, the value determined according to the first-order PARCOR coefficient is a first-order quantized PARCOR coefficient obtained by quantizing the first-order PARCOR coefficient, and the value determined according to the second-order PARCOR coefficient is 2 A second-order quantized PARCOR coefficient obtained by quantizing the next PARCOR coefficient, wherein the parameter is the first variable value or the second variable value satisfying the relational expression;
The step (A) decodes the code, and decodes at least the parameter and a first-order quantized PARCOR coefficient or a second-order PARCOR coefficient that is a decoded value corresponding to the first-order PARCOR coefficient. Generating a second-order quantized PARCOR coefficient that is a value;
In the step (B), a second order is obtained by the sum of a first multiplication value obtained by multiplying a predetermined weighting coefficient by an inverse quantization value of the primary quantization PARCOR coefficient and an inverse quantization value of the parameter. Or a second multiplication value obtained by multiplying a predetermined weighting coefficient by an inverse quantization value of the second-order quantization PARCOR coefficient, and an inverse quantum of the parameter. Or a first multiplication value obtained by multiplying a predetermined weighting factor by the first-order quantized PARCOR coefficient, and Or a second multiplication value obtained by multiplying a predetermined weight coefficient by the second-order quantized PARCOR coefficient. A step of calculating the restored value of the parameter and the sum of the inverse quantization to 1-order PARCOR coefficients,
A decoding method characterized by the above. (A) decoding a code corresponding to the primary PARCOR coefficient and generating a decoded value corresponding to the primary PARCOR coefficient;
(B) The absolute value of the decoded value of the code corresponding to the first order PARCOR coefficient is compared with a predetermined threshold value, and the second order by the decoding method corresponding to the predetermined first variable length coding method. A code corresponding to the PARCOR coefficient is decoded, or a code corresponding to the second PARCOR coefficient is decoded by a decoding method corresponding to a predetermined second variable length encoding method different from the first variable length encoding method. Determining whether to do;
The decoding method characterized by including. The decoding method according to claim 15, wherein
The first variable length coding method is configured such that when the absolute value of the first encoding target is closer to the predetermined first value than the absolute value of the second encoding target, the first encoding target is the first encoding target. Encoding with a frequency that a code having a shorter code length than the code to be encoded is higher than a frequency with which a code having a longer code length than the code to be encoded is assigned to the first encoding object Is the way
When the absolute value of the first encoding target is closer to a predetermined second value than the absolute value of the second encoding target, the second variable-length encoding method may include the first encoding target as the first encoding target. Encoding with a frequency that a code having a shorter code length than the code to be encoded is higher than a frequency with which a code having a longer code length than the code to be encoded is assigned to the first encoding object Is the way
The closer the absolute value of the second-order PARCOR coefficient is to 1, the larger the second-order quantized PARCOR coefficient corresponding thereto, the first value is greater than the second value,
The closer the absolute value of the second-order PARCOR coefficient is to 1, the smaller the second-order quantized PARCOR coefficient corresponding thereto, the first value is smaller than the second value.
A decoding method characterized by the above. The decoding method according to claim 15 or 16, comprising:
(C) The absolute value of the decoded value of the code corresponding to the first order PARCOR coefficient is compared with a second predetermined threshold value, and the second order PARCOR coefficient is determined using a predetermined first inverse quantization method. The decoded value obtained by decoding the code corresponding to 1 is inversely quantized, or a predetermined second inverse quantization method having a quantization step size larger than that of the first inverse quantization method is used. Determining whether to decode the decoded value obtained by decoding the code corresponding to the next PARCOR coefficient;
A decoding method characterized by the above. A linear prediction analysis unit for calculating at least a first-order PARCOR coefficient and a second-order PARCOR coefficient by performing linear prediction analysis on the input time-series signal;
A parameter calculation unit for calculating a parameter determined according to a relational expression established between a value determined according to the first-order PARCOR coefficient calculated by the linear prediction analysis unit and a value determined according to the second-order PARCOR coefficient;
An encoding unit that generates a code corresponding to information including the parameter and either the primary PARCOR coefficient or the secondary PARCOR coefficient;
An encoding device comprising: A linear prediction analysis unit for calculating at least a first-order PARCOR coefficient and a second-order PARCOR coefficient by performing linear prediction analysis on the input time-series signal;
When the absolute value of the primary PARCOR coefficient is equal to or greater than a predetermined threshold, the first variable length encoding method is selected as an encoding method for generating a code corresponding to the secondary PARCOR coefficient. When the absolute value of the primary PARCOR coefficient is less than the threshold value, a second encoding method for generating a code corresponding to the secondary PARCOR coefficient is different from the first variable length encoding method. An encoding method selection unit for selecting a variable length encoding method;
Using a coding method selected by the coding method selection unit, a second-order quantized PARCOR coefficient obtained by quantizing the second-order PARCOR coefficient is coded, and a code corresponding to the second-order PARCOR is obtained. An encoding unit to generate;
An encoding device comprising: A parameter determined according to a relational expression established between a value determined according to the first-order PARCOR coefficient and a value determined according to the second-order PARCOR coefficient; And decoding a code corresponding to the information including at least one of the parameters and a decoded value corresponding to the first-order PARCOR coefficient or a decoded value corresponding to the second-order PARCOR coefficient. And
The parameter and the decoded value corresponding to the first-order PARCOR coefficient are used to calculate a restored value of the second-order PARCOR coefficient, or the parameter and the decoded value corresponding to the second-order PARCOR coefficient are used. A PARCOR coefficient calculation unit for calculating a restoration value of the first-order PARCOR coefficient;
A decoding device comprising: A decoding unit that decodes a code corresponding to the primary PARCOR coefficient and generates a decoded value corresponding to the primary PARCOR coefficient;
The absolute value of the decoded value of the code corresponding to the first-order PARCOR coefficient is compared with a predetermined threshold value, and the second-order PARCOR coefficient is determined by a decoding method corresponding to the first variable-length encoding method. Whether to decode a corresponding code or to decode a code corresponding to the second-order PARCOR coefficient by a decoding method corresponding to a predetermined second variable-length encoding method different from the first variable-length encoding method A decoding unit for determining;
A decoding device comprising:   The program for making a computer perform each step of the encoding method of Claim 1 or 8.   The program for making a computer perform each step of the decoding method of Claim 11 or 15.   A computer-readable recording medium storing a program for causing a computer to execute each step of the encoding method according to claim 1 or 8. A computer-readable recording medium storing a program for causing a computer to execute each step of the decoding method according to claim 11 or 15.

Images