Highly efficient coding transmission equipment

[Detailed Description of the Invention]

[Summary of the Invention] The optimal quantization is switched by choosing a line intermediary To have coding part for every frame among a plurality of adaptive coding parts by which parallel arrangement was carried out, It is the high-efficiency-coding transmission equipment which sets various parameters of the optimal selected coding part as other coding parts, and prevented the discontinuity of the reproduction signal by that cause at the time of the change. [Industrial Application] The present invention relates to the high-efficiency-coding device which carries out high efficiency coding of an audio signal, the image signal, etc., for example, carry out two or more owners of the adaptive coding part from which the renewal speed of step size of an adaptive quantizer differs especially, for every frame, switch by choosing a line intermediary To have adaptive coding part, and excite the optimal quantization -- it is alike and an improvement of the coding characteristic is related more with I figured high-efficiency-coding transmission equipment. For example, high-efficiency-coding transmission equipment of an audio signal is a device which performs information compression as [ It was / which maintains quality for an audio signal ]. In transmission of Voice information, reduction of memory capacity for accumulation is made [ when ] again voice accumulation and an audio response system for reduction of circuit cost possible in accumulation of Voice information when with mobile communications, satellite communication, or communication in a company. [Description of the Prior Art] The adaptive-predictive-coding machine as high-efficiency-coding transmission equipment of the conventional audio signal is shown in Drawing 5. The subtraction machine with which 80 output intermediary prediction error E (n) among a figure in the difference of voice input signal X (n) and predicted value (n) X ', As for the adaptive quantizer which 81 carries out adaptive quantizing of the prediction error E (n), and outputs quantization prediction error signal I (n) as a transmission output, the adaptation dequantization machine with which 82 carries out dequantization of the quantization prediction error signal I (n), and 83, an adding machine and 84 are zero predicted values X (n)._zThe adapted type zero prediction machine which generates ', and 85 are predicted values X (n) very much._pThe adapted type pole predicted value which generates ', and 86 are zero predicted values X (n)._zIt is [ ' and ] predicted value X (n) very much._pIt is an adding machine which adds ' and generates predicted value (n) X '. In this adaptive-predictive-coding machine, quantization machine 81, dequantization machine 82, zero prediction machine 84, and pole predicted value 85 have received adaptive control. in quantization machine 81 and dequantization machine 82 -- quantization step size delta (n) -- a lower type -- therefore, the adaptive control calculated and updated is performed. delta(n+1) =delta (n) =xM (I (n)) The thing of a value with gamma near [ here / in the coefficient which reduces the influence of line failure gradually ] Then and one or less 1, M (I (n)) is a renewal coefficient of step size (or renewal speed of step size), and is Then, For example, when quantization prediction error signal I (n) of quantization machine 81 is 1 A bit output, when I (n) is "0", value alpha near [ 1 ] 1 of smallness, for example, alpha= 0.93, is taken, and when I (n) is "1", value beta near [ 1 ] adult 1, for example, beta= 1.31, is taken. Renewal coefficient M of step size (I (n)) becomes large, so that the number denoted by the two or more A bit is size, when quantization prediction error signals I (n) are two or more A bit. In the prediction-coding machine of this Drawing 5, since values alpha and beta which determine it are given fixed, renewal speed M of step size in quantization machine 81 and dequantization machine 82 (I (n)) is a fixed value. For this reason, also when updating speed of quantization step size delta (n) which is a threshold for quantizing prediction error E (n) cannot fully respond to change of prediction error E (n) and cannot perform optimal quantization to it in quantization machine 81, it produces. Since the character of a signal changes every moment like a voiced sound, a voiceless sound, or silent etc when especially an input signal is an audio signal, at a fixed renewal speed of step size, it cannot respond to change of this audio signal, and optimal quantization is not performed. In order to solve this problem, in Japanese Patent Application No. 60-142307 concerning these people, the high-efficiency-coding transmission equipment which can perform optimal quantization is proposed. The high-efficiency-coding transmission equipment to apply is shown in Drawing 4. The inside of a figure, 7₁~7_kRenewal coefficient M of step size of a quantization machine [ in / at the adaptive-predictive-coding part of Are plural / each / Then and ] (I (n))₁~M(I(n))_kAlthough (values alpha and beta which were mentioned above in more detail) differ respectively and the basic composition has them with the adaptive-predictive-coding machine shown in Drawing 5, it is quantization prediction error signal I (n) as the output signal as difference. [ same ]₁~I(n)_kQuantization error e (n) which is a difference value with prediction error E (n) and output value [ of dequantization machine 82 ] E(n) ' at of etc.₁~e(n)_kIt is having the subtraction machine to calculate. Each adaptive-predictive-coding part 7₁~7_kQuantization prediction error signal I of Or and others (n)₁~I(n)_kIt is led to Is a selector 4 and is quantization error e (n).₁~e(n)_kIt is led to Is quantized error power calculation part 2. It is quantization error power calculation section 2 in the end of the frame for every (every [ for example, ] 16 samples) frame of input signal X (n), and is quantization error e (n).₁~e(n)_kQuantization error electric power is computed about Respectively, and the result is sent to the optimal quantization deciding part 3. The optimal quantization deciding part 3 is quantization error e (n).₁~e(n)_kof -- what has the inner smallest quantization error electric power is judged, in the frame concerned, the optimal quantization is determined for a corresponding adaptive-predictive-coding part as a line intermediary To have coding part, and indication signal op which directs the number of the coding part is sent to selecting part 4 and multiplexing part 5. Selecting part 4 is quantization prediction error signal I (n).₁~I(n)_kIt has the buffer memory stored by one frame, quantization prediction error signal I (n) of the coding part directed by indication signal op is chosen, and it sends to multiplexing part 5. Multiplexing part 5 gives indication signal op to selected quantization prediction error signal I (n), and sends it out to a receiving side. In a receiving side, it is prediction-coding part 7 of the transmitting side.₁~7_kIt has several dequantization machines with which renewal coefficients M of step size (I (n)) differ respectively corresponding to Quantizer, and a dequantization machine is chosen based on received indication signal op, and quantization prediction error signal I (n) is decoded. The above data processing is usually realized by what is called digital signal processor (DSP) by software as known well. A setup of the above parameters is performed by writing in a predetermined field of RAM which the above-mentioned digital signal processor (DSP) uses. A digital signal processor (DSP) reads the parameter set as this field if needed, and uses it for its voice data processing. Thereby, voice data processing based on the set parameter is performed. Thus, in the device of Drawing 4, a plurality of prediction-coding machines which have an adaptive quantizer in which updating speed differs are operated in parallel to an input signal, A line intermediary To have prediction-coding machine is chosen for the optimal quantization that makes the minimum electric power of quantization error e (n) for every every frame, and it is figure intermediary To have about an improvement of a line intermediary and the characteristic in a change. [Problem(s) to be Solved by the Invention] Each prediction-coding machine 7₁~7_kBecause it is a quantization machine which can be boiled and set, a dequantization machine, a zero prediction machine, and the thing as for which the prediction machine has received adaptive control very much, respectively, Parameters, such as a prediction coefficient of quantization step size [ of the parameter which determines the internal state of a coding part, for example, a quantization machine, and a dequantization machine ] delta (n), and a prediction machine, and A tap data, are each prediction-coding part 7 like the processing fault of an input signal therefore.₁~7_kIt changes to every one by one, and becomes a different Noodle thing, respectively. Therefore, when the selection change of the prediction-coding part chosen as the former on the boundary of the frame which chooses and switches the optimal prediction-coding part is carried out at other prediction-coding parts, the above-mentioned parameters of these prediction-coding part are a different Noodle thing and intermediary To have respectively. As a result, when performing the selection change of a prediction-coding part, based on the difference of this parameter, discontinuity arises in processing of an input signal on the boundary of a frame, therefore the discontinuity of a reproduction signal arises in a receiving side on a frame boundary, and the quality of a reproduction sound deteriorates. Thus, there is a problem that optimal processing is not necessarily performed, with 4th [ The ] figure device. [Means for Solving the Problem] Drawing 1 is a principle block diagram of high-efficiency-coding transmission equipment concerning the present invention. The inside of a figure, 101₁~101_kIt is a Is adaptive coding part and carries out hidden composition of an adaptive quantizer which outputs quantization output signal I according to input signal X, and the error calculation circuit which computes quantization error e of an adaptive quantizer, respectively. These adaptive coding machines 101₁~101_kRenewal speed M of step size of Adaptive quantizer₁~M_kAre different intermediary To have. 102 is a deciding part and is adaptive coding part 101.₁~101_kQuantization error e of Or and others₁~e_kAn adaptive coding part in which a Respectively input is carried out and optimal quantization is performed based on them is determined for every frame of input signal X. 103 is a selection sending part and is adaptive coding part 101.₁~101_kQuantization output signal I of Or and others₁~I_kA Respectively input is carried out, a quantization output signal of optimal adaptive coding part is chosen based on a determination result of deciding part 102, a determination result is attached, and it sends out to a receiving side. 104 is a parameter selecting part and is Then and adaptive coding part 101.₁~101_kParameter P which determines Or et al. and a part state₁~P_kA Respectively input is carried out and a parameter of an adaptive coding part determined that it is the optimal in the end of each frame is sent out to other adaptive coding parts. In each adaptive coding part, it is constituted so that a parameter sent from the parameter selecting part in advance of a start of processing of the following frame may be set as a self parameter. [Function] Input signal X is each adaptive coding part 101.₁~101_kIt is alike, and is inputted and they are these adaptive coding parts 101.₁~101_kIt comes out and is processed in parallel. As a result, adaptive coding part 101₁~101_kOr and others is quantization output signal I.₁~I_kQuantization error e₁~e_kAnd parameter P₁~P_kIt is outputted to Respectively selection sending part 103, deciding part 102, and parameter selecting part 104. Deciding part 102 is quantization error e.₁~e_kIt is alike, and it is based, a line intermediary To have adaptive coding part is determined for the optimal quantization to input signal X in the end of a frame, and the determination result is sent out to selection sending part 103 and parameter selecting part 104. Selection sending part 103 chooses quantization output signal I of the determined adaptive coding part, attaches a determination result, and sends it out to a receiving side. On the other hand, parameter selecting part 104 is sent to other adaptive coding parts which choose the parameter of a line intermediary To have adaptive coding part for the optimal quantization with the frame concerned, and arrange it in parallel based on the determination result. In each adaptive coding part, the parameter is set as a self parameter in advance of processing of the input signal of the following frame. The parameter in the end of the frame concerned of the adaptive coding part which appeared by this and was determined that it is the optimal with the frame, It will always be in agreement with the parameter in the start of the frame of the adaptive coding part determined that it is the optimal with the following frame concerned, and, as a result, the discontinuity of the reproduction signal in the boundary of a frame is lost. [Example] Hereinafter, the example of the present invention is described with reference to drawings. The high-efficiency-coding transmission equipment by the ADPCM method as one example of the present invention is shown in Drawing 2. In Drawing 2, it is 1.₁~1_kIt is a Is adaptive predictive coding part and voice input signal X (n) is inputted into each. This adaptive-predictive-coding part 1₁~1_kRenewal coefficient M of step size of The (I (n))₁~M(I(n))_kAre different from each other. This adaptive-predictive-coding part 1₁~1_kof -- detailed composition is shown in Drawing 3. In Drawing 3, subtraction machine 10 is sent to adaptive quantizer 11 in quest of input signal X (n) and prediction error E (n) which is difference with predicted value (n) X '. Adaptive quantizer 11 quantizes prediction error E (n), and outputs quantization prediction error signal I (n). This quantization prediction error signal I (n) is led to dequantization machine 12, and dequantization is performed there. Output value [ of the result ] E(n) ' is led very much to prediction machine 15 via adding machine 13 while it is led to zero prediction machine 14 and subtraction machine 17. Zero prediction machine 14 is zero predicted value X (n)._zCalculating and outputting ', prediction machine 15 is predicted value X (n) very much._p' is calculated and outputted and it is these zero predicted value X (n)._zIt is [ ' and ] predicted value X (n) very much._p' is added with adding machine 16 and predicted value (n) X ' is sent out to subtraction machine 10. Subtraction machine 17 outputs intermediary quantization error e (n) in difference with prediction error E (n) and output value E(n) '. the quantization step adaptation machine which quantization machine 11 and dequantization machine 12 are adapted type things, and it does not illustrate -- quantization step size delta (n) -- a lower type -- therefore, it is constituted so that it may be updated. delta(n+1) =delta (n) =xM (I (n)) Here, the coefficient and M (I (n)) in which gamma reduces the influence of line failure gradually are a renewal coefficient of step size. The parameter which determines the state of this quantization machine 11 and dequantization machine 12 serves as this quantization step size delta (n). It is zero prediction machine 14 and a thing of an adapted type [ machine / 15 / prediction ] very, and the degree which comprises a delay element and a coefficient unit is the l-th circuit [ m-th ], respectively. It has A tap data which has zero prediction machine 14 and a prediction coefficient by which prediction machine 15 is updated very much one by one, and holds the history of a fault in respect of the transfer function as a filter. That is, zero prediction machine 14 is zero prediction coefficient C of each coefficient unit by the prediction coefficient adaptation machine which is not illustrated._z(1,n)~C_zAlthough (l, n) used the lower type, respectively, renewal of an intermediary is carried out. C_z(i,n+1)=L_zxC_z(i,n)+D_zxsgn(E (n) ') xsgn (E(n-i) ') -- here -- L_zAnd D_zThe number of Is fixed and sgn are with the numerals function showing the numerals of positive/negative. Parameter P which determines the internal state of this zero prediction machine 14 is above-mentioned zero prediction coefficient C._z(1,n)~C_zIt is with (l, n), and A tap data E(n-1) '~E(n-l) ' into which it is put by each delay element. In a similar manner, very much, although number Cpof pole predicted value coefficients (1, n)~Cp (m, n) of each coefficient unit used the lower type with the prediction coefficient adaptation machine which is not illustrated, respectively, renewal of an intermediary of the prediction machine 15 is carried out. Cp(i, n+1) =LpxCp(i, n)+Dpxsgn(S (n) ') xsgn (S(n-i) ') -- Lp and Dp are the numbers of fixation here. Parameter P which determines the internal state of this pole prediction machine 15 is very much with prediction coefficient Cp(1, n)~Cp (m, n) and A tap data S(n-1) '~S(n-m) ' into which it is put by each delay element. In Drawing 2, as for each component of Drawing 3, quantization machine 11 is Q.₁~Q_kDequantization machine 12 is Q.₁^-1~Q_k^-1Hz1~Hzk and pole prediction machine 15 are expressed for zero prediction machine 14 by Hp1~Hpk, respectively. In Drawing 2, it is adaptive-predictive-coding part 1.₁~1_kQuantization error e of Or and others (n)₁~e(n)_kIt is led to Is quantized error power calculation part 2, respectively. Quantization error power calculation section 2 is crossed to one frame of input signal X (n), and it is quantization error e (n).₁~e(n)_kIt is a circuit which holds, computes those quantization error electric power in the end of a frame, and sends the calculation result to the optimal quantization deciding part 3. The optimal quantization deciding part 3 is based on the calculation result, and it is quantization error e (n).₁~e(n)_kof -- inner quantization error electric power judges the optimal quantization for the adaptive-predictive-coding part of the smallest thing in the frame concerned to be a line intermediary To have thing, and sends out to selecting part 4, multiplexing part 5, and parameter copying control part 6 by making the decision result of , into indication signal op. Adaptive-predictive-coding part 1₁~1_kQuantization prediction error signal I of Or and others (n)₁~I(n)_kIt is led to Are selecting part 4. This selecting part 4 is quantization prediction error signal I (n).₁~I(n)_kIt has a memory which can be stored by one frame, and sends out quantization prediction error signal I (n) for one frame of the adaptive-predictive-coding part directed by the indication signal op to multiplexing part 5 by receiving selection indication signal op. After multiplexing part 5 gives indication signal op to this quantization prediction error signal I (n) and changes it into transmission way numerals, it is sent out to a receiving side. In parameter copying control part 6, it is adaptive-predictive-coding part 1.₁~1_kParameter P which determines Or et al. and a part state₁~P_kNamely, quantization step size delta (n)₁~delta (n) _kZero prediction coefficient Cz₁~Cz_kIt is prediction coefficient Cp very much.₁~C_pkA tap data E'₁~E'_kAnd A tap data S'₁~S'_kA Respectively input is carried out. And according to indication signal op, parameter copying control part Pop of the directed adaptive-predictive-coding part is chosen, and they are all the adaptive-predictive-coding parts 1.₁~1_kIt is alike, and it is constituted so that it may send out. Operation of this example device is explained below. Input signal X (n) is each adaptive-predictive-coding part 1.₁~1_kIt is alike, and is inputted in parallel, prediction-coding processing is performed in each, and, thereby, it is quantization prediction error signal I (n).₁~I(n)_kAnd quantization error e (n)₁~e(n)_kIt Output. Quantization error e (n) as which quantization error power calculation section 2 was inputted₁~e(n)_kPower is computed for every frame and the calculation result is sent out to the optimal quantization deciding part 3. The optimal quantization deciding part 3 distinguishes adaptive-predictive-coding part 1op to which the quantization error electric power outputs quantization error e (n) which is the minimum, The adaptive-predictive-coding part 1op determines the optimal quantization as line intermediary To have, and sends out indication signal op which showed the number of the adaptive-predictive-coding part 1op to selecting part 4, multiplexing part 5, and parameter copying control part 6. Thereby, selecting part 4 sends quantization prediction error signal I(n) op for one frame of optimal adaptive-predictive-coding part 1op directed by that indication signal op to multiplexing part 5, and multiplexing part 5 gives indication signal op to this quantization prediction error signal I(n) op, and transmits to a receiving side. Parameter P into which parameter copying control part 6 is inputted on the other hand₁~P_kof -- from inside, parameter Pop of adaptive-predictive-coding part 1op corresponding to indication signal op is chosen, and it is sent out to all the adaptive-predictive-coding parts. All the adaptive-predictive-coding parts of the others which this arranges in parallel set sent parameter Pop as a self parameter in the end of a frame. That is, in each adaptive-predictive-coding part, quantization step size delta(n) op is set to quantization machine 11 and dequantization machine 12, A tap data S'op is set to the coefficient unit of zero prediction machine 14 at zero prediction coefficient Czop and its delay element, and is set to the coefficient unit of A tap data E'op and pole prediction machine 15 at a zero prediction machine, pole prediction coefficient Cpop, and its delay element, respectively. As a result, in the start time of processing of the following frame, it will have the internal state as the selected adaptive-predictive-coding part with all the same adaptive-predictive-coding parts, therefore is for example, the present frame, and is adaptive-predictive-coding part 1.₁It is chosen as a But optimal thing, is the following frame, and is adaptive-predictive-coding part 1.₂But -- adaptive-predictive-coding part [ in / even when chosen as optimal thing / the end of the present frame ] 1₁of parameter P₁Adaptive-predictive-coding part 1 in the start of the following frame₂of parameter P₂What's -- it will be in agreement and the discontinuity of the data in the boundary of a frame is lost. Probably, also in the high-efficiency-coding transmission equipment of the present invention, it will usually be clear for a person skilled in the art that above data processing's software may be realized using a digital signal processor (DSP), as the item of [Description of the Prior Art] was described previously. For example, a setup of the above parameters is performed by writing in the predetermined field of RAM which the above-mentioned digital signal processor (DSP) uses. [Effect of the Invention] According to the present invention, losing the discontinuity of the signal in the frame boundary for the selection change of the optimal adaptive coding part for every frame, signal processing can be performed, always using the optimal quantization characteristic according to the character of the input signal, and quality, such as a reproduction sound or a reproduction picture, can be improved.

[Brief Description of the Drawings]

Drawings 2 are a principle block diagram of the high-efficiency-coding transmission equipment which requires Drawing 1 for the present invention, and a block diagram showing the high-efficiency-coding transmission equipment as one example of the present invention, The block diagram of the high-efficiency-coding transmission equipment relevant to the present invention in the block diagram and Drawing 4 showing the details of an adaptive-predictive-coding part [ in / in Drawing 3 / 2nd / The / figure device ] and Drawing 5 are block diagrams showing the adaptive-predictive-coding machine as conventional high-efficiency-coding transmission equipment. 1₁~1_k,7₁~7_k...... An adaptive-predictive-coding part, 2 ...... A quantization error power calculation section, 3 ...... The optimal quantization deciding part, 4 ...... A selecting part, 5 [ ...... A quantization machine, 12 / ...... A dequantization machine, 13, 16 / ...... An adding machine 14 / ...... A zero prediction machine, 15 / ...... It is a prediction machine and 101 very much. ] ...... A multiplexing part, 6 ...... A parameter copying control part, 10, 17 ...... A subtraction machine, 11₁~101_k...... [ ...... Parameter copy part. ] An adaptive-predictive-coding part, 102 ...... A deciding part, 103 ...... A selection sending part, 104