Digital audio transmission.

[Detailed Description of the Invention]

[Industrial Application] The present invention relates to the data communications which used sub band coding. Especially the present invention is suitable for transmission of the multiplexed data and the audio signal coded in digital one. Sub band coding divides into two or more subbands the frequency band of the basis over which voice information and other spectrum are distributed, and means coding each frequency band separately. [Description of the Prior Art] Pulse code modulation (pulse code modulation, PCM) or an Adaptive Differential Pulse Code Modulation (adaptive differentialPCM ADPCM) is used for the main advantages of sub band coding, for example, (a) Without it limits the quantization noise generated in each subband in the frequency band and carries out [ sound / of other zones / of a low level ] a mask at the time of decryption, (b) the signal of each subband -- an auditory standard -- the energy distribution -- therefore, code and be in the ability to provide good sound quality auditorily as the whole subband. How to multiplex a data signal to such a signal by which sub band coding was carried out is considered. for this reason -- being alike -- there is a method of inserting a data signal in one channel of a subband fixed or the method of using A bit currently assigned to the subband with few amounts of energy for transmission of a data signal. [Problem(s) to be Solved by the Invention] However, in the method of inserting a data signal only in one channel, the excessive signal for [ the ] data signals is needed, the effect of data multiplexing is fully acquired, and there is a fault. The fault for which accompanying information is needed in order to judge the subband in which the data signal was inserted in the method of assigning a subband with few amounts of energy is Oh. The present invention solves the above fault and an object of the present invention is to provide the numerals transmission equipment which codes the audio signal and data signal of a subband efficiently, and can transmit. [Means for Solving the Problem] In a numerals transmission method which an invention of the first of the present invention is a numerals transmission method, and divides an input signal into two or more frequency bands, i.e., a subband, codes separately and transmits a signal of each frequency band for every zone of the, The number of bits required for coding is beforehand assigned to each frequency band, and it is at the transmitting side, It reduces accommodatively, and transmits in a bit position which has responded to a value of a signal by which numerals were carried out, and was able to define beforehand the number of bits of a coded signal in a gap or one or more frequency bands, and a value of a received signal determines the number of bits required for decryption in a receiving side. That is, a subband determines reduction of A bits with reference to A bit transmitted regularly on both sides of the transmitting side and a receiving side. A subband divided from other subbands may be sufficient as itself, and a subband directly divided from a frequency band of voice information and other bases may be sufficient as a subband. It is the method of inserting and transmitting a data signal to the A bit in an invention of the second of the present invention rather than reducing the number of bits. That is, it is while dividing an input signal into two or more frequency bands, coding separately and transmitting a signal of each frequency band for every zone of the, In a numerals transmission method which inserts and transmits a data signal to a signal with which at least one frequency band was coded, The number of bits required for coding is beforehand assigned to each frequency band, and it is at the transmitting side, A data signal is inserted in a bit position where it has responded to a value of a coded signal, and a coded signal in a gap or one or more frequency bands was defined beforehand, and it transmits to it, and is at a receiving side, Signals other than A bit equivalent to the bit position appointed beforehand perform determination of the number of bits required for decryption, and separation of a data signal. That is, in order to unify a method of insertion and separation of a data signal, and coding and decryption, A bit in which a data signal is not inserted, i.e., A bit regularly transmitted in the subband, is referred to on both sides of the transmitting side and a receiving side. The present invention is suitable for multiplexing transmission with an audio signal and a data signal. In that case, an input signal is an audio signal. It is desirable to judge an amount of energy of the frequency band from a coded signal, and for an amount of energy to insert a data signal in a signal of few frequency bands. The judgment of an amount of energy does not need to be a strict judgment, and, in higher rank A bit of a coded signal, or embedding adaptation differentiation pulse code abnormal conditions, it can judge it from coding step size. It is desirable to measure an amount of energy between two adjoining frequency bands. As for comparison of an amount of energy, comparing with a difference by a ratio is desirable. When a difference or a ratio of an amount of energy of two frequency bands is smaller than a value defined beforehand, it is desirable to insert a data signal in a signal coded by both frequency bands. When embedding a signal of each frequency band and coding by adaptation differentiation pulse code abnormal conditions, a rate of an amount of energy of each frequency band can be calculated by comparing step size of this coding. It uses for a voice transmission system which a zone codes an audio signal which is 7 kHz by 8 A bit, and it transmits at 64 kbits/s, and is in a case, This 7-kHz zone is divided into two 3.5-kHz adjoining frequency bands, At a frequency band by the side of low-pass, an input signal is coded by 5 A bit and an input signal is coded by 3 A bit by a frequency band by the side of a high region, A data signal can be accommodatively inserted in one A bit of one side of numerals of 5 A bit obtained by a frequency band by the side of low-pass, and numerals of 3 A bit obtained by a frequency band by the side of a high region, or both sides. An invention of the third of the present invention is a device which carries out the first invention mentioned above, is provided with a sending set and a receiving set which transmit and receive a coded signal, and is a sending set, It has a filtering means to divide an input signal into two or more frequency bands, and an encoding means which codes a signal of each frequency band separately, and is a receiving set, In numerals transmission equipment including a decoding means which receives a signal from a sending set and it decrypts for every frequency band, it is a sending set, A transmitting side judging means which judges a relative amount of energy of each frequency band by A bit which is contained in an output of an encoding means, and which was defined beforehand is included, and it is a receiving set, A receiving side judging means which judges a relative amount of energy of each frequency band with reference to the A bit defined beforehand is included. The present invention is a device which carries out the second invention mentioned above, and is provided with a sending set and a receiving set which transmit and receive a coded signal, and an invention of the fourth of the present invention is a sending set, It has a filtering means to divide an input signal into two or more frequency bands, and an encoding means which codes a signal of each frequency band separately, and is a receiving set, Numerals transmission equipment including a decoding means which receives a signal from a sending set and it decrypts for every frequency band is characterized by comprising: A transmitting side judging means a sending set judges a relative amount of energy of each frequency band by A bit which is contained in an output of an encoding means, and which was defined beforehand to be, A receiving side judging means a receiving set judges a relative amount of energy of each frequency band with reference to the A bit defined beforehand including a data inserting means which inserts a data signal in a signal coded by this decision result by at least one frequency band to be, A means to separate a data signal based on an output of this receiving side judging means. The encoding means can include a means by which a transmitting side judging means and a receiving side judging means compare quantization step size of an embedding adaptation difference pulse code, respectively, including an embedding Adaptive Differential Pulse Code Modulation numerals machine. As for a transmitting side judging means and a receiving side judging means, it is desirable respectively to include a comparison means to measure a relative amount of energy of a signal of two adjoining frequency bands. As for a comparison means, it is desirable to include a means to calculate a ratio of a relative amount of energy. As for a data inserting means, it is desirable for A bit which a transmitting side judging means and a receiving side judging means of a signal which were coded refer to to be composition which inserts a data signal in a different bit position, It is desirable that it is the composition which inserts a data signal in a signal coded by a frequency band with few relative amounts of energy alternatively. As for a data inserting means, it is desirable that it is the composition which inserts a data signal in a signal of both sides where a difference of an amount of energy of two frequency bands was coded by two frequency bands by value defined beforehand when small. As for a filtering means, it is desirable to include a Deisil filter which divides an input signal into a plurality of frequency bands. [Function] The number of bits which codes the subband changes accommodatively with some of amounts of energy in which a subband includes the numerals transmission method of the present invention. However, A bit always transmitted to the subband performs a judgment to some of amounts of energy, and unfixed A bit does not use for a judgment which subband is transmitted. For this reason, the signal which arrives at a receiver can also be judged only by A bit always transmitted to that subband. Thus, since all the A bit are not used for the judgment of the amount of energy of a subband, the data signal as accompanying information can be inserted in an audio signal using A bit which is not used for this judgment, and the subband in which the data signal is inserted can be easily judged in a receiver. [Example] Drawings 1 and 2 are block lineblock diagrams of the first example numerals transmission equipment of the present invention, Drawing 1 shows a transmitter and Drawing 2 shows a receiver. Audio signal 1 by which PCM coding was carried out at sampling speed of 16 kHz is inputted into the transmitter of this example numerals transmission equipment. The frequency spectrum of this inputted audio signal 1 is divided into a high region subband and a low-pass subband by high pass filter 2 and low-pass filter 3. these two subbands -- "Decimator" -- a down sample is carried out to 8 kHz by 4 and 5, respectively. In order to explain simply, quantization of the inputted audio signal assumes that it is what is performed in the accuracy of 4 A bit with both subbands. Actually, much more numbers of bits need to be used for coding, and the coding number of bits of each subband does not need to be still the more nearly same. Multiplexer 6 corrects and outputs a set of output A bit from Decimator 4 and 5 to the form of having been suitable for transmission. Some Time-Division-Multiplexing communication apparatus may be sufficient as multiplexer 6, and a simple parallel in-series converter may be sufficient as it. Multiplexer 6 inputs the data of 8 A bit in this example. One A bit of these 8 A bit is everlastingly assigned to transmission of spare data signal D. With, Also and this are not the essence of the present invention. Left 7 A bit is accommodatively assigned to two channels which are the inputs of multiplexer 6. Here, one channel is quantized in the accuracy of 3 A bit, and other channels are quantized in the accuracy of 4 A bit. Therefore, triple higher rank beams h of the channel by the side of a high region₃Or h₁And triple higher rank beams l of the channel by the side of low-pass₃Or l₁It is assigned to each input of But and multiplexer 6. (On the other hand, the lowest A bit of the channel as which the 8th input of multiplexer 6 was chosen via selection circuit 8, i.e., least significant digit h by the side of a high region)₀Or least significant digit l by the side of low-pass₀of one side is received. The receiver of Drawing 2 is the mirror image and intermediary To have of a transmitter fundamentally. The higher rank beam input of 3 A bit of Up sample circuits 14 and 15 is supplied, the output of Up sample circuits 14 and 15 is supplied to interpolation filters 12 and 13, the output of interpolation filters 12 and 13 is added with adding machine 10, and demultiplexer 16 serves as decryption output 11. Explanation is returned to the transmitter of Drawing 1. Selection circuit 8 is looked like [ decision circuit 9 ], and is controlled more, and decision circuits 9 are the double lowest beams h.₀Or l₀It is chosen whether Afterlife and others is transmitted. This judgment is determined as the signal between [ two ] channels, or a function of an amount of energy. Selection corresponding to selection with a transmitter is performed by 1 A bit demultiplexer 18, and it is functionally controlled by a receiver by equivalent decision circuit 19 with decision circuit 9. Decision circuit 19 of a receiver is higher rank beam h of two channels, in order "to pursue" correctly in operation of a transmitter.₃Or h₁And l₃Or l₁A Using judging is performed. Naturally decision circuit 19 of a subband decoder is least significant digit h.₀l₀It is alike, and it does not access and it "is not concerned" whether it was chosen before which A bit's judging. The output of the least significant digit of one channel is suspended so that noise may not become so large, even if it investigates which high channel decision circuits 9 and 19 have energy with, and there is no least significant digit by this. Since noise does not increase so much even if it reduces the number of bits of a side with low energy in the case of an audio signal, the output of the least significant digit of the side here is suspended. This operation is performed for every specimen. A simple comparison machine is enough for this. In an actual device, the comparison average of the energy presumption for every cycle is carried out. This is explained in more detail below. Drawings 3 and 4 show the second example numerals transmission equipment of the present invention, Drawing 3 shows the block lineblock diagram of a transmitter, and Drawing 4 shows a receiver block lineblock diagram. In the transmitter of this example, it assigns 4 A bit at a time to two channels using eight A bit. Spare data signal D is least significant digit h of two channels.₀Or l₀It is accommodatively inserted in Position of. Decision circuit 9' of this example outputs two outputs x and y. At the time more than a threshold with a low energy ratio of the high region side channel and a low-pass side channel, output x becomes effective, and output y becomes effective when it is below a high threshold. This is explained with reference to Drawing 9. A horizontal axis expresses energy [ of an energy ratio, i.e., the [high region side channel, ] ]/[energy of a low-pass side channel]. In order to explain simply, suppose that the energy ratio of two thresholds is 1±delta (0< delta<<1). When an energy ratio is more than 1-delta, decision circuit 9' validates output x. Thereby, selection circuit 8 by the side of low-pass is least significant digit l.₀Data signal D is chosen as instead of. On the other hand, when an energy ratio is below 1+delta, decision circuit 9' validates output y, and it is least significant digit h by the side of a high region.₀Data signal D is chosen as instead of. When an energy ratio is between 1-delta and 1+delta, it originates in calculation accuracy, a limit, or a transmitting error, and may make a different Ivy judgment by the transmitting side and a receiving side. In such a case, a data signal may be lost, if a data signal is transmitted and received only by one channel and it will carry out. Then, the data signal is inserted in both channels. Thereby, the errors of a data signal are reducible. The still more practical example of the transmitter of this example and a receiver is described. However, many portions explained below are feasible similarly in the first example. A transmitter is suitable for inputting a wide band sound signal with a bandwidth of 7 kHz sampled at 16 kHz, and transmits at 64 kbits/s using an Adaptive Differential Pulse Code Modulation. When A bit is usually eliminated, in order that a difference numerals machine may have a fault with a large signal to noise ratio and may correct this, an embedding Adaptive Differential Pulse Code Modulation (Embedded ADPCM) is used. About embedding coding (Embedded coding), David Jee Goodman (David) J. Coodman "the IEEE report about communication (IEEE Trans.Comm.) the [" ] -- COM- 28 No. 7 (July, 1980) -- the 1040th, thru/or crowded to Burial for "variable A bit degree degree transmission announced to the 1046th page -- difference pulse code modulation (Embedded DPCM) for Variable bit rate transmission" It is alike and is explained in detail. However, the numerals transmission method currently explained by this literature is not adaptive coding, and is not used for sub band coding, either. It embeds fundamentally and the low signal noise ratio by difference pulse code modulation not having exact prediction and quantization step size high by this are needed, Therefore, quantization noise solved problems, such as a large thing, and has eliminated all the A bit which should be removed by transmission from both prediction return circuit of a transmitter and a receiver. Drawings 5 and 6 show the numerals transmission equipment of the third example of the present invention, Drawing 5 shows the block lineblock diagram of a transmitter, and Drawing 6 shows the block lineblock diagram of a receiver. The transmitter of this example numerals transmission equipment is provided with digital filter 22. This digital filter 22 divides into two subbands the signal inputted from input terminals 23, such as a signal by which comprised high region Foil 22a and low-pass filter 22b, for example, PCM coding was carried out. The output of high pass filter 22a and low-pass filter 22b is inputted into down sample circuits 24a and 24b of "Decimator" 24, respectively, and a down sample is carried out to them. The output of down sample circuit 24a is supplied to ADPCM code machine 25, and the output of down sample circuit 2b is supplied to ADPCM code machine 26. ADPCM code machines 25 and 26 output numerals word outputs 27 and 28, respectively. ADPCM code machines 25 and 26 supply the signal which shows step size to decision circuit 31 by the outputs 29 and 30. It is connected to data insertion circuits 32 and 33, and the output of decision circuit 31 performs control which is explained below. The output of data insertion circuits 32 and 33 turns into an input of multiplexer 34 of 8 A bit, and the output of this multiplexer 34 is outputted as a transmitted signal. Digital filter 22 comprises a rectangular mirror image filter (quadrature mirror filter), and it is used in order to divide input spectrum into two zones (0~4 kHz and 4~8 kHz) with which a part overlaps. The constant of a suitable filter is shown in the 1st table. The 1st table is constant h of the low-pass filter of 32 bit configurations._L(n) is shown. The constant of a low-pass filter and the constant of a high pass filter are h._u(n) =-1 and h_L(n) There is connection of. here -- n= 0, 1, and 2 -- it is ...... (N-1) and N is the number of bits.

[Table]

[Table]

The signal included in each of two subbands is supplied to down sample circuits 24a (4~8 kHz) and 24b (0~4 kHz), respectively. These down sample circuits 24a and 24b sample per second 8000 specimens, and output them to two independent ADPCM code machines 25 and 26, respectively. ADPCM code machines 25 and 26 are publicly known in itself, and coding of two subbands in ADPCM code machines 25 and 26 is enough here, if cautious of it not being symmetrical. For example, a low-pass side subband is coded by ADPCM by the prediction circuit of the fourth Udah of fixation optimized to the sound of the robust Jayant quantization circuit (robust Jayant quantiser) of 5 bit configurations, and 4-kHz bandwidth. On the other hand, a high region side subband (4~8 kHz) is coded using the robust Jayant quantization circuit of 3 bit configurations, and the prediction circuit of the first Udah of fixation. In a receiver, demultiplexer 55 separates into a low-pass side and a high region side subband, and processes a signal in an order contrary to a transmitter fundamentally. Digital filter 42, interpolation filters 42a and 42b, adding machine 43, Up sample circuits 44, 44a, and 44b, ADPCM decoders 45 and 46, decision circuit 51, data signal regenerative circuits 52 and 53, and a demultiplexer, Fundamentally, they are digital filter 22 and high pass filter 22a, respectively, It is a mirror image of low-pass filter 22b, input terminal 23, Decimator 24, down sample circuits 24a and 24b, ADPCM code machines 25 and 26, decision circuit 31, data signal insertion circuits 32 and 33, and a multiplexer. Therefore, it dissociates, and a 64 kbits/s signal is decrypted and interpolated, and is processed with digital filter 42. be absorbed -- be alike -- the signal of 7-kHz bandwidth is reproduced more. Two channels are connected with adding machine 43 next. The transmitter and the receiver equip additionally decision circuit 31 for data signal insertion, and decision circuit 41 for data signal reproduction with data signal insertion circuits 32 and 33 and data signal regenerative circuits 52 and 53, respectively. 3 A bit and 5 A bit output of Everyone of ADPCM code machines 25 and 26 are supplied to data signal insertion circuits 32 and 33. data signal insertion circuits 32 and 33 -- the position of the 5th coding A bit (the lowest A bit) of a low-pass side channel output, or the position of the 3rd coding A bit of the high region side channel -- on the other hand -- or data signal D is inserted in both sides, and it replaces with the A bit there to them. The method of insertion of data signal D is explained below. The output of the transmitter at the time of inserting data signal D in the channel of a high region side, a low-pass side, or both sides, It becomes 5+ (2+D), and +3 or (4+D) (4+D)+ (2+D). Decision circuits 31 and 51 can be carried out with a simple comparison machine, and compare adaptation step size delta from ADPCM code machines 25 and 26 or ADPCM decoders 45 and 46 (scaling factor), respectively. The block lineblock diagram of ADPCM code machines 25 and 26 is shown in Drawing 7, and the block lineblock diagram of ADPCM decoders 45 and 46 is shown in Drawing 8. The signal coded to one input of addition circuit 60 is inputted. The reversal input of the output of prediction circuit 67 is carried out at the input of another side of addition circuit 60. The output of addition circuit 60 is connected to one input of multiplication circuit 61. One output (1/delta, however delta are adaptation step sizes) of step size adaptation circuit 64 is inputted into another side of multiplication circuit 61. The output of multiplication circuit 61 is connected to quantization circuit 62. The output of quantization circuit 62 turns into an output of this ADPCM code machine. The output of quantization circuit 62 is connected to coefficient circuit 63. The output of coefficient circuit 63 is connected to step size adaptation circuit 64 and addition circuit 65. Addition circuit 65 adds 0.5 to the output of coefficient circuit 63, and inputs it into one side of multiplication circuit 66. Step size adaptation circuit 64 outputs adaptation step size delta and its reciprocal, and the output of the adaptation step size delta is inputted into multiplication circuit 66. The output of multiplication circuit 66 is connected to prediction circuit 67 for performing adaptation prediction of an audio signal. The adaptation step size delta output of step size adaptation circuit 64 is used for the judgment of the channel which data signal D inserts. The audio signal inputted into the ADPCM decoder is inputted into addition circuit 70 and coefficient circuit 71. The output of coefficient circuit 71 is inputted into step size adaptation circuit 72 and addition circuit 73. Addition circuit 73 adds 0.5 to the output of coefficient circuit 71, and inputs it into multiplier 74. The output of step size adaptation circuit 72 is inputted into multiplication circuit 74 and multiplication circuit 75, and is further used for the judgment of the channel in which data signal D was inserted. The output of multiplication circuit 74 is inputted into prediction circuit 76. The output of multiplication circuit 75 and the output of prediction circuit 76 are inputted into addition circuit 77. The output of addition circuit 77 turns into an output of this ADPCM decoder. The deciding method of a judgment of the data signal insertion by such an ADPCM code machine and a decoder and a data signal reproducing method are explained. It is necessary to judge in which channel the data signal is included without the accompanying information for dissociating. Back adaptation data multiplexing is used for such a judgment. The judgment for choosing the data channel of multiplexing and separation by back adaptation data multiplexing is drawn by each step size adaptation circuit 64 and 72 of a subband numerals machine and a subband decoder. The channel which inserts a data signal is determined as follows. Spectrum envelope presumption obtained from each subband signal by measuring an amount of energy fundamentally is obtained. By comparing these spectrum envelope presumption, the signal judges either owner sounds (vowel etc.), a non-sound (consonant) and an intermediate tone. When an audio signal is an owner sound, a data signal can be inserted in the high region side channel (4~8 kHz), and a low-pass side channel with many amounts of energy can be sampled by 5 A bit. On the other hand, when a non-sound, or voice sound or a non-sound does not have a clear audio signal, a data signal is inserted in a low-pass side channel (0~4 kHz), and this codes the audio signal of the zone by the side of low-pass by 4 A bit. At this time, the high frequency of an audio signal can be correctly coded by the high region side channel using all of 3 A bit. In order to judge the channel which inserts a data signal, spectrum envelope information is acquired for a short time using the step size parameter of the quantization circuit of two channels. This information can be derived in back adaptation mode, when using a Jayant algorithm, The n+1st step size parameters delta l (n+1) and delta u (n+1) of a low-pass side and each quantization circuit of the high region side channel, deltal(n+1) =deltal(n) = and Ml (Il (n)) ... (1), and deltau(n+1) =deltau(n) = and Mu (Iu (n)) ... (2) It is come out and given. Here, Il (n) is the n-th output-codes word from the quantization circuit of a low-pass side channel, and Iu (n) is the n-th output-codes word from the quantization circuit of the high region side channel. Ml (Il (n)) and Mu (Iu (n)) are the functions of Il (n) and Iu (n), respectively. A coefficient value by coefficient circuit 63 is shown. The value over the quantization circuit in the various numbers of bits is shown in the 2nd table. gamma is a "leak factor", and it is required in order to erase the effect of a transmitting error. Typically, gamma is chosen as 0.984.

[Table]

(1) type and (2) types are shown in the form of logarithm, and complicated index operation is eliminated so that conveniently [ explanation of a numerals machine ]. Namely, dl(n+1) = gamma-dl (n)+ml (Il (n)) ... (3) and du(n+1) = gamma-du (n)+mu (Iu (n)) ... (4) It comes out and expresses. however, dl (n), du (n), and ml (Il (n)) and mu (Iu (n)) -- respectively -- the logarithm of deltal (n), deltau (n), and Ml (Il (n)) and Mu (Iu (n)) -- it is a value. The energy envelope of a paragraph can be measured by an above-mentioned recurrence algorithm for a short time [ of the sound divided into the zone of each subband ]. In a subband decoder, it is determined as follows in which subband the data signal is inserted. First, the ratio of the quantization step size of a low-pass side subband and a high region side subband, V(n) = mu-delta u (n) /deltal (n) ... (5) Obtainable with -- having . Here, mu is a constant and V (n) is a signal which is applied to continuation data bit assignment and which carries out time change. in order to simplify digital processing -- (5) types -- the formula of logarithm -- v(n) =log(deltau (n))+log(mu)-log (deltal (n)) ... (6) =du(n)-dl(n)+log (mu) ... (7) It expresses. In the experiment, the value of mu of (7) types was set as 2. When the threshold of a judgment is set as v(n) =0, it is clear that the judgment with voice sound and voice sound is performed. Regrettably, in such a strict judgment, when a judgment level has a near value of v (n), a possibility that the error of the channel selection in which the data signal was inserted will occur systematically arises. This is based on a limit of accuracy of measurement of an energy envelope. Function v (n) by a signal inputted into a subband decoder originates in not synchronizing with function v (n) which a subband decoder outputted completely. For this reason, when function v (n) comes between threshold [ with the upper bottom ] u, and l, a data signal is simultaneously inserted in both subbands. In this case, in order to insert a data signal in both channels, the accuracy of an audio signal is degraded momentarily, but the transmitting errors of a data signal can be reduced and the systematic errors by the arrangement difference in the step size parameter which occurs with a subband numerals machine and a subband decoder can be reduced. The data signal insertion method explained above was Simulate(ed) by floating a small number of point FORTRAN on the mini computer, and it compared with fixed data bit assignment. In fixed data bit assignment, a data signal is continued for all the time, and it is multiplexed by one subband numerals word by the side of a high region or low-pass. The limit of the limited accuracy of NEC Corp. make muPD7720 microprocessor used with a usual speech coding transmitting machine and receiver was simulated, and the irregular error was inputted into the output by which the (channel) of the subband numerals machine was made in-series. As a result, for example, it became clear that it was convenient for inserting an 8 kbits/s data signal in the channel of the 56 kbits/s voice band by which sub band coding was carried out. The subband numerals method of the present invention can make data-communications capacity increase by performing adaptation prediction of a low zone, for example. The present invention is 1. Composition is comparatively easy and it can use for the maximum the treatment capacity obtained with the digital signal-processing micro computer used for a numerals machine, 2 It is not Necessary to Transmit Accompanying Information for Control Required for Data Multiplexing Function, 3 In Above-mentioned Example, Fixed Data-Communications Capacity of 8 Kbits/s is Realizable, 4 it is possible to also make data-communications capacity increase the present invention to 216 kbits/s or more, combining other art -- etc. -- there is an advantage. [Effect of the Invention] As explained above, the present invention is comparatively easy to constitute, the treatment capacity obtained with the digital signal-processing micro computer used for a numerals machine can be used for the maximum, and there is an effect it becomes unnecessary to transmit the accompanying information for control required for a data multiplex function. As compared with fixed-data multiplexing which fixes for example, to a low-pass side channel, and assigns a data signal, the present invention has an effect which can reduce notably distortion of the sound which can be perceived.

[Brief Description of the Drawings]

Drawing 1 is a block lineblock diagram of the transmitter of the first example numerals transmission equipment of the present invention. Drawing 2 is a block lineblock diagram of a receiver. Drawing 3 is a block lineblock diagram of the transmitter of the second example numerals transmission equipment of the present invention. Drawing 4 is a block lineblock diagram of a receiver. Drawing 5 is a block lineblock diagram of the transmitter of the third example numerals transmission equipment of the present invention. Drawing 6 is a block lineblock diagram of a receiver. Drawing 7 is a block lineblock diagram of an ADPCM code machine. Drawing 8 is a block lineblock diagram of an ADPCM decoder. Drawing 9 is a figure explaining the operation which judges the insertion point of a data signal. 2 ... high pass filter, 3 ... a low-pass filter and 4, 5 ... Decimator, 6 ... multiplexer, 8 ... selection circuit, 9 ... decision circuit, 10 ... adding machine, 12, 13 ... An interpolation filter, 14, 15 ... Up sample circuit, 16 ... demultiplexer, 18 ... 1 A bit demultiplexer, 19 ... decision circuit, 22 ... digital filter, 22 a ... high pass filter, 22 b ... low-pass filter, 23 ... input terminal, 24 ... Decimator, 24a, 24b ... A down sample circuit, 25, 26 ... ADPCM code machine, 31 ... a decision circuit and 32, 33 ... data signal insertion circuit, 34 ... multiplexer, 42 ... a digital filter and 42a, 42 b ... interpolation filter, 43 ... an adding machine, and 44 and 44a, 44 b ... Up sample circuit, 45, 46 ... An ADPCM decoder, 51 ... A decision circuit, 52, 53 ... Data signal regenerative circuit, 55 ... demultiplexer, 60 ... addition circuit, 61 ... multiplication circuit, 62 ... quantization circuit, 63 ... coefficient circuit, 64 ... step size adaptation circuit, 65 ... addition circuit, 66 ... multiplication circuit, 67 ... prediction circuit 70 [ ... An addition circuit, 74 / ... A multiplier, 75 / ... A multiplication circuit, 76 / ... A prediction circuit, 77 / ... Addition circuit ] ... An addition circuit, 71 ... A coefficient circuit, 72 ... A step size adaptation circuit, 73