JPH0595333A - Packet transmitter, its transmission reception section and packet transmission method and its transmission reception method - Google Patents

Abstract

PURPOSE:To minimize the deterioration in the transmission quality by allowing a receiver side to take an error in a reproduced signal to be tentative even when a transmission error takes place in an ADPCM (adaptive differential pulse modulation) code. CONSTITUTION:A prediction signal/quantization width addition circuit 4 in a transmission section 500 uses a packet processing circuit 3 to add a prediction signal and a quantization width to a code at the head of a code string subject to block processing. A prediction signal/quantization width read circuit 57 of a reception section 600 reads a prediction signal 85 and a quantization width 88 added by a transmission section from a packet in a buffer 51 and gives them respectively to selection circuits 59, 58. The selection circuit 59 selects the prediction signal 85 at the reproduction of a code of the head in the packet and selects the prediction signal 84 generated in a prediction circuit 56 to the succeeding codes. The selection circuit 58 selects the quantization width 88 at the reproduction of the head code similarly and selects the quantization width 87 of the quantization width decision circuit 53 at the reproduction of the succeeding codes.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a packet transmission device, and more particularly, to a packet transmission device for adaptively digitizing voice signal information by adaptive differential pulse modulation (ADPCM) coding, and further transmitting the packet by dividing it into packets. And a packet transmission method and a transmission / reception method thereof.

[0002]

2. Description of the Related Art In a packet transmission apparatus for packetizing and transmitting digitized voice signal information and the like, adaptive differential pulse code modulation (AD) is used to reduce the amount of information to be transmitted.
Information may be encoded by the PCM) method.

In the transmitting section of such an ADPCM packet transmission device, the quantization width is made coarse in the section where the level fluctuation of the input signal is large, and conversely, the quantization width is made fine in the section where the level fluctuation is small. The difference value between the input signal value and the predicted signal value is encoded while changing the quantization width according to the level change of the input signal.

The ADPCM code string obtained in this way is divided into a predetermined length, and control information such as the source and destination addresses is added and packetized and transmitted. On the other hand, the receiving unit generates the same prediction signal value and quantization width as the transmitting side based on the code string read from the received packet, adds the prediction signal value to the difference value obtained by encoding the code, and reproduces it. You are getting the signal value.

[0005]

However, in the above-mentioned conventional packet transmission apparatus using the ADPCM system, the predicted signal value and the quantization width generated in the receiving section are based on all the codes received in the past. Since it is decided, if an error occurs in the code due to a failure in the transmission path, or if a part of the code string is lost due to packet discard, the prediction that occurs in the receiving unit even if the code string to be received later is correct After that, the signal value and the quantization width are all wrong. Therefore, the reproduction signal value obtained by adding the prediction signal value to the difference value obtained by encoding the received code becomes different from the original signal encoded at the transmission side thereafter. was there.

An object of the present invention is to eliminate the above-mentioned conventional problems and to minimize the influence of the error of the transmission code even if an error occurs in the transmission code due to a failure of the transmission line or the like. It is to provide a packet transmission device, a transmission / reception unit therefor, a packet transmission method, and a transmission / reception method thereof.

[0007]

In order to solve the above problems, in the first solution of the present invention, the packet transmission device is configured as follows. Prediction means for calculating and outputting a prediction signal value in the transmission section of the packet transmission device, subtraction means for subtracting the prediction signal value from the pulse code modulated signal input to the transmission section, and an output signal of the subtraction means Quantization /
Quantization / encoding means for encoding and outputting a code string that has been subjected to adaptive differential pulse code modulation, and packetization for dividing this code string into a packet of a predetermined length and adding control information to the packet for output. Means, a quantizing width coefficient is obtained from the code, and a quantizing width determining means for calculating and outputting a quantizing width for the next process from the quantizing width coefficient and the past quantizing width, and the quantizing width. Decoding means for obtaining a difference value from the quantization width output from the determining means and the quantized coefficient obtained from the code, and addition means for adding the difference value and the prediction signal value and outputting to the prediction means. And a prediction signal / quantization width adding means for adding the prediction signal value and the quantization width for the code at the head of the packet output by the packetizing means to the packet. ..

Further, the receiving unit of the packet transmission device temporarily stores the received packet, extracts a code from the packet and outputs the buffer, and decodes it from the quantization coefficient and the quantization width obtained from this code. Decoding means for obtaining a difference value by means of the above, quantization width determining means for obtaining a next quantization width from the quantization width coefficient obtained from the code and the quantization width, and the difference value and the prediction signal value are added. And a reproduction means for outputting a reproduction signal value, and a prediction means for obtaining a next prediction signal value from the reproduction signal value. The reception prediction signal value and the reception quantization width added in the packet from the buffer. And a predictive signal value / quantization width reading means for reading, and a reception quantization width and a quantization width output by the quantization width determining means are selectively switched and output to the quantization width determining means and the decoding means. First Selection means, characterized in that it comprises a second selecting means for outputting to said adding means is a switching select the prediction signal value output of the prediction means and the received prediction signal value.

In the second solution of the present invention, the packet transmission method is as follows. The packet transmission method includes an adaptive differential pulse code modulation coding step of coding a differential value between an input signal value and a predicted signal value while changing a quantization width according to a level change of an input signal, and the adaptive differential pulse code modulation. A packetizing step of dividing the coded sequence generated by the encoding step into a predetermined length and adding control information such as a transmission source address and a transmission destination address, and further including a packetizing step at the beginning of the code sequence stored in the packet. It is characterized by including a prediction signal value / quantization width adding step of adding a prediction signal value and a quantization width necessary for obtaining a reproduction signal value from the code to the packet.

Further, the packet receiving method uses a difference value obtained by decoding the code string extracted from the received packet,
An ADPCM decoding step of generating a prediction signal value and a quantization width from the code string, adding the prediction signal value and the quantization width to the difference value and outputting a reproduction signal value, and the reception prediction signal value and the reception quantization width added to the packet. And the reception prediction signal value and the reception quantization width for the head code of the code string stored in the packet, and the prediction signal value for codes other than the head code of the packet. And a prediction signal value / quantization width selection step of selecting a quantization width.

[0011]

On the packet transmitting side, the differential value between the input signal value and the predicted signal value is coded while changing the quantization width according to the level fluctuation of the input signal to perform adaptive differential pulse code modulation. This encoded coded sequence is divided into a predetermined length, control information such as source and destination addresses is added to the packet, and the code at the beginning of the code sequence stored in the packet is reproduced from that code. The predicted signal value and the quantization width required to obtain the value are added to the packet. After that, the packet is transmitted.

On the side receiving the packet, the code string extracted from the received packet is decoded to obtain the difference value. A prediction signal value and a quantization width are generated from the code string and added to the difference value to obtain a reproduction signal value. At this time, only when the head code of the packet is reproduced, the reproduction signal is obtained using the reception prediction signal value and the reception quantization width added to the packet on the transmitting side. Then, for the remaining code strings of the packet, the reception side reproduces the signal by using the predicted signal value and the quantization width obtained from the code string to minimize the influence of the error of the code received in the past. It works to limit it.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings.

1 and 2 are block diagrams showing an embodiment of the packet transmission apparatus of the present invention. FIG. 1 is a block diagram of a transmission section of the packet transmission apparatus, and FIG. 2 is a reception section of the packet transmission apparatus. It is a block diagram. Therefore, the transmission terminal 22 of FIG. 1 and the output terminal 72 of FIG. 2 are connected by a signal transmission medium (not shown).

In FIG. 1, the input signal value 31 from the input terminal 21 and the prediction signal value 36 which is the output signal of the prediction circuit 6 are shown.
[Yp (n)] is input to the subtraction circuit 11, and the pre-quantization difference value 32 [Dd which is the output of the subtraction circuit 11
(N)] (n represents the order of the time series, and n = 1, 2, ...
(..., the same applies to n below) is input to the quantization / encoding circuit 1. Further, the quantizing / encoding circuit 1 has a quantizing width 37 output from the quantizing width determining circuit 7.
[G (n)] is also input. This quantization / encoding circuit 1
The ADPCM code 33 [X (n)], which is the output of, is input to the packetization circuit 3, the decoding circuit 5, and the quantization width determination circuit 7. The output of the control information creation circuit 2 and the output of the timing control circuit 10 are input to the packetization circuit 3, and the output of the packetization circuit 3 is input to the prediction signal / quantization width addition circuit 4. The timing control signal 38 of the timing control circuit 10 is the prediction signal / quantization width adding circuit 4
2 is also input to the control information creation circuit 2, the output of the prediction signal / quantization width addition circuit 4 is output to the transmission terminal 22, and further via a signal transmission medium such as a communication cable (not shown) of FIG. It is output to the reception terminal 71. A quantization coefficient table 8 and a quantization width coefficient table 9 are connected to the decoding circuit 5 and the quantization width determination circuit 7, respectively. The difference value 34 which is the output of the decoding circuit 5
[D (n)] is the prediction signal value 3 which is the output of the prediction circuit 6.
6 [Yp (n)] and the reproduction signal value 35 [Yo (n)], which is the output of the addition circuit 12, are input to the prediction circuit 6.
Has been entered into. The quantization width 37 [G (n)] output from the quantization width determination circuit 7 is the same as the prediction signal / quantization width addition circuit 4, the decoding circuit 5, and the own quantization width determination circuit 7.
Have been entered respectively.

The operation of the transmitting unit having the above configuration will be described below.

Input signal value 3 input from the input terminal 21
1 [Yi (n)] is a signal such as voice information encoded by pulse code modulation (PCM), and the input signal value 31 [Yi (n)] is the subtraction circuit 11 and the prediction circuit 6
Predicted signal value generated based on the past signal in
The difference Yi (n) −Yp (n) from [Yp (n)] is obtained and output to the quantization / encoding circuit 1 as the pre-quantization difference value 32 [Dd (n)].

The quantizing / encoding circuit 1 uses the difference value 32 [Dd (n)] before quantization and the quantizing width 37 [G (n)] input from the quantizing width determining circuit 7 to Qd ( n) =
After calculating Dd (n) / G (n), this Qd (n) is quantized and encoded to produce an ADPCM code 33 [X (n)].
Output as. The value Q (X (n)) obtained by quantizing Qd (n) will be referred to as a quantization coefficient.

AD output from the quantization / encoding circuit 1
The PCM code 33 [X (n)] is input to the packetizing circuit 3, the decoding circuit 5 and the quantization width determining circuit 7, respectively. The decoding circuit 5 uses the ADPCM code 33 [X
(N)] and the quantization width 37 [G (n)] output from the quantization width determination circuit 7, the difference value 34 [D (n)].
Is output. The method is as follows: ADPCM code 33 [X
Based on (n)], the quantization coefficient table 8 is searched to obtain the quantization coefficient 41 [Q (X (n))]. Next, the quantization coefficient 41 [Q (X (n))] and the quantization width determination circuit 7
And the quantization value 37 [G (n)]
Find [D (n)].

As a result, the difference value 34 [D (n)] is the difference value 32 [Dd before quantization] which is the output of the subtraction circuit 11.
(N)] is quantized.

The difference value 34 output from the decoding circuit 5
[D (n)] is added to the prediction signal value 36 [Yp (n)] output from the prediction circuit 6 in the adder circuit 12 and becomes the reproduction signal value 35 [Yo (n)].

The prediction circuit 6 uses the reproduction signal value 35 [Yo
(N)], the predicted signal value 3 used in the next processing
6 [Yp (n + 1)] is calculated. Next predicted signal value 3
6 [Yp (n + 1)] is the reproduction signal value 35 [Yo
(N)] is multiplied by the prediction coefficient a (= 0.8).

The quantizing width determining circuit 7 is based on the ADPCM code 33 [X (n)] and the quantizing width 37 [G (n)] output from the quantizing / encoding circuit 1, and in the next processing. The quantization width 37 [G (n + 1)] to be used is calculated. The method is to search the quantum width coefficient table 9 based on the ADPCM code 33 [X (n)] and to quantize the width coefficient 42 [M (X
(N))] is calculated. Next, the quantization width coefficient 42 [M
Quantization width 37 [G (n + 1]] is obtained by multiplying (X (n))] by the quantization width 37 [G (n)].

The packetizing circuit 3 receives the input ADPC
The M code 33 [X (n)] is divided into a certain number (512 times, for example) and divided into blocks, and control information such as the source and destination addresses output by the control information generation circuit 2 is added to each block. Then, a packet is created and output to the prediction signal / quantization width adding circuit 4.

The prediction signal / quantization width adding circuit 4 includes a packet from the packetizing circuit 3, a prediction signal value 36 [Yp (n)] from the prediction circuit 6 and a quantization width from the quantization width determining circuit 7. 37 [G (n)] is input. The prediction signal / quantization width adding circuit 4 is the AD of the head (n = N at this time) of the code string blocked by the packetizing circuit 3.
Predicted signal value 36 for PCM code 33 [X (N)]
[Yp (N)] and the quantization width 37 [G (N)] are added to the packet and output to the transmission terminal 22.

The control information generation circuit 2, the packetization circuit 3, and the prediction signal / quantization width addition circuit 4 receive the timing control signal 38 from the timing control circuit 10 and synchronize so that the above-described processing is performed at a constant cycle. And be operated.

Next, the structure of the receiving section shown in FIG. 2 will be described. The buffer 5 to which the transmission signal is input from the reception terminal 71
1 and there is a reference numeral 81 which is the output of this buffer 51.
[X (n)] is a decoding circuit 52 and a quantization width determining circuit 53.
Entered in and. Further, the buffer 51 outputs the synchronization signal 90 to the timing control circuit 61 and the data 100 to the prediction signal value / quantization width reading circuit 57, respectively. A quantization coefficient table 54 and a quantization width coefficient table 55 are connected to the decoding circuit 52 and the quantization width determination circuit 53, respectively. The difference value 82 [D (n)] that is the output of the decoding circuit 52 is input to the addition circuit 60 together with the prediction signal value 86 [Y′p (n)] that is the output of the selection circuit 59. The reproduction signal value of this adding circuit 60 is 83 [Yo
(N)] is output to the output terminal 72 and the prediction circuit 56. The prediction signal value 84 [Yp that is the output of the prediction circuit 56
(N)] is input to the selection circuit 59. This selection circuit 5
A prediction signal value 85 [Yp (N)], which is the output of the prediction signal value / quantization width reading circuit 57, and the timing control signal 93 of the timing control circuit 61 are input to 9.

The quantization width 87 [G (n)] output from the quantization width determination circuit 53, the quantization width 88 [G (N)] of the prediction signal value / quantization width reading circuit 57, and the timing The timing control signal 93 of the control circuit 61 is input to the selection circuit 58. The quantization width 89 [G ′ (n)] output from the selection circuit 58 is input to the decoding circuit 52 and the quantization width determination circuit 53. The timing control signal 93 of the timing control circuit 61 is also input to the prediction signal value / quantization width reading circuit 57.

With respect to the above configuration, the operation of the receiving section will be described below.

The packet transmitted from the transmitting unit shown in FIG. 1 is input from the receiving terminal 71 to the buffer 51, and a plurality of packets are temporarily stored in this buffer 51.
Then, the buffer 51 outputs the code 81 [X (n)] from the packet in the input order to the decoding circuit 52 and the quantization width determining circuit 53. Further, at this time, if the leading code [X (N)] in the packet becomes available for output, the buffer 5
1 outputs the synchronization signal 90 to the timing control circuit 61. The timing control circuit 61 outputs the timing control signal 93 in synchronization with the synchronization signal 90 and outputs the predicted signal value / quantization width reading circuit 57, the selection circuit 58, and the selection circuit 59.
Control the synchronous behavior of.

Under the control of the timing control circuit 61, the prediction signal value / quantization width reading circuit 57 controls the prediction signal value 85 [added in the transmitting section from the packet output next by the buffer 51 by the timing control signal 93]. Yp
(N)] and the quantization width 88 [G (N)] are read out,
It outputs to the selection circuit 59 and the selection circuit 58, respectively.

Under the control of the timing control circuit 61, the selection circuit 58 selects either the quantization width 87 [G (n)] or the quantization width 88 [G (N)] by the timing control signal 93. The quantization width 89 [G '(n)] is output. That is, if n = N (N is the number of the leading code in the packet), the quantization width 88 [G (N)] added to the received packet is selected, and if n ≠ N, the quantization width is selected. 8
7 [G (n)] is selected.

Similarly, the selection circuit 59 also receives the control of the timing control circuit 61 and receives the predicted signal value 84 [Yp (N)] or the predicted signal value 85 [Y] by the timing control signal 93.
p (N)], the predicted signal value 86 [Y'p
(N)] is output to the adder circuit 60. That is, n = N
If so, the predicted signal value 85 [Yp (N)] is selected, and n ≠
In the case of N, the predicted signal value 84 [Yp (N)] is selected.

The buffer 51 sequentially reads the code 81 [X (n)] from the received packet and outputs it to the decoding circuit 52 and the quantization width determining circuit 53. In the decoding circuit 52, similarly to the decoding circuit 5 of the transmission unit, the code 81 [X (n)] supplied from the buffer 51 and the quantization width 89 [G ′ (n)] output from the selection circuit 58 are set. Base, difference value 8
2 [D (n)] is obtained and output.

The difference value 82 [D (n)] is added to the expected signal value [Y'p (n)] output from the selection circuit 59 by the adder circuit 60 as in the signal transmission section, and the reproduced signal value 83 [is obtained]. Y
o (n)] is output to the output terminal 72.

On the other hand, the predicting circuit 56, like the transmitting section,
Based on the reproduction signal value 83 [Yp (n)], the predicted signal value 84 [Yp (n + 1)] used in the next process is calculated. The predicted signal value 84 [Yp (n
+1)] is the reproduction signal value 83 [Yo (n)] multiplied by the prediction coefficient a (= 0.8).

The quantizing width determining circuit 53 also has a code 81 [X] supplied from the buffer 51, as in the transmitting section.
(N)] and the quantization width 89 [G ′ output from the selection circuit 58.
Quantization width 87 used in the next processing based on (n)]
Calculate [G (n + 1)].

The above-mentioned transmitter and receiver can be executed by a program of a digital signal processor (DSP). Therefore, the operation flow of the transmission processing and the reception processing of the second embodiment of the present invention by the DSP is shown in FIGS. 3 and 4, and the flow will be described below. An example of the information format in the packet is shown in FIG.

First, the operation flow of the transmitting DSP of FIG. 3 will be described. Predicted signal value Yp in block 200
(N) and the initial value of the quantization width G (n) are set. For example, when n = 1, Yp (1) = 0, G (1) = 1 and so on. Next, in block 201, the quantized difference value Dd (n) of the predicted signal value Yp (n) is obtained from the input signal Yi (n) pulse-code modulated. Qd at block 202
(N) = Dd (n) / G (n) is calculated, and this Qd
(N) is quantized / encoded to obtain an ADPCM code X (n). In block 203, the ADPCM code X (n) is sequentially stored in the packet. At block 204, the ADPCM
The quantization table is searched based on the code X (n) to obtain the quantization coefficient Q (X (n)). In block 205, the quantized difference value D (n) is obtained from the quantized coefficient Q (X (n)) and the quantized width G (n). At block 206, the difference value D (n) is added to the prediction signal value Yp (n) to obtain the reproduction signal value Yo (n). Playback signal value Yo in block 207
From (n), the predicted signal value Yp (n +
1) is asked. ADPCM code X in block 208
The quantization coefficient table is searched from (n) to obtain the quantization width coefficient M (X (n)). In block 209, the quantization width G (n + 1) to be used in the next process is obtained from the quantization width coefficient M (X (n)) and the quantization width G (n). Block 21
Check the value of n at 0, and if n = N, go to block 211, otherwise go to block 201 and input next signal Y
The ADPCM code X (n) of i (n + 1) is obtained.

In block 211, n in block 210
= N, the ADPCM code X
Since (n) is the head of the packet, the predicted signal value Yp
(N) and the quantization width G (N) are added to the packet. At the next block 212, the value of n is examined, and the value of n is N +.
When it is determined that the number becomes 511, the next block 213
Then, control information such as source and destination address data is added to the packet, and then the packet is sent to the receiving unit. If it is determined in this block 212 that n ≠ N + 511, the process returns to block 201,
ADPCM code X for the next input signal Yi (n)
Find (n).

Next, the operation flow of the receiving DSP will be described below with reference to FIG. In block 300, the predicted signal value Y′p (n) and the initial value of the quantization width G ′ (n) are set. For example, when n = 1, Y'p (1) = 0, G '
(1) = 1 and so on. In block 302, the ADPCM code X (n) is extracted from the transmitted packet, and the quantization table is searched based on the ADPCM code X (n) to obtain the quantization coefficient Q (X (n)). Block 3
In 03, the value of n is examined to determine whether n = N, that is, whether the taken out ADPCM code X (n) is the head of the packet. Then, if n = N, block 304
The predicted signal value Y'p (n) and the quantization width G '(n) are
Predicted signal value Yp added to the beginning of the packet by the transmitter
(N) and the quantization width G (N). If n ≠ N in the determination of block 303, block 305
Process is transferred to. In block 305, the difference value D (n) after quantization is quantized with a quantization coefficient Q (X (n)) and a quantization width G ′.
Calculate from (n). Predicted signal value Y'p in block 306
The difference value D (n) is added to (n) and the reproduction signal value Yo
Find (n).

Next, in block 307, the reproduction signal value Yo
From (n), the predicted signal value Yp (n +
1) is asked. ADPCM code X at block 308
The quantization coefficient table is searched from (n) to obtain the quantization width coefficient M (X (n)). In block 309, the quantization width G ′ (n + 1) to be used in the next process is obtained from the quantization width coefficient M (X (n)) and the quantization width G ′ (n). After the above processing, the value of n is checked in block 310, and n = N + 51
If it is 1, the process is completed, and if not, the process returns to block 301 and the next transmission ADPCM code X (n + 1) is sent.
Signal reproduction processing is performed.

In this way, the signal reproduction processing of the ADPCM code X (n) in the transmitted packet is sequentially performed.

Next, the information format in the packet will be briefly described. Box 401 to Box 4 in FIG.
04 is the predicted signal value Yp (N) added to the beginning of the packet
And the quantization width G (N), each box has a size of 8 bits. Further, the boxes 405 to 416 represent ADPCM codes X (n) obtained by quantizing / encoding the input signal value Yi (n) in the transmitting unit, and each box has a size of 4 bits. There is.

Although the number of packet codes X (n) in the above embodiment is 512, the number is not limited to this number.

[0046]

As described above, in the packet transmission device of the present invention, the prediction signal value and the quantization width corresponding to the leading ADPCM code are added to each packet, and the leading ADPC code is added.
When the M code X (n) is reproduced, since the predicted signal value and the quantization width added to the packet are used, the M code X (n) is reproduced without being affected by the error of the code X (n) received before the reproduction process. Further, since a new predicted signal value and a new quantization width are obtained from the reproduced signal, the predicted signal value and the quantization width generated in the receiving unit do not depend on all past codes. Therefore, even if a code error occurs due to a transmission path failure or the like, the reproduction signal value for the last code in the packet in which the error occurs in the reproduction signal value obtained based on the prediction signal value and the quantization width is sufficient. As a result, unlike the prior art, there is an effect that the deterioration of the quality of the transmission path can be suppressed to the minimum without causing all the reproduced signal values thereafter to be incorrect.

For example, if the ADPCM code in the packet is 5,
If there are twelve, the error in the reproduced signal value will be 512 at maximum.

[Brief description of drawings]

FIG. 1 is a diagram showing a block of a transmission unit according to a first embodiment of the present invention.

FIG. 2 is a diagram showing a block of a receiving unit according to the first embodiment of the present invention.

FIG. 3 is a transmission process flowchart of the second embodiment of the present invention.

FIG. 4 is a flowchart of a receiving process according to the second embodiment of the present invention.

FIG. 5 is a diagram showing an information format of a packet according to the present invention.

FIG. 6 is a conceptual block diagram of the entire packet transmission system of the present invention.

[Explanation of symbols]

 1 Quantization / Encoding Circuit 2 Control Information Creation Circuit 3 Packetization Circuit 4 Prediction Signal / Quantization Width Addition Circuit 5,52 Decoding Circuit 6,56 Prediction Circuit 7,53 Quantization Width Determining Circuit 8,54 Quantization Coefficient Table 9,55 Quantization width coefficient table 10,61 Timing control circuit 11 Subtraction circuit 12,60 Addition circuit 21 Input terminal 22 Transmission terminal 31 Input signal value 32 Pre-quantization difference value 33 ADPCM code 34,82 Difference value 35,83 Playback signal value 36, 84, 86 Prediction signal value 37, 87, 89 Quantization width 38, 93 Timing control signal 41, 91 Quantization coefficient 42, 92 Quantization width coefficient 51 Buffer 57 Prediction signal value / quantization width reading circuit 58, 59 selection circuit 71 reception terminal 72 output terminal 90 synchronization signal 500 transmission unit 600 reception unit

Claims (6)

[Claims] 1. Prediction means for calculating and outputting a prediction signal value,
Subtracting means for subtracting the predicted signal value from the pulse code modulated signal input to the transmission section, and quantizing for quantizing / encoding the output signal of the subtracting means to output a code string subjected to adaptive differential pulse code modulation. / Encoding means, packetizing means for dividing this code string into lengths of a predetermined value and packetizing it, and adding control information to the packet for output, and quantizing width coefficient from the code, Quantization width determination means for calculating and outputting the quantization width for the next process from the quantization width coefficient and the past quantization width, and the quantization width output from the quantization width determination means and the code. In a transmitter of a packet transmission device having a decoding means for obtaining a difference value from a quantized coefficient and an adding means for adding the difference value and the prediction signal value and outputting the result to the prediction means, the packetization means is Out A transmitting unit of a packet transmission device, comprising: a prediction signal / quantization width adding means for adding the prediction signal value and the quantization width for the code at the head of the input packet to the packet. 2. A buffer for temporarily storing a received packet, extracting a code from the packet and outputting the code, and a decoding means for decoding a quantized coefficient and a quantized width obtained from the code to obtain a difference value. A quantization width determining means for obtaining the next quantization width from the quantization width coefficient obtained from the code and the quantization width, and the difference value and the prediction signal value are added to output a reproduction signal value. In the receiving section of the packet transmission device having an adding means and a predicting means for obtaining a next predicted signal value from the reproduced signal value, the received predicted signal value and the reception quantization width added in the packet from the buffer are calculated. A predictive signal value / quantization width reading means for reading, a reception quantization width and a quantization width output by the quantization width determining means are selectively switched and output to the quantization width determining means and the decoding means. And a second selecting means for selectively switching the received predicted signal value and the predicted signal value which is the output of the predicting means to output to the adding means. Department. 3. A packet transmission device comprising the transmission unit of the packet transmission device according to claim 1 and the reception unit of the packet transmission device according to claim 2. 4. An adaptive differential pulse code modulation coding step for coding a differential value between an input signal value and a predicted signal value while changing a quantization width according to a level fluctuation of an input signal, and the adaptive differential pulse code modulation. A packet transmission method comprising a packetization step of dividing a coded sequence generated by the encoding step into a predetermined length and adding control information such as a transmission source address and a transmission destination address, the code sequence stored in the packet. A packet transmission method including a prediction signal value / quantization width adding step of adding to the packet a prediction signal value and a quantization width required to obtain a reproduction signal value for the first code of the packet. 5. A prediction signal value and a quantization width are generated from the code string to a difference value obtained by decoding a code string extracted from the received packet, added to the difference value, and a reproduction signal value is output. A packet receiving method including an ADPCM decoding step for reading the reception prediction signal value and the reception quantization width added to the packet, and the reception code for the head code of the code string stored in the packet. A predicted signal value / quantization width selection step for selecting a predicted signal value and a received quantization width, and selecting the predicted signal value and the quantization width for codes other than the head code of the packet. Packet reception method. 6. A packet transmission method, wherein a packet is transmitted by the packet transmission method according to claim 4 and the packet reception method according to claim 5.