JP4348324B2 - Signal encoding apparatus, method, program, and recording medium - Google Patents

Description

  The present invention relates to a signal encoding apparatus, method, program, and recording medium for encoding a plurality of sample values.

  FIG. 1 shows a functional configuration example of a conventional integer signal encoding unit. FIG. 2 shows a processing flow (S840) of the integer signal encoding unit 840. The integer signal encoding unit 840 includes an interval dividing unit 8401, a linear prediction analysis unit 8402, a linear prediction coefficient encoding unit 8403, a linear prediction coefficient decoding unit 8404, an inverse filter 8407, a sample buffer 8408, and a residual signal encoding unit 8409. , And an integration unit (Multiplexer) 8410. The section dividing unit 8401 assembles a plurality of samples of the input integer signal into a frame and sets it as a digital sampling value in units of frames (S8401). The linear prediction analysis unit 8402 performs analysis for linear prediction on the integer signal input in the form of a frame (hereinafter referred to as “input integer signal”), and outputs a linear prediction coefficient (S8402). Here, let P be the order of the linear prediction coefficient. The linear prediction coefficient encoding unit 8403 encodes the linear prediction coefficient according to the result of the linear prediction analysis unit 8402, and outputs a prediction coefficient code (S8403). The linear prediction coefficient decoding unit 8404 decodes the output from the linear prediction coefficient encoding unit 8403 and outputs a P-th order linear prediction coefficient (S8404). In this example, the output from the linear prediction coefficient encoding unit 8403 is decoded by the linear prediction coefficient decoding unit 8404 to obtain a quantized linear prediction coefficient. However, the linear prediction coefficient decoding unit 8404 may be omitted, and a prediction coefficient code and a corresponding quantized linear prediction coefficient may be obtained from the linear prediction coefficient encoding unit 8403. The inverse filter 8407 uses the P-order quantized linear prediction coefficient output from the linear prediction coefficient decoding unit 8404 and the sample value of the previous frame stored in the sample buffer 8408 as the signal transmitted with the prediction coefficient code. And using the sample value of the current frame. Further, the signal transmitted with the prediction coefficient code restored from the input integer signal is subtracted, and a residual signal is output (S8407). Further, at least the last P sample of the sample value of the current frame is held in the sample buffer 8408 (S8408). The residual signal encoding unit 8409 encodes the residual signal output from the inverse filter 8407 and outputs a residual code (S8409). The integration unit (Multiplexer) 8410 integrates the prediction coefficient code output from the linear prediction coefficient encoding unit 8403 and the residual code output from the residual signal encoding unit 8409, and outputs the result as an integer signal code (S8410). Note that the linear prediction analysis unit 8402 may also use the last P sample of the immediately preceding frame for the linear prediction analysis. In this case, as indicated by a dotted line in FIG. 1, the value of the last P sample of the immediately preceding frame is received from the sampling buffer 8408.

  FIG. 3 shows a functional configuration example of a conventional integer signal decoding unit. FIG. 4 shows a processing flow (S920) of the integer signal decoding unit 920. The integer signal decoding unit 920 includes a division unit (Demultiplexer) 9201, a linear prediction coefficient decoding unit 9202, a residual signal decoding unit 9203, a sample buffer 9206, and a synthesis filter 9207. The encoded data is received and accumulated by a demultiplexer 9201, and is divided into a prediction coefficient code and a residual code (S9201). The linear prediction coefficient decoding unit 9202 decodes the prediction coefficient code and outputs a linear prediction coefficient (S9202). The residual signal decoding unit 9203 decodes the residual code and outputs a residual signal (S9203). The synthesis filter 9207 synthesizes the signal using the linear prediction coefficient output from the linear prediction coefficient decoding unit 9202 and the sample value of the previous frame and the sample value of the current frame held in the sample buffer 9206 (S9206). Further, the restored signal and the residual signal are added to obtain an integer signal (S9207).

As an encoding method for losslessly encoding an integer signal, for example, as described in Non-Patent Document 1, there is a method for performing linear prediction and losslessly encoding a linear prediction coefficient and a linear prediction residual. In the encoding method of Non-Patent Document 1, a linear prediction coefficient is obtained for each frame in which a plurality of input data sample value sequences are put together, the linear prediction coefficient is encoded, and is quantized in the encoding process. An inverse filter (also referred to as an analysis filter) is configured using the linear prediction coefficient thus obtained, a linear prediction residual signal is obtained, and the linear prediction residual signal is encoded.
Tilman Liebchen and Yuriy A. Reznik, "MPEG4-ALS: an Emerging Standard for Lossless Audio Coding," Proceedings of the Data Compression Conference (DCC'04), pp1068-0314 / 04, 2004.

Problems of the conventional method will be described with reference to FIG. FIG. 5 is an integer signal encoding device that encodes 24 bits, and shows an image when one frame is composed of 10 samples. The shaded portions in the figure are 0 or 1 bits (bits having substantially information), and the other portions are 0 (bits having substantially no information). In addition, a range surrounded by a dotted line in the figure is an encoding target. FIG. 5 shows that all bits are to be encoded.
However, in the case of a signal originally obtained by amplifying an input signal having a small amplitude, the lower bit may be 0. For example, in the frame 1 of FIG. 5, all the bits are 0 in the lower 2 bits of the digit, and in the frame 2 in the lower 4 bits of the digit. In the prior art, such lower bits having substantially no information are also encoded. That is, in the case of an input signal in which the lower bits are 0, it cannot be said that the compression efficiency is necessarily good.

In the present invention, when an integer signal is encoded, the number of digits with all 0s in the frame is from the least significant digit, or the number of digits with one bit is 1 from the least significant digit. Is detected and the information is output. In addition, a higher-order digit including a digit in which any bit in the lowest digit is 1 is encoded, and a code string is output. FIG. 6 shows an image of encoding when the present invention is applied. In the case of frame 1, all the lower two digits are 0 in the case of frame 2 and in the case of frame 2.
Also, when decoding an integer signal, the integer signal is decoded from the code string, and the number of all zero bits in the frame is from the least significant digit, or one of the bits is 1 Using the information on the number of the least significant digit from the least significant digit, an integer signal to be output is obtained by adding a digit having all 0 bits to the lower order of the decoded integer signal.

  In this way, when the encoding target range is different for each frame, there is a new problem that inter-frame prediction is not properly performed and compression efficiency is deteriorated in the case of linear predictive encoding for performing inter-frame prediction. Therefore, in the present invention, when the detection result of the immediately preceding frame is different from the detection result of the current frame, frequent changes in the shift amount are suppressed in consideration of the past detection results. Alternatively, linear predictive coding is performed in consideration of the previous shift amount.

  According to the present invention, since the lower bits having substantially no information are not encoded when the integer signal is encoded, the compression efficiency is increased. Also, frequent changes in the shift amount are eliminated, and the compression rate in the case of compression encoding using inter-frame prediction can be improved. Furthermore, in the case of linear predictive coding that performs inter-frame prediction, even if the position of a bit with information is shifted between the previous and subsequent frames, the shift amount can be corrected, so that the compression efficiency is maintained high. it can.

Below, in order to avoid duplication of description, the same number is given to the structural part which has the same function, and the process step which performs the same process, and description is abbreviate | omitted.
[First Embodiment]
FIG. 7 shows a functional configuration of the encoding processing unit of the present invention. FIG. 8 shows a processing flow of the encoding processing unit 400. The encoding processing unit 400 includes a frame buffer 810, a shift amount calculation unit 420, an integer signal shift processing unit 430, an integer signal encoding unit 840, and an integration unit (Multiplexer) 460. The frame buffer 810 accumulates the integer signal (input signal) and outputs it to the shift amount calculation unit 420 and the integer signal shift processing unit 430 in a plurality of sample value units (frame units) (S810). The shift amount calculation unit 420 determines the shift amount so that all the higher-order digits including the lowest-order digit in which any bit is 1 are to be encoded (S420). A specific determination method will be described later. The integer signal shift processing unit 430 shifts the encoding range according to the determination of the shift amount calculation unit 420 (S430). The integer signal encoding unit 840 encodes the integer signal shifted in step S430 by the method described with reference to FIGS. 1 and 2 as a conventional technique (S840). The integration unit (Multiplexer) 460 integrates and outputs the encoded integer signal and the shift amount (S460).

FIG. 9 shows details of the shift amount determination step (S420) of the shift amount calculation unit 420. The shift amount calculation unit 420 takes in as many samples as the number of samples (N _F ) constituting the frame (S4201). The shift amount ΔE is initialized to 0 (S4202). When the input signal sequence is shifted to the right by ΔE, the number n of samples in which the bit in the least significant bit plane (the set of bits of the same digit of the samples in the frame) becomes 1 is detected (S4203). Whether n = 0 is checked. If Yes, the process proceeds to step S4205. If No, step S420 is terminated (S4204). In step 4205, the shift amount ΔE is set to ΔE + 1 (S4205). It is checked whether ΔE is equal to the number Q of encoded bits. If Yes, the process proceeds to step S4207. If No, the process returns to step S4203. In step S4207, ΔE is returned to 0, and step S420 is ended. In this step, since all the bits in the frame are 0, the shift amount ΔE is set to 0.

FIG. 10 shows a functional configuration of the decoding processing unit of the present invention, and FIG. 11 shows a processing flow of the decoding processing unit 700. The decoding processing unit 700 includes a demultiplexer 910, an integer signal decoding unit 920, and an inverse shift processing unit 950. The division unit (Demultiplexer) 910 receives the encoded data, divides it into encoded integer signals and shift amounts, and acquires digit information (shift amount) shifted on the encoding device side (S910). . The integer signal decoding unit 920 decodes the integer signal by the method described with reference to FIGS. 3 and 4 as a conventional technique (S920). The inverse shift processing unit 950 shifts the decoded integer signal in the direction opposite to the shift amount to obtain an output signal (S950).
As described above, according to the present invention, when the integer signal is encoded, the lower bits having substantially no information are not encoded, so that the compression efficiency is increased.

[Second Embodiment]
This embodiment is a combination of the method of making the current shift amount the same as the immediately preceding shift amount and the first embodiment when the difference in shift amount from the immediately preceding frame is within a predetermined range. .

FIG. 12 shows a functional configuration example of the encoding processing unit of this embodiment. The encoding processing unit 100 includes a frame buffer 810, a shift amount candidate calculation unit 120, a shift amount selection unit 130, and a frame shift amount storage buffer 140, a shift amount determination unit 110, an integer signal shift processing unit 430, an integer signal code 840 and an integration unit (Multiplexer) 460. A difference from the encoding processing unit 400 in FIG. 7 is a shift amount determination unit 110.
The processing flow of the encoding processing unit 100 is the processing flow of FIG. 8, in which step S420 is replaced with step S110 (FIG. 13). FIG. 13 shows a processing flow (step S110) of the shift amount determination unit 110. In step S110, first, the shift amount candidate calculation unit 120 obtains the shift amount candidate ΔE from the least significant digit to the most significant digit where all the bits of the sample value in the frame are 0 (S120). . The shift amount selection unit 130 determines whether the current frame is the first frame or a random access frame (RA frame: a frame that does not use prediction from a past frame) (S140). In the case of the first frame or the random access frame, the shift amount selecting unit 130 sets the shift amount candidate ΔE as the shift amount of the current frame (S150). If it is neither the first frame nor the random access frame, the shift amount selecting unit 130 reads the shift amount S _j−k (k is an integer of 1 or more) of one or more past frames from the frame shift amount holding buffer. The shift amount S _j of the current frame is determined using the shift amount of the past frame and the shift amount candidate ΔE (S130).

FIG. 14 shows a detailed process flow example of the process (step S130) of the shift amount selecting unit 130. The shift amount selection unit 130 reads the shift amount S _j−1 of the immediately preceding frame from the frame shift amount holding buffer 140 and the shift amount candidate ΔE from the shift amount candidate calculation unit 120 (S1301). S _j-1 > ΔE is confirmed (S1302). In the case of No, S _j-1 + α <ΔE is confirmed (S1303). Here, α is a predetermined threshold value. A step S1302 and S1303 are both in the case of No, the shift amount _{S j-1} of the previous frame and the shift amount _{S j} of the current frame (S1304). If either of steps S1302 and S1303 is Yes, the shift amount candidate ΔE is set as the shift amount S _j of the current frame (S1305).

α is a threshold for changing the shift amount only when the shift amount fluctuates above a certain level, and is set to 3 in advance, for example. When α = 3, the shift amount compensation ΔE obtained by analyzing the maximum amplitude of the frame is smaller than the shift amount S _{j−1 of} the previous frame or is smaller than S _j−1 +3. This corresponds to changing the shift amount only when the value becomes large.
By determining the frame shift amount S _j in this way, frequent shift amount changes are eliminated, and the compression rate in the case of compression encoding using inter-frame prediction can be improved.

[Modification 1]
In the second embodiment, the shift amount selection unit 130 of the shift amount determination unit 110 determines a threshold value in advance as shown in FIG. 14, and the difference between the shift amount of the immediately preceding frame and the shift amount candidate of the current frame. If is within the threshold, the shift amount of the current frame is made the same as the previous frame. In this modification, the shift amount selection unit 130 of the shift amount determination unit 110 calculates the data amount after encoding at each shift amount from the shift amount of the previous frame to the shift amount candidate of the current frame, A shift amount with a small amount of data is set as a shift amount of the current frame.

FIG. 15 shows a processing flow (step S130 ′) of the shift amount selecting unit 130, which is an alternative to step S130. The shift amount selection unit 130 reads the shift amount S _j−1 of the immediately preceding frame from the frame shift amount holding buffer 140 and the shift amount candidate ΔE from the shift amount candidate calculation unit 120 (S1301). S _j-1 > ΔE is confirmed (S1302). When step S1302 is No, _{Dmin is set} to infinity, and i is set to the shift amount _Sj-1 of the immediately preceding frame (S1311). However, infinity may be greater than or equal to the maximum value that can be taken as the code amount. It obtains a code amount of the code amount and the error signal of the integral signal when the shift amount was i, obtaining the code amount D _i of the coded data when integrated (S1312). D _min is to confirm whether greater than _{D i} (S1313). When D _min is larger than D _i , D _{i is set} to D _min and i is set to i _min (S1314). If D _min is less than _{D i,} the process proceeds to step S1315. It is confirmed that i <ΔE (S1315). If step S1315 is Yes, i + 1 is substituted for i (S1316). If step S1315 is No, the shift amount S _{j is set} to i _min (S1317). If step S1302 is Yes, the shift amount _Sj is set as a shift amount candidate ΔE (S1305).
If processing is performed in this way, processing time is required, but a shift amount with a small code amount can be selected with certainty.

[Modification 2]
In this modification, the shift amount selection unit 130 of the shift amount determination unit 110 records the shift amounts of the past N frames (N is an integer of 2 or more). If the shift amount candidate is less than or equal to the nth (n is an integer less than or equal to 1 and less than N) shift amount in the past N frame shift amounts and greater than or equal to the shift amount of the immediately preceding frame, Let the shift amount of the immediately preceding frame be the shift amount of the current frame. When the shift amount candidate is larger than the nth (n is an integer greater than or equal to 1 and less than N) shift amount in the past N frame shift amounts, or smaller than the shift amount of the previous frame Uses the shift amount candidate as the shift amount of the current frame.

FIG. 16 shows a processing flow (step S130 ″) of the shift amount selecting unit 130 instead of step S130. The shift amount selecting unit 130 shifts the past frame shift amount S _j−k ( k = 1 to N), the shift amount candidate ΔE is read from the shift amount candidate calculation unit 120 (S1321), where N is an integer greater than or equal to 2. The threshold α is the N past shift amounts. The shift amount is set to the nth smallest amount (S1322), S _j-1 > ΔE is confirmed (S1302), and if <No, α <ΔE is confirmed (S1323). to, the shift amount _{S j-1} of the previous frame and the shift amount _{S j} of the current frame (S1304). Furthermore, one of the steps S1302 and S1323 is In the case of Yes, the shift amount candidate ΔE is set as the shift amount S _j of the current frame (S1305).

  In the present embodiment, the threshold is not determined in advance, but is obtained from the past shift amount. Therefore, the threshold value can be changed in consideration of the characteristics of the input signal.

[Third Embodiment]
In 1st Embodiment and 2nd Embodiment, it changed by shifting the range of encoding object. This is because, in integer signal encoding processing, encoding is usually performed by ignoring 0 that is higher than the highest 1 of each sample, so even if a digit of 0 is added to the higher order, integer signal encoding is performed. This is a process assuming that the compression efficiency of the process does not deteriorate.

As another method, when all the bits of N digits (N is a positive integer) including the least significant digit of the sample in the frame are 0, there is a method of deleting the N digit including the least significant digit. According to this method, for example, in the integer signal encoding process, even when encoding is performed including 0 that is higher than the highest one, the compression efficiency of the integer signal encoding unit can be improved.
FIG. 17 is a diagram illustrating a functional configuration example of an encoding processing unit 300 including a deletion amount calculation unit 320 that calculates a deletion amount instead of determining a shift amount, and an integer signal deletion processing unit 330 that performs a deletion process instead of a shift process. It is.

In the present embodiment, the deletion amount is obtained, but the method of obtaining the deletion amount may be the same as the method of obtaining the shift amount in the first embodiment or the second embodiment. Instead of simply shifting, all the bits with zeros are deleted.
In addition, since the deletion amount is received instead of the shift amount on the decoding processing unit side, instead of performing the reverse shift processing, only the processing for adding all the digits of 0 is changed.

[Modification]
FIG. 18 shows a modification of the third embodiment. The deletion processing is modified not to be performed by the integer signal deletion processing unit 330 but to be performed by the integer signal encoding unit 340.

[Fourth Embodiment]
The integer signal encoding unit 840 of the first embodiment is the conventional technique shown in FIG. In the present embodiment, an integer signal encoding unit that takes into account the shift amount of the immediately preceding frame is used. FIG. 19 shows a functional configuration of the encoding processing unit of the present embodiment. The difference from the encoding processing unit 400 shown in FIG. 7 is an integer signal encoding unit 240.

  FIG. 20 shows a functional configuration example of the integer signal encoding unit of the present embodiment. The integer signal encoding unit 240 includes an interval dividing unit 8401, a linear prediction analysis unit 8402, a linear prediction coefficient encoding unit 8403, a linear prediction coefficient decoding unit 8404, an interframe correction processing unit 2405, a shift amount buffer 2406, and an inverse filter 8407. , A sample buffer 2408, a residual signal encoding unit 8409, and an integration unit (Multiplexer) 8410. The difference from the integer signal encoding unit 840 in FIG. 1 is that an inter-frame correction processing unit 2405 and a shift amount buffer 2406 are added to correct a difference in shift amount between frames, and a sample value can be shifted. The sample buffer 2408 is used. Note that the linear prediction analysis unit 8402 may also use the last P sample of the immediately preceding frame for the linear prediction analysis. In this case, as indicated by a dotted line in FIG. 8, the value of the last P sample of the immediately preceding frame in which the shift amount described later is matched with the shift amount of the current frame is received from the sampling buffer 2408.

FIG. 21 shows a processing flow of the integer signal encoding unit 240. The shift amount buffer 2406 and the sample buffer 2408 are initialized in advance (the state in which there is no previous frame information). The section dividing unit 8401 further finely divides the sequence of digital sampling values in units of frames of the input integer signal into subframes (S8401). However, when the subframe is not used, the section dividing unit 8401 is not necessary. In the following, it is also expressed as framing including sub-framing. The linear prediction analysis unit 8402 performs an analysis for linear prediction on the framed input integer signal, and outputs _P linear prediction coefficients (a _{1 to} a _P ) (S8402). Here, let P be the order of the linear prediction coefficient. The linear prediction coefficient encoding unit 8403 encodes the linear prediction coefficient according to the result of the linear prediction analysis unit 8402, and outputs a prediction coefficient code (S8403). The linear prediction coefficient decoding unit 8404 decodes the output from the linear prediction coefficient encoding unit 8403, and outputs P-th order quantized linear prediction coefficients (a ₁ ＾ to a _P ）) (S8404). In this example, the output from the linear prediction coefficient encoding unit 8403 is decoded by the linear prediction coefficient decoding unit 8404 to obtain a quantized linear prediction coefficient. However, the linear prediction coefficient decoding unit 8404 may be omitted, and a prediction coefficient code and a corresponding quantized linear prediction coefficient may be obtained from the linear prediction coefficient encoding unit 8403. The inter-frame correction processing unit 2405 receives the shift amount S _j of the current frame from the shift amount calculation unit 820 (S24051). The inter-frame correction processing unit 2405 records the shift amount S _j of the current frame in the shift amount buffer 2406, and reads the shift amount S _j−1 of the immediately preceding frame from the shift amount buffer 2406 (S2406). The inter-frame correction processing unit 2405 calculates the shift amount difference S _j −S _j−1 and sets the last P samples of the immediately preceding frame held by the sample buffer 2408 to the right by S _j −S _j−1. Alternatively, it is shifted to the left (corrected) (S24052). Whether to shift right or left depends on whether the right shift is defined as the positive direction or the left shift is defined as the positive direction in the shift amount calculation method. With this correction, even when the shift amount of the immediately preceding frame is different from the shift amount of the current frame, the values of the last P samples of the immediately preceding frame used for linear prediction of the first sample of the current frame (y _{−1 to} y _−P ) is a sample value (y ′ _{−1 to} y ′ _−P ) having the same shift amount as the current frame. If the current frame is the first frame or a random access frame (RA frame: a frame that does not use prediction from a past frame), neither the shift amount of the immediately preceding frame nor the sample value exists. As a corresponding method, a method of substituting 0 as the value of the last P samples (y _{−1 to} y _−P ) of the immediately preceding frame in initialization, or a shift in the case of the first frame or random access frame There is a method not to change the amount. However, it is not necessary to limit to these.

The inverse filter 8407 holds at least the last P sample of the sample value of the current frame in the sample buffer 2408. Further, the last P sample values of the immediately preceding frame are read from the sample buffer 2408 (S2408). The inverse filter 8407 reads the signal transmitted with the prediction coefficient code from the P-order quantized linear prediction coefficients (a ₁ ˜a _P ) output from the linear prediction coefficient decoding unit 8404 and the sample buffer 2408. Calculation is performed using the last P sample values of the immediately preceding frame and the sample value of the current frame. Specifically, the predicted value of the i-th sample of the current frame of the signal is the quantized linear prediction coefficient (a ₁ 〜 to a _P ）) of the current frame and the sample value (y ′ ₋₁ ) of the immediately preceding frame. ˜y′− _P ) and a sample value (y _{1 to} y _i−1 ) of the current frame are calculated as follows.

逆フィルタ８４０７は、さらに、入力整数信号ｙ_ｉから復元した予測係数符号で伝達される信号を減算し、残差信号ｒ_ｉを出力する（Ｓ８４０７）。したがって、残差信号ｒ_ｉは、次式のようになる。

Inverse filter 8407 is further adapted to subtract the signals transmitted by the prediction coefficient code that has been restored from the input integer signal _{y i,} and outputs a residual signal _{r i} (S8407). Therefore, the residual signal r _i is given by

残差信号符号化部８４０９は、逆フィルタ８４０７から出力された残差信号ｒ_ｉを符号化し、残差符号を出力する（Ｓ８４０９）。統合部（Multiplexer）８４１０は、線形予測係数符号化部８４０３が出力した予測係数符号と残差信号符号化部８４０９が出力した残差符号とを統合して整数信号符号として出力する（Ｓ８４１０）。
図２２に、本実施形態の復号化処理部の機能構成例を示す。図１０に示した復号化処理部７００との違いは、整数信号復号化部６２０である。

Residual signal encoding unit 8409 is a residual signal _{r i} which is outputted from the inverse filter 8407 encodes and outputs the residual code (S8409). The integration unit (Multiplexer) 8410 integrates the prediction coefficient code output from the linear prediction coefficient encoding unit 8403 and the residual code output from the residual signal encoding unit 8409, and outputs the result as an integer signal code (S8410).
FIG. 22 shows a functional configuration example of the decoding processing unit of the present embodiment. The difference from the decoding processing unit 700 shown in FIG. 10 is an integer signal decoding unit 620.

  FIG. 23 shows a functional configuration example of the integer signal decoding unit of the present embodiment. FIG. 24 shows a processing flow of the integer signal decoding unit 620. The integer signal decoding unit 620 includes a demultiplexer 9201, a linear prediction coefficient decoding unit 9202, a residual signal decoding unit 9203, an interframe correction processing unit 6204, a shift amount buffer 6205, a sample buffer 6206, and a synthesis filter 9207. Consists of The difference from the conventional integer signal decoding unit 920 in FIG. 3 is that an inter-frame correction processing unit 6204 and a shift amount buffer 6205 are added, and that the sample buffer 6206 can change the shift amount of the sample value.

In the integer signal decoding unit 620, the shift amount buffer 6205 and the sample buffer 6206 are initialized in a state where there is no previous frame information. The demultiplexer 9201 receives and accumulates the encoded data, and divides it into a prediction coefficient code and a residual code (S9201). The linear prediction coefficient decoding unit 9202 decodes the prediction coefficient code and outputs _P linear prediction coefficients (a _{1 to} a _P ) (S9202). Residual signal decoding unit 9203 decodes the residual code, and outputs a residual signal _{r i} (S9203). On the other hand, the interframe correction processing unit 6204 receives the shift amount S _j of the current frame from the demultiplexer 9201 or other communication means (S62041). The inter-frame correction processing unit 6204 stores the shift amount of the current frame in the shift amount buffer 6205 and reads the shift amount S _j−1 of the immediately preceding frame from the shift amount buffer 6205 (S6205). The inter-frame correction processing unit 6204 calculates the shift amount difference S _j −S _j−1 and the last P sample values (y _{−1 to} y _−P ) of the immediately preceding frame held in the sample buffer 6202. ) Is shifted (corrected) to the right or left by S _j −S _j−1 (S62042). Whether to shift right or left depends on whether the right shift is defined as a positive direction or the left shift is defined as a positive direction in the shift amount calculation method as described above. With this correction, even when the shift amount of the immediately preceding frame is different from the shift amount of the current frame, the values of the last P samples of the previous frame used for linear prediction of the first sample of the current frame (y _{−1 to} y _{− P} ) is a sample value (y'- _{1 to} y'- _P ) having the same shift amount as the current frame. If the current frame is the first frame or a random access frame, neither the shift amount of the immediately preceding frame nor the sample value exists. As a corresponding method, a method of substituting 0 as the value of the last P samples (y _{−1 to} y _−P ) of the immediately preceding frame in initialization, or a shift in the case of the first frame or random access frame There is a method not to change the amount. However, it is not necessary to limit to these.

The synthesis filter 9207 holds at least the last P sample of the sample value of the current frame in the sample buffer 6206. Further, the last P sample values of the immediately preceding frame are read from the sample buffer 6206 (S6206). The synthesis filter 9207 stores the quantized linear prediction coefficients (a ₁ 〜 to a _P ＾) output from the linear prediction coefficient decoding unit 9202, held in the sample buffer 9206, and the previous frame corrected by the interframe correction processing unit 6204. Is used to synthesize an integer signal y _i using the following sample equation (y ′ _{−1 to} y ′ _−P ), the current frame sample value (y _{1 to} y _i−1 ), and the residual signal r _i (S9207).

このように、直前のフレームのシフト量と符号化対象のフレームのシフト量とを考慮して線形予測符号化のフレーム間予測を行うので、効率的に符号化でき、符号量を少なくすることができる。
なお、上記の実施形態はコンピュータに、上記方法の各ステップを実行させるプログラムを読み込ませ、実施することもできる。また、コンピュータに読み込ませる方法としては、プログラムをコンピュータ読み取り可能な記録媒体に記録しておき、記録媒体からコンピュータに読み込ませる方法、サーバ等に記録されたプログラムを電気通信回線等を通じてコンピュータに読み込ませる方法などがある。

In this way, since inter-frame prediction of linear predictive coding is performed in consideration of the shift amount of the immediately preceding frame and the shift amount of the encoding target frame, it is possible to efficiently encode and reduce the code amount. it can.
In addition, said embodiment can also read and implement the program which makes a computer perform each step of the said method. Also, as a method for reading into the computer, the program is recorded on a computer-readable recording medium, and the program is read from the recording medium into the computer, or the program recorded in the server or the like is read into the computer through an electric communication line or the like. There are methods.

The figure which shows the function structural example of the conventional integer signal encoding part. The figure which shows the processing flow of the conventional integer signal encoding part. The figure which shows the function structural example of the conventional integer signal decoding part. The figure which shows the processing flow of the conventional integer signal decoding part. The figure explaining the problem of the conventional method. The figure which shows the image of an encoding at the time of applying this invention. The figure which shows the function structure of the encoding process part of 1st Embodiment. The figure which shows the processing flow of the encoding process part. The figure which shows the detail of the shift amount determination step (S420) of the shift amount calculation part 420. The figure which shows the function structure of the decoding process part of 1st Embodiment. The figure which shows the processing flow of the decoding process part. The figure which shows the function structural example of the encoding process part of 2nd Embodiment. The figure which shows the processing flow (step S110) of the shift amount determination part 110. FIG. The figure which shows the detail of the process step (S130) of the shift amount selection part 130. FIG. The figure which shows the processing flow (S130 ') of the shift amount selection part 130 used instead of step S130. The figure which shows the processing flow (S130 ") of the shift amount selection part 130 used instead of step S130. The figure which shows the function structural example of the encoding process part of 3rd Embodiment. The figure which shows the function structural example of the encoding process part of the modification of 3rd Embodiment. The figure which shows the function structure of the encoding process part of 4th Embodiment. The figure which shows the function structural example of the integer signal encoding part of 4th Embodiment. The figure which shows the processing flow of the integer signal encoding part 240. FIG. The figure which shows the function structural example of the decoding process part of 4th Embodiment. The figure which shows the function structural example of the integer signal decoding part of 4th Embodiment. The figure which shows the processing flow of the integer signal decoding part 620.

Claims (12)

The digital signal samples represented in the input plurality of bits, each frame constituted by a plurality of samples, a signal encoding apparatus for encoding,
A frame buffer for forming a plurality of the digital signal frames of the or stopped samples,
The value of the bit position is 0 in all the digital signal samples in a frame, the range of the least significant bit of n bit (n is an integer of 1 or more), the determined, a shift amount candidate calculating unit that n is a shift amount candidate ,
A frame shift amount holding buffer for recording a shift amount of at least one past frame;
A shift amount selection unit that determines a shift amount of the current frame according to a predetermined criterion using the shift amount candidate and the shift amount recorded in the frame shift amount holding buffer;
A shift unit that shifts all digital signal samples in the frame according to a shift amount determined by the shift amount selection unit;
The portion excluding the all-digital signal samples the least significant bit or brush shift amount of the bit range of le in the frame encoding, and integer signal encoding unit to obtain a code sequence,
A signal encoding device comprising: an output unit that outputs information on a shift amount obtained by the shift amount selection unit and a code string obtained by the integer signal encoding unit. The signal encoding device according to claim 1 ,
When the relationship between the shift amount candidate and one shift amount recorded in the frame shift amount holding buffer is within a predetermined range, one shift amount recorded in the frame shift amount holding buffer is set. As the shift amount of the current frame,
When the relationship between the shift amount candidate and one shift amount recorded in the frame shift amount holding buffer is outside a predetermined range, the shift amount candidate is set as the shift amount of the current frame. A signal encoding device comprising the shift amount selection unit. The signal encoding device according to claim 1 ,
When the shift amount candidate is smaller than the shift amount of the immediately preceding frame recorded in the frame shift amount holding buffer, or when the difference between the shift amount candidate and the shift amount of the immediately preceding frame is larger than a predetermined threshold value Is the shift amount candidate of the current frame as the shift amount candidate,
If the shift amount candidate is equal to or greater than the shift amount of the immediately preceding frame and the difference between the shift amount candidate and the shift amount of the immediately preceding frame is equal to or less than a predetermined threshold, the shift amount of the immediately preceding frame is set to the current frame. A signal encoding device comprising the shift amount selecting unit. The signal encoding device according to claim 1 ,
The shift amount of the immediately preceding frame is recorded in the frame shift amount holding buffer,
If the shift amount candidate is greater than or equal to the shift amount of the previous frame, calculate the data amount after encoding for each shift amount from the shift amount of the previous frame to the shift amount candidate, and the data amount is the largest. The smaller shift amount is used as the current frame shift amount.
The signal encoding device including the shift amount selection unit, wherein the shift amount candidate is set as the shift amount of the current frame when the shift amount candidate is smaller than the shift amount of the immediately preceding frame. The signal encoding device according to claim 1 ,
The shift amount of the past N frames (N is an integer of 2 or more) is recorded in the frame shift amount holding buffer,
In the case where the shift amount candidate is nth (n is an integer greater than or equal to 1 and less than N) and less than or equal to the shift amount of the previous frame, among the shift amounts of the past N frames, Let the shift amount of the previous frame be the shift amount of the current frame,
When the shift amount candidate is larger than the nth (n is an integer greater than or equal to 1 and less than N) shift amount smaller than the shift amount of the previous N frames, or smaller than the shift amount of the immediately preceding frame. Includes a shift amount selection unit, wherein the shift amount candidate is the shift amount of the current frame. The digital signal samples represented in the input plurality of bits, each frame constituted by a plurality of samples, a signal encoding method for encoding,
A framing step of forming a plurality of the digital signal frames of the or stopped samples,
The value of the bit position is 0 in all the digital signal samples in a frame, the range of the least significant bit of n bit (n is an integer of 1 or more), the determined, a shift amount candidate calculating step of the n shift amount candidate ,
A shift amount selection step of determining a shift amount of the current frame according to a predetermined criterion using the shift amount candidate and the shift amount of at least one past frame ;
A shift step of shifting all digital signal samples in the frame according to a shift amount determined in the shift amount selection step;
Encodes the portion excluding the least significant bit or brush shift amount of the bit range of the entire digital signal samples in the frame, and an integer signal encoding step of obtaining a code sequence,
A signal encoding method comprising: an output step of outputting information on a shift amount obtained in the shift amount selection step and a code string obtained in the integer signal encoding step. A signal encoding method according to claim 6 , comprising:
When the relationship between the shift amount candidate and one shift amount recorded in the frame shift amount holding buffer is within a predetermined range, one shift amount recorded in the frame shift amount holding buffer is set. As the shift amount of the current frame,
When the relationship between the shift amount candidate and one shift amount recorded in the frame shift amount holding buffer is outside a predetermined range, the shift amount candidate is set as the shift amount of the current frame. A signal encoding method comprising the shift amount selecting step. A signal encoding method according to claim 6 , comprising:
When the shift amount candidate is smaller than the shift amount of the immediately preceding frame recorded in the frame shift amount holding buffer, or when the difference between the shift amount candidate and the shift amount of the immediately preceding frame is larger than a predetermined threshold value Is the shift amount candidate of the current frame as the shift amount candidate,
If the shift amount candidate is equal to or greater than the shift amount of the immediately preceding frame and the difference between the shift amount candidate and the shift amount of the immediately preceding frame is equal to or less than a predetermined threshold, the shift amount of the immediately preceding frame is set to the current frame. A signal encoding method comprising the shift amount selecting step. A signal encoding method according to claim 6 , comprising:
The shift amount of the immediately preceding frame is recorded in the frame shift amount holding buffer,
If the shift amount candidate is greater than or equal to the shift amount of the previous frame, calculate the data amount after encoding for each shift amount from the shift amount of the previous frame to the shift amount candidate, and the data amount is the largest. The smaller shift amount is used as the current frame shift amount.
When the shift amount candidate is smaller than the shift amount of the immediately preceding frame, the shift amount candidate is set as the shift amount of the current frame. A signal encoding method according to claim 6 , comprising:
The shift amount of the past N frames (N is an integer of 2 or more) is recorded in the frame shift amount holding buffer,
In the case where the shift amount candidate is nth (n is an integer greater than or equal to 1 and less than N) and less than or equal to the shift amount of the previous frame, among the shift amounts of the past N frames, Let the shift amount of the previous frame be the shift amount of the current frame,
When the shift amount candidate is larger than the nth (n is an integer greater than or equal to 1 and less than N) shift amount smaller than the shift amount of the previous N frames, or smaller than the shift amount of the immediately preceding frame. Includes a shift amount selection step, wherein the shift amount candidate is a shift amount of the current frame. The program which implement | achieves the apparatus in any one of Claim 1 to 5 with a computer. The computer-readable recording medium which recorded the program of Claim 11 .

Images