JP2008209637A - Multichannel signal encoding method, encoding device using it, and program by same method, and recording medium for same program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve encoding efficiency. <P>SOLUTION: A PARCOR coefficient and a prediction residual signal are generated from a master channel signal by a linear prediction analysis means 10. The PARCOR coefficient for a slave-channel signal is determined so that a reference value of difference of the slave-channel residual signal and the master channel residual signal may become the smallest, by the linear prediction analysis means 40 using inter-channel correlation, and the residual signal of a slave-channel is generated based on the slave-channel PARCOR coefficient. A residual difference signal is generated by performing subtracting with weight between the master channel residual signal and the slave-channel residual signal by a subtraction processing means with weight 50. By encoding means 20 and 60, the master channel residual signal, the master channel PARCOR coefficient, the slave-channel PARCOR coefficient, the residual difference signal and a weight coefficient are encoded and output. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a method for compressing and encoding a multi-channel audio signal, an apparatus therefor, a program for executing the method on a computer, and a recording medium therefor.

As shown in Non-Patent Document 1, in conventional compression coding of a multi-channel signal, for example, Levinson's signal is used so as to minimize a reference value (energy, etc.) of a linear prediction residual signal closed to each channel. Linear prediction analysis is performed by Durbin, etc., and the residual signal obtained through the filter using the linear prediction coefficients a _j (i = 0, 1,…, P), (a ₀ = 1) obtained by the analysis The weighted subtraction process was performed. The concept of the encoding device is shown in FIG. In FIG. 1, the left channel signal and the right channel signal are respectively subjected to linear prediction analysis by the linear prediction analysis units 10 and 30 to generate respective residual signals, and one of the residual signals is encoded by the encoding unit 20. The two residual signals are weighted and subtracted by the weighted subtraction processing unit 50, and the residual difference signal resulting from the subtraction is encoded by the encoding unit 60. However, it is assumed that the linear prediction coefficient here is converted into a PARCOR coefficient or quantized and includes a converted one, and is processed for each N-sample frame for each channel. For example, the left channel input signal is x ^L (n) (n = 1, 2,…, N), the residual signal is e ^L (n) (n = 1, 2,…, N), and the P ^L order The prediction coefficient is a _i ^L (i = 1, 2,…, P ^L ), the right channel input signal is x ^R (n) (n = 1, 2,…, N), and the residual signal is e ^R (n ) (n = 1, 2, ..., n), P and ^R order prediction coefficients ^{_{a i R (i = 1,}} 2, ..., when the P ^R), the residual signal of each channel,

と表すことができる。ただし、a₀ ^L=1, a₀ ^R=1とする。前記非特許文献１では、各チャネルごとに、そのチャネルの基準値（前向き及び後ろ向き予測誤差のエネルギーなど）を最小化するようにPARCOR係数を求め、そのPARCOR係数を線形予測係数a_i ^L, a_i ^Rに変換して予測残差信号を求めていた。ここで求められた基準値の小さいチャネルを親チャネル（引くチャネル）、大きいチャネルを子チャネル（引かれるチャネル）として重みつき減算処理

It can be expressed as. However, it is assumed that a ₀ ^L = 1 and a ₀ ^R = 1. In Non-Patent Document 1, for each channel, a PARCOR coefficient is obtained so as to minimize a reference value (such as forward and backward prediction error energy) of the channel, and the PARCOR coefficient is calculated as a linear prediction coefficient a _i ^L , a It is converted to _i ^R had sought prediction residual signal. Weighted subtraction processing with the channel with the smaller reference value determined here as the parent channel (subtracted channel) and the larger channel as the child channel (channel to be subtracted)

を行って重み付き差分信号を求め、これを子チャネルの残差信号の代わりに符号化対象とする。この方法によって、残差信号そのものを符号化の対象とする場合よりも、子チャネルの基準値を減少させることにより、符号量を減少させていた。重み係数γは、例えば、減算処理後のエネルギー

として求める。

To obtain a weighted difference signal, which is used as an encoding target instead of the residual signal of the child channel. With this method, the code amount is reduced by reducing the reference value of the child channel, compared to the case where the residual signal itself is the target of encoding. The weighting factor γ is, for example, energy after subtraction processing

Asking.

In order to obtain a PARCOR coefficient that minimizes the reference value of each channel, processing using a well-known lattice filter as shown in FIG. In FIG. 2, an arbitrary multi-channel input signal is represented by X (n), and a P-stage ST ₁ , ST ₂ ,..., ST _P lattice type filter that obtains coefficients up to the P order is shown. . The processing of the i-th stage ST _i (i = 1, 2,..., P) for calculating the i-th order PARCOR coefficient k _i is performed by multiplication units 6F _i−1 , 6B _i−1 and an addition unit 7F _{i−. 1} , 7B _i-1 , the sample delay unit 8 _i-1, and the PARCOR coefficient calculation unit 5 _i . However, the i = 1 stage ST _1, to calculate the coefficients k ₁ by PARCOR coefficient calculating unit 5 ₁ directly from X (n). A specific example of the PARCOR coefficient calculation process will be described below.

First-order PARCOR coefficient k ₁ calculation:
First, let the input signal X (n) itself be a forward 0th order prediction error F ₀ (n) and a backward 0th order prediction error B ₀ (n). The sample delay unit 8 _0, backward 0-order prediction error B ₀ input (n) (n = 1, 2, ..., N) is a one-sample delay and outputs the B ₀ (n-1). PARCOR coefficient calculating unit 5 _1, the inputted forward 0-order prediction error _{F 0 (n) (n =} 1, 2, ..., N) and backward zero-order prediction error _{B 0 (n-1) (} n = 1, 2,..., N) is used to calculate and output the first-order PARCOR coefficient k ₁ from Equation (6) or Equation (7) described later.

Calculation of second order PARCOR coefficient k ₂ :
Using the calculated first-order PARCOR coefficient k ₁ , forward-looking first order prediction error
F ₁ (n) = F ₀ (n) + k ₁ B ₀ (n-1)
And backward primary prediction error
B ₁ (n) = k ₁ F ₀ (n) + B ₀ (n-1)
Ask for.

Similar to the above, sample delay unit 8 ₁ In retrospective 1 input order prediction error _{B 1 (n) (n =} 1, 2, ..., N) was delayed by one sample and outputs the B ₁ (n-1) . PARCOR coefficient calculating unit 5 ₂ forward primary predicted inputted error _{F 1 (n) (n =} 1, 2, ..., N) and backward primary prediction error _{B 1 (n-1) (} n = 1, 2,..., N) and the second-order PARCOR coefficient k ₂ is output from the following equation (6) or equation (7).

The same processing as the calculation of the second-order PARCOR coefficient k ₂ is continued until the P-th order. In the final stage, the P-th order PARCOR coefficient is obtained as follows.
Calculation of Pth order PARCOR coefficient k _P :
In the sample delay unit 8 _P-1 , the input backward P-1 order prediction error B _P-1 (n) (n = 1, 2,..., N) is delayed by one sample and B _P-1 (n− 1) is output. The PARCOR coefficient calculation unit 5 _P inputs the forward P-1 order prediction error F _P-1 (n) (n = 1, 2,..., N) and the backward P-1 order prediction error B _P-1 (n -1) A P-th order PARCOR coefficient k _P is output from the following formula (6) or formula (7) using (n = 1, 2,..., N).

により、PARCOR係数k_i(i=1, 2, …, P)を求めている。このようにしてPARCOR係数k_i(i=1, 2, …, P)を求めることは格子法と呼ばれている。式(6)又は(7)を使って格子法によりPARCOR係数を求めることにより、式(6)の場合は前向き予測誤差のエネルギーと後向き予測誤差のエネルギーの相乗平均を最小化する、又は式(7)の場合は前向き予測誤差のエネルギーと後向き予測誤差のエネルギーの相加平均を最小化するようにPARCOR係数を決めることになる。 In each PARCOR coefficient calculator 5 _i (i = 1, 2,…, P), the Itakura equation

Thus, the PARCOR coefficient k _i (i = 1, 2,..., P) is obtained. Obtaining the PARCOR coefficient k _i (i = 1, 2,..., P) in this way is called a lattice method. By calculating the PARCOR coefficient by the lattice method using Equation (6) or (7), in the case of Equation (6), minimize the geometric mean of the energy of the forward prediction error and the energy of the backward prediction error, or In the case of 7), the PARCOR coefficient is determined so as to minimize the arithmetic mean of the energy of the forward prediction error and the energy of the backward prediction error.

In the prior art, at the X ^R (n) in place of X (n) in the grid system with respect to the input signal of the R channel X ^R (n), the same applies to the input signal of the L channel X ^L (n) X ( Instead of n), X ^L (n) is used to obtain the PARCOR coefficient for each channel.
A specific procedure of multi-channel signal encoding processing using the conventional technique described in Non-Patent Document 1 will be described with reference to FIG. Here, the parent channel is an R channel and the child channel is an L channel.

PARCOR coefficients of the linear prediction analyzer 11R from the input parent channel original signal x ^R (n) P ^R Next to the order of the predetermined right channel from the primary by grid method ^{_{k i R (i = 1,}} 2 ,..., P ^R ) are calculated and output as a PARCOR coefficient sequence K ^R = (k ₁ ^R , k ₂ ^R ,..., K _PR ^R ). The quantization unit 12R quantizes the PARCOR coefficient sequence K ^R and outputs a quantized PARCOR coefficient sequence ^ K ^R = (^ k ₁ ^R , ^ k ₂ ^R ,..., ^ K _PR ^R ). The conversion unit 13R converts the input quantized PARCOR coefficient sequence ^ K ^R into a linear prediction coefficient sequence ^ a _i ^R = (^ a ₁ ^R , ^ a ₂ ^R , ..., ^ a _PR ^R ) and outputs the result. . The linear prediction filter 14R uses the linear prediction coefficient sequence ^ a ^R as a filter coefficient, and the input parent channel original signal x ^R (n) (n = 1, 2,..., N)

でフィルタリングして予測残差e^R(n)を得る。ただし^a₀ ^R=1とする。残差符号化部２２Ｒは予測残差e^R(n)を例えばエントロピー符号化し残差符号C_e ^Rを出力する。係数符号化部２３Ｒは量子化済PARCOR係数系列^K^R=(^k₁ ^R, ^k₂ ^R, …, ^k_PR ^R)を例えばエントロピー符号化し係数符号C_k ^Rを出力する。符号合成部２４Ｒは残差符号C_e ^Rと係数符号C_k ^Rを合成し、親チャネル合成符号C_g ^Rを出力する。符号の合成は、単に符号の結合でよい。

To obtain a prediction residual e ^R (n). However, ^ a ₀ ^R = 1. The residual encoding unit 22R entropy encodes the prediction residual e ^R (n), for example, and outputs a residual code C _e ^R. The coefficient encoding unit 23R performs entropy encoding on the quantized PARCOR coefficient sequence ^ K ^R = (^ k ₁ ^R , ^ k ₂ ^R ,..., ^ K _PR ^R ), for example, and outputs a coefficient code C _k ^R. The code combining unit 24R combines the residual code C _e ^R and the coefficient code C _k ^R , and outputs a parent channel combined code C _g ^R. The code composition may be simply code combination.

The linear prediction analysis unit 31L uses the PARCOR coefficients k ₁ ^L , k ₂ ^L ,... From the input child channel signal x ^L (n) to the P ^L order which is the order of the left channel determined in advance by the lattice method. k _PL ^L is calculated and output as a PARCOR coefficient sequence K ^L = (k ₁ ^L , k ₂ ^L ,..., k _PL ^L ). The quantization unit 32L quantizes the PARCOR coefficient sequence K ^L and outputs a quantized PARCOR coefficient sequence ^ K ^L = (^ k ₁ ^L , ^ k ₂ ^L ,..., ^ K _PL ^L ). The conversion unit 33L converts the input quantized PARCOR coefficient sequence ^ K ^L into a linear prediction coefficient sequence ^ a ^L = (^ a ₁ ^L , ^ a ₂ ^L ,..., ^ A _PL ^L ) and outputs the result. The linear prediction filter 34L uses the linear prediction coefficient sequence ^ a ^L = (^ a ₁ ^L , ^ a ₂ ^L ,..., ^ A _PL ^L ) as filter coefficients to input the child channel signal x ^L (n) (n = 1, 2,..., N) are filtered by the following expression to obtain a prediction residual e ^L (n). However, ^ a ₀ ^L = 1.

残差符号化部６５Ｌは予測残差e^L(n)を例えばエントロピー符号化し、残差符号C_e ^Lを出力する。係数符号化部６６Ｌは量子化済PARCOR係数系列^K^L=(^k₁ ^L, ^k₂ ^L, …, ^k_PL ^L)を符号化し、係数符号C_k ^Lを出力する。符号合成部６７Ｌは残差符号C_e ^Lと係数符号C_k ^Lを合成し、子チャネル合成符号C_g ^Lを出力する。

The residual encoding unit 65L performs entropy encoding on the prediction residual e ^L (n), for example, and outputs a residual code C _e ^L. The coefficient encoding unit 66L encodes the quantized PARCOR coefficient sequence ^ K ^L = (^ k ₁ ^L , ^ k ₂ ^L ,..., ^ K _PL ^L ), and outputs a coefficient code C _k ^L. The code synthesis unit 67L synthesizes the residual code C _e ^L and the coefficient code C _k ^L , and outputs a child channel synthesis code C _g ^L.

重み量子化部５２は重み係数γを量子化し、量子化済重み係数^γを出力する。重み付き減算部５３は、予測残差e^R(n)及びe^L(n)と量子化済重み係数^γとを用いて次式(11)より、残差差分信号e^M(n)を得る。 The weight calculation unit 51 uses the prediction residual e ^R (n) and the prediction residual e ^L (n) to obtain the weight coefficient γ from the following equation (10).

The weight quantization unit 52 quantizes the weight coefficient γ and outputs a quantized weight coefficient ^ γ. The weighted subtraction unit 53 uses the prediction residuals e ^R (n) and e ^L (n) and the quantized weight coefficient ^ γ to calculate the residual difference signal e ^M (n) from the following equation (11). obtain.

残差符号化部６１Ｍは残差差分信号e^M(n)を例えばエントロピー符号化し、残差符号C_e ^Mを出力する。重み符号化部６２Ｍは量子化済重み係数^γを符号化し重み符号C_w ^Mを出力する。符号合成部６３Ｍは残差符号C_e ^Mと重み符号C_w ^Mと係数符号化部６６Ｌで求めた係数符号C_k ^Lを合成し、減算子チャネル合成符号C_g ^Mを出力する。

The residual encoding unit 61M performs, for example, entropy encoding on the residual difference signal e ^M (n), and outputs a residual code C _e ^M. The weight encoding unit 62M encodes the quantized weight coefficient ^ γ and outputs a weight code C _w ^M. The code combining unit 63M combines the residual code C _e ^M , the weight code C _w ^M, and the coefficient code C _k ^L obtained by the coefficient encoding unit 66L, and outputs a subtractor channel combined code C _g ^M.

The code amount comparison unit 68 compares the code amounts of the normal child channel composite code C _g ^L and the subtractor channel composite code C _g ^M , and outputs the smaller one together with the code representing it as a child channel code. This is the conventional method.
JP2005-115267 Yu Kamamoto, Takehiro Moriya, Takuya Nishimoto, Shigeki Hiyama, “Reversible compression coding of multi-channel signals using inter-channel correlation”, Transactions of Information Processing Society of Japan (Vol.46, No.5, pp.1118- 1128)

In the prior art, even when input is two or more channels, linear prediction analysis is performed using a PARCOR coefficient obtained so that the energy of the linear prediction residual is reduced for each channel. However, the energy of the residual difference signal shown in Equation (11), which is to be encoded in the child channel, is not minimum, and the amount of code when the residual difference signal is entropy encoded is It cannot always be reduced, and it cannot be said that efficient encoding is performed.
An object of the present invention is to provide a multi-channel signal encoding method capable of encoding more efficiently than before, an apparatus therefor, a program according to the method, and a recording medium for recording the program.

According to the present invention, a multi-channel encoding method for generating a code corresponding to a signal of a plurality of channels input for each frame composed of a plurality of samples,
A first channel linear prediction analysis step of linearly predicting and analyzing a signal of at least one channel, hereinafter referred to as a first channel, to generate a first channel PARCOR coefficient and a first channel residual signal;
A second channel PARCOR coefficient is obtained by a correlation grid method so that the residual signal approaches the first channel residual signal from a signal of at least one channel other than the first channel, hereinafter referred to as a second channel, A linear prediction analysis step using inter-channel correlation that generates a residual signal of the second channel signal as a second channel residual signal based on a second channel PARCOR coefficient;
A weighted subtraction step of generating a residual difference signal by a weighted subtraction process between the first channel residual signal and the second channel residual signal;
The first channel PARCOR coefficient and the first channel residual signal are encoded to output a first channel code, and at least the second channel PARCOR coefficient and the residual difference signal are encoded to output a second channel code. Encoding step,
It is characterized by including.

  According to the present invention, since the PARCOR coefficient of the child channel is determined so that the child channel residual signal approaches the parent channel residual signal, it is possible to make the reference value of the residual difference signal smaller than that in the prior art. It is possible to perform more efficient coding with a small code amount of the child channel.

Principle of the Invention According to the present invention, in order to reduce the amount of code required for the residual difference signal generated by weighted subtraction of the parent channel residual signal and the child channel residual signal, the total reference value (child channel, The PARCOR coefficient for determining the child channel residual signal is determined so that the sum of the energy of the forward and backward prediction errors after the subtraction process is minimized. That is, the PARCOR coefficient is calculated so that the residual signal of the child channel approaches the residual signal of the parent channel.
Here, as an example, a method will be described in which a parent channel obtains a PARCOR coefficient and a residual signal by a conventional method, and determines the PARCOR coefficient of the child channel so that the residual signal of the child channel approaches the residual signal of the parent channel. In other words, the PARCOR coefficient of the child channel is determined so as to minimize the energy per frame of the residual difference signal e ^M (n).

Correlation grid method 1
FIG. 4 shows a first method for determining the PARCOR coefficient by the lattice method in the linear prediction analysis of the child channel signal x ^L (n) so that the residual signal of the child channel resembles the residual signal e ^M (n) of the parent channel. The description will be given with reference. In this embodiment, in the conventional lattice method shown in FIG. 2, the PARCOR coefficient is calculated from the first order to the P-1 order using the conventional lattice method, and only the P order of the final stage is different from the conventional method. Calculate the PARCOR coefficient. Therefore, FIG. 4 shows only the first stage and the last stage. Here, the child channel signal x ^L (n) is given to the first stage of the lattice method processing. The final stage of the grid method processing is not in the conventional method, and the weighting factor calculator 9W and the gain section 9X are provided, PARCOR coefficient calculating unit 5 _P the use a residual signal of the parent channel e ^M (n) and that calculates the PARCOR coefficients k _P and is different from the conventional.

式(12)のカッコ内の第１項は、子チャネル信号の前向き予測誤差F_P-1(n)と親チャネル残差信号e^R(n)との相互相関を親チャネル残差信号のエネルギーで規格化した表現となっている。第２項は、子チャネル信号の後向き予測誤差B_P-1(n-1)と親チャネル残差信号e^R(n)との相互相関を親チャネル残差信号のエネルギーで規格化した表現となっている。ここでは重み係数ｇはこれら２つの規格化相互相関の相加平均で表されているが、重み付き平均としてもよい。 The weighted coefficient calculation unit 9W includes a forward P-1 order prediction error F _P-1 (n) (n = 1, 2,..., N) from the addition unit 7F _P-1 , and a sample delay unit 8 _P-1. The weight coefficient g is calculated from the following equation (12) using the backward P-1 order prediction error B _P-1 (n-1) and the parent channel residual signal e ^R (n).

The first term in parentheses in equation (12) represents the cross-correlation between the forward prediction error F _P-1 (n) of the child channel signal and the parent channel residual signal e ^R (n), and the energy of the parent channel residual signal. It has become a standardized expression. The second term is a representation in which the cross-correlation between the backward prediction error B _P-1 (n-1) of the child channel signal and the parent channel residual signal e ^R (n) is normalized by the energy of the parent channel residual signal. It has become. Here, the weight coefficient g is expressed as an arithmetic average of these two normalized cross-correlations, but may be a weighted average.

ゲイン部９Ｘは、親チャネルの残差信号e^R(n)に重み係数ｇを乗算し、親チャネル重み付き残差信号g´e^R(n)を出力する。PARCOR係数計算部５_Ｐは、加算部7F_P-1からの前向きP-1次予測誤差F_P-1(n)(n=1, 2, …, N)と、サンプル遅延部8_P-1からの後向きP-1次予測誤差B_P-1(n-1)と、重み付き親チャネル残差信号g´e^R(n)とを用いて、次式(14)によりPARCOR係数k_Pを算出する。 Or you may represent the weighting coefficient g by the geometric mean of these two normalization cross correlations like following Formula (13).

The gain unit 9X multiplies the parent channel residual signal e ^R (n) by a weighting factor g, and outputs a parent channel weighted residual signal g′e ^R (n). The PARCOR coefficient calculation unit 5 _P includes a forward P-1 order prediction error F _P-1 (n) (n = 1, 2,..., N) from the addition unit 7F _P-1 , and a sample delay unit 8 _P-1 Using the backward P-1 order prediction error B _P-1 (n-1) and the weighted parent channel residual signal g´e ^R (n), the PARCOR coefficient k _P is calculate.

ここで、式(7)に示したBurg法の変形として相加平均での求め方を示したが、式(6)に示した相乗平均によるItakura法の変形として次式(15)によって計算してもよい。

で示すように前向き予測誤差F_P-1(n)と後向き予測誤差B_P-1(n-1)の重み比率αと(1-α)（ただし１≧α≧０）を変えるなど、他の基準を最小化するように求めてもよい。いずれにしても、親チャネル残差信号e^R(n)に対し子チャネルの残差信号が似るように子チャネルの線形予測フィルタに対する予測係数を決めればよい。

Here, an arithmetic mean was obtained as a modification of the Burg method shown in Equation (7), but it was calculated by the following equation (15) as a variant of the Itakura method based on the geometric mean shown in Equation (6). May be.

The weight ratio α and (1-α) (where 1 ≧ α ≧ 0) of the forward prediction error F _P-1 (n) and the backward prediction error B _P-1 (n-1) are changed as shown in FIG. You may ask to minimize the criterion. In any case, the prediction coefficient for the linear prediction filter of the child channel may be determined so that the residual signal of the child channel resembles the parent channel residual signal e ^R (n).

As described above, the weighting factor g expressed by the equation (12) or (13) is the cross-correlation between the parent channel residual signal e ^R (n) and the forward prediction error F _P-1 (n), and the parent channel residual. This invention is represented by the cross-correlation between the difference signal e ^R (n) and the backward prediction error B _P-1 (n-1), and is therefore represented by the formula (14), (15) or (16) The grid method for calculating the PARCOR coefficient according to is based on the cross-correlation between the parent channel residual signal and the forward and backward prediction errors. The PARCOR coefficient calculation processing method according to the present invention is called the correlation grid method. To.

Modification 1 of the correlation grid method 1
In the correlation grid method 1, only the P-order PARCOR coefficient k _P which is the final stage of processing by the grid method is calculated by a method different from the conventional method. PARCOR coefficients of all orders of the calculation of the PARCOR coefficients from the primary implemented in the preceding stage to P next k _1, k _2, ..., may be determined k _P. Alternatively, for any stage, for example, 11 ≦ P, the same method as the P order of the correlation grid method 1 is used for calculating the PARCOR coefficient from the first order to the tenth order, and from the 11th order to the P order, the conventional method is used. The PARCOR coefficient may be calculated. In short, if the calculation of one or more arbitrary order PARCOR coefficients is performed in the same manner as the P order of the correlation grid method 1, the minimum effect of the present invention can be obtained.

Modification 2 of the correlation grid method 1
In the first modification of the correlation grid method 1, forward and backward prediction error by weighting factor calculator 9W _i-1 similar to FIG. 4 in any stage (i-th stage), as shown in FIG. 5 F _i-1 (n ), B _i-1 (n-1) and the parent channel residual signal e ^R (n), the weight coefficient g is obtained by the equation (12), and the residual signal e is obtained by the gain unit 9X _i-1 using the weight coefficient g. ^R (n) is weighted, and this weighted residual signal g′e ^R (n) is used by the PARCOR coefficient calculation unit 5 _i to replace i with P in the formula (14), (15) or (16). The PARCOR coefficient k _i is obtained as follows. Further, the polarity inversion unit 9Y _i is provided in the prediction error calculation unit EE including the multiplication units 6F _i and 6B _i and the addition units 7F _i and 7B _i shown by the dotted frame in FIG. 5, and the weighted residual signal g′e ^R By reversing the polarity of g′e ^R (n) with respect to (n) by the polarity reversing unit 9Y _i and giving it to the adding unit 7B _i , the forward i-th order prediction error F _i (n) and the backward i used in the next stage Next prediction error B _i (n) is obtained. That is, the forward i-th order prediction error F _i (n) and the backward i-th order prediction error B _i (n) are obtained by the following equations.
F _i (n) = F _i-1 (n) + k _{i '} B _i-1 (n-1)-g'e ^R (n) (17)
B _i (n) = k _{i '} F _i (n) + B _i-1 (n-1)-g'e ^R (n) (18)

Correlation grid method 2
In this correlation grid method 2, the PARCOR coefficient is calculated using the conventional grid method from the first order to the P-1 order as in the correlation grid method 1 in FIG. Calculate the PARCOR coefficient. Hereinafter, only parts different from the correlation grid method 1 of FIG. 4 will be described with reference to FIG. This correlation grid method 2, the correlation grid method 1 of Figure 4, omitting the weight calculator 9W and the gain unit 9X, giving the parent channel residual signal e ^R (n) is directly in PARCOR coefficient calculating unit 5 _P PARCOR and to calculate the coefficient k _P.

The PARCOR coefficient calculation unit 5 _P includes a forward P-1 order prediction error F _P-1 (n) (n = 1, 2,..., N) from the addition unit 7F _P-1 , and a sample delay unit 8 _P-1 From the backward P-1 order prediction error B _P-1 (n-1) (n = 1, 2,…, N) and the parent channel residual signal e ^R (n) by the following equation (19) to calculate the coefficient k _P.

Comparing the above equation (19) and equation (14), in equation (19), the forward prediction error F _P-1 (n) and the backward prediction error B _P-1 (n-1) in equation (14) The weighted subtraction of the parent channel residual signal e ^R (n) is not performed for each, and in the numerator, the cross-correlation between the forward prediction error F _P-1 (n) and the parent channel residual signal e ^R (n) The product of the cross-correlation between the backward prediction error B _P-1 (n-1) and the parent channel residual signal e ^R (n) is normalized by the parent channel residual signal energy and subtracted. In the denominator, the cross-correlation between the forward prediction error F _P-1 (n) and the parent channel residual signal e ^R (n), the backward prediction error B _P-1 (n-1) and the parent channel residual The sum of squares of the cross-correlation between the signals e ^R (n) is normalized by the parent channel residual signal energy and subtracted.

Difference term including forward prediction error F _P-1 in the denominator of equation (19) to (n) in common, forward prediction error F _P-1 (n) as the parent channel residual signal e in the denominator of formula (14) Corresponding to the weighted subtraction between ^R (n), the difference in terms that commonly include the backward prediction error B _P-1 (n-1) in the denominator of Equation (19) is the backward prediction in the denominator of Equation (14). This corresponds to weighted subtraction between the error B _P-1 (n-1) and the parent channel residual signal e ^R (n). In Equation (19), the PARCOR coefficient k _P can also be obtained by weighted averaging the terms that commonly include the denominator F _P-1 (n) and the terms that commonly include B _P-1 (n-1). Good.

Modification 1 of the correlation grid method 2
In the correlation grid method 2 shown in FIG. 6, only the P-order PARCOR coefficient k _P which is the final stage of the processing by the grid method is calculated by a method different from the conventional method, but the same method as the P-order of the correlation grid method 2 is used. The PARCOR coefficients may be calculated in all stages to obtain the PARCOR coefficients k ₁ , k ₂ ,..., K _P for all orders from the first order to the P order. Alternatively, for any stage, for example, 11 ≦ P, the same method as the P order of the correlation grid method 1 is used for calculating the PARCOR coefficient from the first order to the tenth order, and from the 11th order to the P order, the conventional method is used. The PARCOR coefficient may be calculated. In short, if the calculation of one or more arbitrary order PARCOR coefficients is performed in the same manner as the P order of the correlation grid method 1, the minimum effect of the present invention can be obtained.

Modification 2 of the correlation grid method 2
Similar to the modified example 2 of the correlated grid method 1 shown in FIG. 5, in the modified example 1 of the correlated grid method 2, after obtaining the PARCOR coefficient k _i at an arbitrary stage, the prediction error calculation indicated by the dotted frame in FIG. The configuration of the part EE may be adopted, and the forward i-th order prediction error F _i (n) and the backward i-th order prediction error B _i (n) used in the next stage may be obtained by equations (17) and (18).

In each of the above-described correlation grid methods and variations thereof, the backward prediction error B _i (n−1) at each processing stage ST _i is delayed at each stage leading to that processing stage. Therefore, the parent channel residual signal e ^R (n) related to the calculation with the backward prediction error B _i (n−1) may be calculated in accordance with the delay amount with a delay.

に置き換えてもよい。また式(14), (15), (16)において、B_P-1(n-1)-g×e^R(n)をB_P-1(n-1)-g×e^R(n-P)と置き換えてもよい。更に、式(18)におけるg×e^R(n)をg×e^R(n-i)に置き換えてもよい。 For example, in equations (12), (13), (19), instead of ΣB _P-1 (n-1) × e ^R (n)

May be replaced. In the equations (14), (15), (16), B _P-1 (n-1) -g × e ^R (n) is replaced with B _P-1 (n-1) -g × e ^R (nP) May be replaced. Furthermore, g × e ^R (n) in equation (18) may be replaced with g × e ^R (ni).

The encoding of the multi-channel signal encoding basic configuration and processing procedure 7 are multi-channel signal according to the present invention, it illustrates the basic functional structure of a coding apparatus when applied to left and right channel signals, the procedure Is shown in FIG. In the present invention, when a multi-channel input signal is encoded, it may be determined in advance which channel signal is input as a parent channel signal or a child channel signal, or is determined by this encoding apparatus. (See Non-Patent Document 1 for details on how to determine parent and child). For example, it is assumed here that the right channel signal is input as a parent channel signal and the left channel signal is input as a child channel signal.
The multi-channel signal encoding apparatus includes a parent channel linear prediction analysis unit 10, a parent channel encoding unit 20, a linear prediction analysis unit 40 using inter-channel correlation, a weighted subtraction processing unit 50, a difference The encoding means 60 is comprised.

First, in the linear prediction analysis means 10, the PARCOR coefficient sequence K ^R is obtained by the conventional lattice method shown in FIG. 2 using the parent channel signal x ^R (n) as an input signal, and further based on the PARCOR coefficient sequence K e ^R (n) is obtained by, for example, equation (2) (step S1). The parent channel residual signal e ^R (n) thus obtained is subjected to a PARCOR coefficient sequence K ^M for the child channel so that the child channel residual signal approaches by the linear prediction analysis means 40 using the inter-channel correlation. Determine (step S2). Specifically, this is to determine the PARCOR series by the correlation grid method according to the present invention described with reference to FIGS. 4 to 6, whereby the parent channel residual signal and the forward and backward prediction errors of the child channel signal are determined. The energy difference between them is made small.

The linear prediction analysis means 40 using the inter-channel correlation further generates a residual signal e ^M (n) of the child channel signal using the obtained child channel PARCOR coefficient sequence K ^M (step S3). Then, weighted subtraction processing unit 50, generates a residual difference signal ~ e ^M (n) by weighted subtraction of the parent channel residual signal e ^R (n) and the child channel residual signal e ^M (n) (Step S4). Next, the parent channel encoding unit 20 encodes the parent channel PARCOR coefficient sequence K ^R and the parent channel residual signal e ^R (n), and the differential encoding unit 60 encodes the child channel PARCOR coefficient sequence K ^M , The residual difference signal ~ e ^M (n) and the weight coefficient γ are each encoded (step S5).

Example 1
FIG. 9 shows a block diagram of an encoding apparatus when the present invention is applied to encoding of a stereo signal. 3 is different from the configuration of the prior art in FIG. 3 in that a linear prediction analysis unit 41M using inter-channel correlation is provided instead of the child channel linear prediction analysis unit 31L in FIG. 3, and a residual encoding unit 65L, a code synthesis unit 67L and the code amount comparison unit 68 are not provided. Also, the quantization unit 32L, the conversion unit 33L, the linear prediction filter 34L, and the coefficient encoding unit 66L in FIG. 3 are replaced with the same configuration units 42M to 44M and 64M in FIG. 9, respectively.

  The configuration 10 including the linear prediction analysis unit 11R, the quantization unit 12R, the conversion unit 13R, and the linear prediction filter 14R in FIG. 9 corresponds to the first channel linear prediction analysis unit in claim 9. The configuration 40 including the linear prediction analysis unit 41M, the quantization unit 42M, the conversion unit 43M, and the linear prediction filter 44M using the inter-channel correlation in FIG. 9 corresponds to the linear prediction analysis unit using the inter-channel correlation in claim 9. To do. The configuration 50 including the weight calculation unit 51, the weight quantization unit 52, and the weighted subtraction unit 53 in FIG. 9 corresponds to the weighted subtraction means in claim 9. The configuration 20 including the residual encoding unit 22R, the coefficient encoding unit 23R, and the code synthesis unit 24R in FIG. 9 corresponds to the first channel encoding unit in claim 9. The configuration 60 including the residual encoding unit 61M, the weight encoding unit 62M, the code synthesis unit 63M, and the coefficient encoding unit 64 in FIG. 9 corresponds to the second channel encoding unit in claim 9. These correspondences also apply to the following embodiments.

The right channel signal of a frame (N samples) is x ^R (n) (n = 1, 2,…, N), and the left channel signal is x ^L (n) (n = 1, 2,…, N). And Here, the right channel is a parent channel and the left channel is a child channel.
The linear prediction analysis unit 11R generates PARCOR coefficients k _i ^R (i = 1, 2,..., P ^R ) from the input parent channel signal x ^R (n) by a conventional lattice method. PARCOR coefficient quantizer 12R is input ^{_{k i R (i = 1,}} 2, ..., P R) is quantized and quantized PARCOR coefficient ^{_{^ k i R (i = 1}} , 2, ..., P R) Is output. Conversion unit 13R is inputted quantized PARCOR coefficient ^{_{^ k i R (i = 1}} , 2, ..., P R) the quantized prediction coefficients ^{_{^ a i R (i = 1}} , 2, ..., P R) Convert to The linear prediction filter 14R uses the quantized prediction coefficient ^ a _i ^R (i = 1, 2,..., P ^R ) as a filter coefficient, and the input parent channel signal x ^R (n) according to the above equation (8). Filter to obtain the prediction residual e ^R (n). However, ^ a ₀ ^R = 1.

The residual encoding unit 22R encodes the prediction residual e ^R (n) and outputs a residual code C _e ^R. The coefficient encoding unit 23R encodes the quantized PARCOR coefficient ^ k _i ^R (i = 1, 2,..., P ^R ) and outputs a coefficient code C _k ^R. The code combining unit 24R combines the residual code C _e ^R and the coefficient code C _k ^R , and outputs a parent channel combined code C _g ^R.
The linear prediction analysis unit 41M using the inter-channel correlation uses the parent channel residual signal e ^R (n) for the child channel signal x ^L (n), and the correlation grid according to the present invention described with reference to FIGS. PARCOR coefficient by any law to P ^M following ^{_{k i M (i = 1,}} 2, ..., P M) is calculated.

The quantization unit 42M quantizes the input PARCOR coefficient k _i ^M (i = 1, 2,..., P ^M ), and the quantized PARCOR coefficient ^ k _i ^M (i = 1, 2,..., P ^M ). Is output. The transform unit 43M converts the input quantized PARCOR coefficient ^ k _i ^M (i = 1, 2,..., P ^M ) into a quantized prediction coefficient ^ a _i ^M (i = 1, 2,..., P ^M ). Convert to The linear prediction filter 44M uses the quantized prediction coefficient ^ a _i ^M (i = 1, 2,..., P ^M ) as a filter coefficient, and filters the input child channel signal x ^L (n) by the following expression to perform prediction. Obtain the residual e ^M (n). However, ^ a ₀ ^M = 1.

重み計算部５１は親チャネルの予測残差e^R(n)と相互相関を考慮した予測残差e^M(n)を用いて以下の式から重み係数γを求める。

重み量子化部５２は重み係数γを量子化し、量子化済重み係数^γを得る。重み付き減算部５３は、残差信号e^R(n)、e^M(n)と量子化済重み係数^γを用いて次式より、残差差分信号~e^M(n)を得る。

The weight calculation unit 51 obtains a weighting coefficient γ from the following equation using the prediction residual e ^R (n) of the parent channel and the prediction residual e ^M (n) considering the cross correlation.

The weight quantization unit 52 quantizes the weight coefficient γ to obtain a quantized weight coefficient ^ γ. The weighted subtraction unit 53 uses the residual signals e ^R (n) and e ^M (n) and the quantized weight coefficient ^ γ to obtain a residual difference signal ~ e ^M (n) from the following equation.

The residual encoding unit 61M encodes the residual difference signal ~ e ^M (n) and outputs a residual code C _e ^M. The coefficient encoding unit 64M encodes the quantized PARCOR coefficient ^ k _i ^M (i = 1, 2,..., P ^M ) and outputs a coefficient code C _k ^M. The weight encoding unit 62M encodes the quantized weight coefficient ^ γ and outputs a weight code C _w ^M. The code combining unit 63M combines the residual code C _e ^M , the weight code C _w ^M, and the coefficient code C _k ^M , and outputs a child channel combined code C _g ^M.

As described above, in the present invention, when the PARCOR coefficients from the child channel signal x ^L (n) to the P ^L order are obtained by the linear prediction analysis unit 41M using cross-correlation, the parent channel remaining in any at least one stage. The PARCOR coefficient is determined so that the energy of the weighted difference between the forward prediction error and the backward prediction error of the difference signal e ^R (n) and the child channel signal x ^L (n) is reduced. The signal resembles the parent channel residual signal, and the amount of code required for encoding the residual difference signal generated by the weighted subtraction according to Equation (22) is reduced.

として残差差分信号を求めてもよいし、時間差（サンプル数間隔τ）を考慮した複数タップの重み付き減算処理を、

として残差差分信号を求めてもよい。 Example 2
In the prior art of FIG. 3 described above, as the processing of the weighted subtraction unit 53, for example, performing a multi-tap weighted subtraction process or a multi-tap weighted subtraction process considering time difference is known (patent). Reference 1). Also in the present invention, a weighted subtraction process of a plurality of taps (j = -1, 0, 1) is used instead of the equation (22).

The residual difference signal may be obtained as a multi-tap weighted subtraction process considering the time difference (sample number interval τ),

A residual difference signal may be obtained as follows.

  As described above, according to the weighted subtraction process with a plurality of taps, the child channel residual signal can be controlled to be closer to the parent channel residual signal, so that the code amount can be reduced accordingly. For example, even when the sound source position of the left and right channel signals is shifted from the center to one side, the weight at the tap position according to the arrival time difference from the sound source to the two microphones can be controlled, so that the code compression rate can be reduced. Can be improved.

Example 3
In the example of FIG. 9, the case where the parent-child relationship has been determined has been described as an example, but each prediction residual may be obtained once and the analysis with the smaller energy as the parent channel may be performed. An example thereof is shown in FIG. In this embodiment, a child channel linear prediction analysis unit including a linear prediction analysis unit 31L, a quantization unit 32L, a conversion unit 33L, and a linear prediction filter 34L similar to FIG. 3 is added to the configuration of FIG. The corresponding configuration corresponds to the second channel second linear prediction analysis means in claim 10), and the same processing is performed on the child channel signal to obtain the residual signal e ^L (n). Further, a comparison unit 45L and an input switching unit 2 are provided. The energy of the residual signal e ^R (n) from the linear prediction filter 14R on the parent channel side and the energy of the child channel residual signal e ^L (n) are represented by the following equations, for example.

により計算し、小さい方のチャネルの入力信号を親チャネル信号、大きい方のチャネルの入力信号を子チャネル信号と決定し、それに従って入力切替部２を切り替え制御する。以下の処理は図９と同様である。なお、比較部４５Ｌによる比較は、残差信号のエネルギーの比較を行う場合を示したが、絶対値の和や符号量の比較を用いてもよい。

The input signal of the smaller channel is determined as the parent channel signal, and the input signal of the larger channel is determined as the child channel signal, and the input switching unit 2 is switched and controlled accordingly. The subsequent processing is the same as in FIG. Note that the comparison by the comparison unit 45L shows a case where energy of residual signals is compared, but a sum of absolute values or a comparison of code amounts may be used.

Example 4
The code amount obtained in the embodiment of FIG. 9 and the right channel signal x ^R (n) in FIG. 9 as the input of the child channel and the left channel signal x ^L (n) as the input of the parent channel (that is, the parent-child relationship) It is also possible to perform the encoding again as shown in FIG. 11 and to output the smaller one compared with the code amount in the case of FIG. For example, as shown in FIG. 12, the encoding device shown in the configuration of FIG. 9 (or FIG. 11) is the encoding unit 3 of this modified example, and the input switching unit 2 is provided on the input side of the encoding unit 3. The selection output unit 4 is provided on the output side. The selection output unit 4 includes storage units 4A and 4B that hold a set of codes from the code synthesis units 24R and 63M in FIG. 9 (or FIG. 11) constituting the encoding unit 3, and the storage units 4A and 4B. A code amount comparison unit 4C that calculates the code amount of the retained code set and determines which is smaller, and a selection unit 4D that selectively outputs the code set that is determined to be smaller are provided. .

First, the input switching unit 2 inputs the right channel signal as a parent channel signal and the left channel signal as a child channel signal to the encoding unit 3, and performs an encoding process as shown in FIG. The output codes C _g ^R and C _g ^M are held in the storage unit 4A, for example.
Next, the input switching unit 2 is switched to input the right channel signal as a parent channel signal and the right channel signal as a child channel signal to the encoding unit 3 and perform encoding processing (L and R are reversed as shown in FIG. 11). Process). M used for a symbol representing a signal or a sign in FIG. 9 is changed to M2 in the process of FIG. Further, the symbol γ in FIG. 9 is changed to γ ₂ . Output codes C _g ^L and C _g ^{M2 obtained} by the encoding process of FIG. 11 are held in the storage unit 4B.

The code amount of the code C _g ^R + C _g ^M held in the storage unit 4A and the code C _g ^L + C _g ^M2 held in the storage unit 4B are respectively calculated by the code amount comparison unit 4C. select the better the selection output section 4D small, a set of the selected code, any channel and outputs the information C ^C indicating whether a parent channel (or child channels).
According to this method, encoding can be performed more efficiently than when the parent channel and child channel are determined by comparing the energy of the right channel residual signal and the energy of the left channel residual signal.

Example 5
For the child channel, C _g ^L obtained by combining the coefficient code C _k ^L obtained by performing the normal linear prediction analysis and the residual code C _e ^L, and C _g ^M obtained in the embodiment of FIG. The code amount may be compared by the code amount comparison unit and the smaller one may be output. A modified embodiment is shown in FIG. This modified embodiment is similar to the embodiment of FIG. 9 in that the same linear prediction analysis unit 31L, quantization unit 32L, conversion unit 33L, linear prediction filter 34L, residual encoding unit 42L, coefficient encoding unit as in FIG. 65L, a code synthesizing unit 66L, and a code amount comparing unit 67 are added, and the same processing as in FIG. 3 is performed.
The configuration including the linear prediction analysis unit 31L, the quantization unit 32L, the conversion unit 33L, and the linear prediction filter 34L in FIG. 13 corresponds to the second channel linear prediction analysis unit in claim 12. The configuration including the residual encoding unit 65L, the coefficient encoding unit 66L, and the code synthesis unit 67L in FIG. 13 corresponds to the second channel second encoding unit in claim 12.
Combined code C _g ^M from composite code C _g ^L and the code combining unit 63M from the code combining unit 67L is given to the code amount comparator 71 is compared their code amount, and select the composite code of the smaller It outputs as the code | symbol of a child channel with the information showing which was selected. According to this embodiment, when the code amount is smaller when the subtraction process is not performed, the result of the normal linear prediction analysis is used, so that the compression rate is not always deteriorated as compared with the conventional method.

Example 6
In principle, the absolute value of PARCOR coefficient k _i (i = 1, 2,, P) calculated by Eqs. (14), (15), (16), (19) based on the correlation grid method is greater than 1. In an actual calculation process, the absolute value of the coefficient k _i may sometimes be larger than 1 due to some cause (for example, rounding down or rounding up in a numerical calculation by a computer). In the embodiment of FIG. 9, if there is a PARCOR coefficient k _i ^M generated by the linear prediction analysis unit 41M using the inter-channel correlation having an absolute value of 1 or more, the linear prediction obtained by converting those coefficients. There is a problem that the operation of the linear prediction filter 44M with the coefficient a _i ^M becomes unstable. Therefore, in the modified embodiment of FIG. 13, instead of comparing the code amounts of the composite codes C _g ^L and C _g ^{M by} the code amount comparison unit 68 and outputting the smaller one, a linear prediction analysis unit using inter-channel correlation Which one to select may be determined depending on the PARCOR coefficient obtained at 41M. An example is shown in FIG. In the modified embodiment of FIG. 14, a code selection unit 69 is provided instead of the code amount comparison unit 68 in the modified embodiment of FIG.

  The configuration including the linear prediction analysis unit 31L, the quantization unit 32L, the conversion unit 33L, and the linear prediction filter 34L in FIG. 14 corresponds to the second channel linear prediction analysis unit in claim 13. The configuration including the residual encoding unit 65L, the coefficient encoding unit 66L, and the code synthesis unit 67L in FIG. 14 corresponds to the second channel second encoding unit in claim 13.

Code selection unit 69 PARCOR coefficient P ^M pieces obtained by the linear prediction analysis unit 41M using inter-channel correlation ^{_{k i M (i = 1,}} 2, ..., P M) of either at least one coefficient When the absolute value is greater than or equal to a threshold (for example, 1), the same composite code C _g ^L as the conventional one is output, and otherwise, the composite code C _g ^M is output as the child channel code. When the former is selected, the quantization unit 42M, the conversion unit 43M, the linear prediction filter 44M, the residual encoding unit 61M, the weight encoding unit 62M, the code synthesis unit 63M, the coefficient encoding unit 64M, and the like need to be performed. Since there is no, the amount of processing can be reduced.

Each of the first to sixth embodiments described above shows the case of a two-channel signal. However, when there are more channels than two channels, as shown in Non-Patent Document 1, for example, the energy or absolute value of the residual signal It is only necessary to determine a pair that reduces the sum of the two and perform the above-described encoding for each pair. In that case, one or a plurality of channels may be overlapped with another channel to form a plurality of pairs, or one or a plurality of channels may be encoded independently. In the case of lossless coding, a code representing the parent channel is also output. However, when the input is two channels, the parent-child relationship is explicit depending on the presence / absence of the weighting coefficient code, and therefore the code representing the parent channel may be omitted.
The encoding method according to each embodiment of the present invention described above may be implemented as a computer-executable program. Alternatively, the program may be recorded on a readable recording medium, and the program read from the recording medium may be executed by a computer.

  FIG. 15 compares the average compression rates for various types of sound source files by the conventional method and the method according to FIG. 7 of the present invention. The sound source was a stereo sound source of 30 seconds, and 48 kHz / 16 bit (15 files), 48 kHz / 24 bit (15 files), 96 kHz / 24 bit (15 files), and 192 kHz / 24 bit (6 files) were used. As shown in the figure, it is shown that the compression ratio (data amount after encoding / data amount before encoding) is smaller in the encoding according to the present invention in any of the prediction orders 10, 30, and 50. Has been.

The block diagram which shows the concept of the conventional multichannel encoding. The figure for demonstrating the determination method of the PARCOR coefficient by the conventional lattice method. The block diagram which shows the example of the conventional multichannel encoding apparatus. The figure for demonstrating the determination method of the PARCOR coefficient by the correlation grid method by this invention. The figure for demonstrating the modification of a correlation lattice method. The figure for demonstrating the other example of a correlation lattice method. 1 is a basic functional configuration block diagram of a multi-channel encoding device according to the present invention. FIG. The flowchart which shows the basic process sequence of the multichannel encoding method by this invention. 1 is a block diagram showing a first embodiment of a multi-channel encoding device according to the present invention. The block diagram which shows Example 2 by this invention. The block diagram which shows the process which replaced the left-right channel signal in FIG. FIG. 12 is a block diagram showing a third embodiment according to FIG. 9 and FIG. 11. The block diagram which shows Example 4 by this invention. The block diagram which shows Example 5 by this invention. The graph for showing the effect of this invention.

Claims (17)

A multi-channel encoding method for generating a code corresponding to a signal of a plurality of channels input for each frame composed of a plurality of samples,
A first channel linear prediction analysis step of linearly predicting and analyzing a signal of at least one channel, hereinafter referred to as a first channel, to generate a first channel PARCOR coefficient and a first channel residual signal;
A second channel PARCOR coefficient is obtained by a correlation grid method so that the residual signal approaches the first channel residual signal from a signal of at least one channel other than the first channel, hereinafter referred to as a second channel, A linear prediction analysis step using inter-channel correlation that generates a residual signal of the second channel signal as a second channel residual signal based on a second channel PARCOR coefficient;
A weighted subtraction step of generating a residual difference signal by a weighted subtraction process between the first channel residual signal and the second channel residual signal;
The first channel PARCOR coefficient and the first channel residual signal are encoded to output a first channel code, and at least the second channel PARCOR coefficient and the residual difference signal are encoded to output a second channel code. Encoding step,
And a multi-channel signal encoding method.   2. The multi-channel signal encoding method according to claim 1, wherein a second channel second linear prediction analysis step of generating a second channel second residual signal by performing a linear prediction analysis on the second channel signal, and the first channel. A comparison control step of comparing a reference value of the residual signal with a reference value of the second channel second residual signal, and controlling switching of the first channel signal and the second channel signal based on the comparison result A multi-channel signal encoding method comprising:   The multi-channel signal encoding method according to claim 1, wherein the total code amount of the first channel code and the second channel code when the first channel signal and the second channel signal are not interchanged with each other A multi-channel signal encoding method further comprising a selection output step of comparing the total code amount, selecting and outputting the smaller one of the total code amounts, and outputting information indicating which one is selected.   2. The multi-channel signal encoding method according to claim 1, wherein the second channel signal is subjected to linear prediction analysis to generate a second channel second PARCOR coefficient and a second channel second residual signal. A second channel second encoding step for encoding the second channel second residual signal and the second channel second PARCOR coefficient, and outputting a second channel second code, the second channel code, and A multi-channel signal encoding method, further comprising: a code amount comparison step of comparing the code amount of the second channel second code and outputting the smaller code amount as the second channel code.   2. The multi-channel signal encoding method according to claim 1, wherein the second channel signal is subjected to linear prediction analysis to generate a second channel second PARCOR coefficient and a second channel second residual signal. A second channel second encoding step of encoding the second channel second residual signal and the second channel second PARCOR coefficient to output a second channel second code, and the second channel PARCOR coefficient A multi-channel signal encoding method, further comprising: a code selection step of selectively outputting the second channel second code or the second channel code depending on whether or not at least one absolute value is equal to or greater than a predetermined value. .   6. The multi-channel signal encoding method according to claim 1, wherein the step of obtaining the second channel PARCOR coefficient includes the second channel signal in at least one stage of a lattice method using a plurality of stages of processing. Determining a weighting factor based on a cross-correlation between the forward prediction error of the first channel residual signal and the cross-correlation between the backward prediction error of the second channel signal and the first channel residual signal; The PARCOR coefficient of the stage based on the forward prediction error and the backward prediction error updated by weighting and subtracting the first channel residual signal with the weighting coefficient for the forward prediction error and the backward prediction error of the second channel signal, respectively. A multi-channel signal encoding method comprising the step of:   7. The multi-channel signal encoding method according to claim 6, wherein the step of determining the weighting factor includes the cross-correlation between the forward prediction error of the second channel signal and the first channel residual signal, the backward prediction error, and the A multi-channel signal encoding method comprising calculating based on an average of cross-correlations between first channel residual signals.   6. The multi-channel signal encoding method according to claim 1, wherein the step of obtaining the second channel PARCOR coefficient is based on a cross correlation between a forward prediction error and a backward prediction error of the second channel signal. A first value obtained by subtracting a normalized value of a product of a prediction error and a cross-correlation between the first channel residual signal and a cross-correlation between the backward prediction error and the first channel residual signal. A value is obtained, and the cross-correlation between the forward prediction error and the first channel residual signal, the backward prediction error, and the first channel residual signal is calculated from the sum of the forward prediction error energy and the backward prediction error energy. Obtaining a second value obtained by subtracting a normalized value of the square sum of the cross-correlation of the cross-correlation, and calculating the PARCOR coefficient based on a ratio between the first value and the second value With features Multichannel signal coding method that. A multi-channel encoding device that generates a code corresponding to a signal of a plurality of channels input for each frame composed of a plurality of samples,
First channel linear prediction analysis means for generating a first channel PARCOR coefficient and a first channel residual signal by performing a linear prediction analysis of a first channel signal of at least one channel, hereinafter referred to as a first channel;
A second channel PARCOR coefficient is obtained from a signal of at least one channel other than the first channel, hereinafter referred to as a second channel, so that the residual signal resembles the first channel residual signal by a correlation lattice method, Linear prediction analysis means using inter-channel correlation for generating a residual signal of the second channel signal as a second channel residual signal based on a second channel PARCOR coefficient;
Weighted subtraction means for generating a residual difference signal by weighted subtraction processing between the first channel residual signal and the second channel residual signal;
First channel encoding means for encoding the first channel PARCOR coefficient and the first channel residual signal and outputting a first channel code;
Second channel encoding means for encoding at least the second channel PARCOR coefficient and the residual difference signal and outputting a second channel code;
A multi-channel signal encoding device.   10. The multi-channel signal encoding apparatus according to claim 9, wherein second channel second linear prediction analysis means for generating a second channel second residual signal by performing linear prediction analysis on the second channel signal, and the first channel. A switching means capable of switching the signal and the second channel signal, a reference value of the first channel residual signal and a reference value of the second channel second residual signal are compared, and the switching is performed based on the comparison result. A multi-channel signal encoding apparatus, further comprising comparison control means for controlling the means.   10. The multi-channel signal encoding apparatus according to claim 9, wherein the input switching means capable of switching the first channel signal and the second channel signal, and the input switching means include the first channel signal and the second channel signal. The total code amount of the first channel code and the second channel code when the code is not replaced is compared with the total code amount when the code is replaced, and the smaller total code amount is selected and output, and which one is selected A multi-channel signal encoding apparatus, further comprising selection output means for outputting information to be expressed.   10. The multi-channel signal encoding apparatus according to claim 9, wherein the second channel signal is subjected to linear prediction analysis to generate a second channel second PARCOR coefficient and a second channel second residual signal. Means, a second channel second encoding means for encoding the second channel second residual signal and the second channel second PARCOR coefficient, and outputting a second channel second code, the second channel code and the above A multi-channel signal encoding apparatus, further comprising: a code amount comparing means for comparing the code amount of the second channel second code and outputting the smaller code amount as the second channel code.   10. The multi-channel signal encoding apparatus according to claim 9, wherein said second channel signal is subjected to linear prediction analysis to generate a second channel second PARCOR coefficient and a second channel second residual signal. And second channel second encoding means for encoding the second channel second residual signal and the second channel second PARCOR coefficient and outputting a second channel second code, and second channel PARCOR coefficient A multi-channel signal encoding apparatus, further comprising code selection means for selectively outputting the second channel second code or the second channel code depending on whether or not at least one absolute value is equal to or greater than a predetermined value. .   The multi-channel signal encoding apparatus according to any one of claims 9 to 13, wherein the linear prediction analysis means using the inter-channel correlation includes a plurality of processing stages based on a lattice method, and in any at least one processing stage, A weighting factor is determined based on the cross-correlation between the forward prediction error of the second channel signal and the first channel residual signal, and the cross-correlation between the backward prediction error of the second channel signal and the first channel residual signal. A weighting factor calculating means for determining, the first channel residual signal is weighted and subtracted by the weighting factor with respect to the forward prediction error and the backward prediction error of the second channel signal, respectively, and based on the obtained difference signal, And a PARCOR coefficient calculating means for calculating a PARCOR coefficient.   15. The multi-channel signal encoding apparatus according to claim 14, wherein the weighting factor calculation means uses the weighting factor as a correlation between the forward prediction error of the second channel signal and the first channel residual signal, and the backward prediction. A multi-channel signal encoding apparatus that calculates based on an average of an error and a correlation between the first channel residual signals. The program which performs the method in any one of Claims 1 thru | or 8 with a computer. A computer-readable recording medium on which the program according to claim 16 is recorded.

Images