JP2001211082A - Pcm signal encoding and decoding method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a PCM signal encoding and decoding method which can realize efficient compression to PCM signal, and also in the case of occurrence of encoding errors, can create a correction encoding signal over full range of an original signal. SOLUTION: In this method, a bit string which is a bit-by-bit reconfiguration of a multi-bit PCM signal is encoded in a way of variable length encoding having a structure capable of selecting the most suitable code book, so an efficient information compression being adequate to a stage of encoded signal. The higher-order bit of a coding string is preferentially multiplexed and protected so as to decode the original signal which is difficult to be decoded in the case of errors.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to an encoding method for compressing data of a PCM signal and a decoding method thereof. In particular, the present invention performs efficient data compression by reconstructing continuous PCM signals for each number of digits of the quantization bit, thereby imparting bias to the redundancy of the signal group, and performing decoding. PCM signal encoding method for generating a corrected decoded signal over the entire area of an original signal by using bits of higher priority (higher-order bits of the number of quantized bit digits) decoded before that when an error occurs. And a decoding method thereof.

[0002]

2. Description of the Related Art When a certain information source is reversibly compressed, the method is roughly divided into two types. One is entropy coding that models information sources into another code based on probability statistics so that they can be represented by short codes on average.The other is prediction using the correlation between samples of the information source. Differential encoding. Here, while predictive difference coding requires correlation between adjacent samples, entropy coding can compress the data of an information source if there is a bias in establishing the appearance. As long as the information source is not a noise component, entropy coding is also effective, for example, for a signal source that has been subjected to differential coding. In fact, in a complex information compression method, entropy coding is used in the last step. There are examples. In general, when information is compressed by a method such as entropy coding of a PCM signal, for example, a typical static Huffman coding, a dictionary is created from the appearance of samples, and a code is generated again using the dictionary. Encoding. In some cases, the establishment of the PCM signal itself is directly converted into a codebook, or when the quantization bit length is short, a plurality of samples are grouped.
Alternatively, if the quantization bit length is long, a codebook may be created after dividing the data into upper bit groups and lower bit groups. There is also a method in which a plurality of codebooks are prepared, the codebook with the highest compression rate is selected, and a code after encoding is transmitted together with the code number, thereby providing a high compression rate.

[0003]

SUMMARY OF THE INVENTION However, PCM
When a signal is subjected to variable-length coding as it is as described above, the code book increases in proportion to the size of the quantization width, resulting in an adverse effect such as an increase in hardware scale and a decrease in search speed. . Also, in the method of preparing a plurality of codebooks and selecting a code having the highest compression rate, if the bias in appearance is small, in other words, in a signal source having a large standard deviation, a code having a short code length is often used. It could not be selected, and as a result, hindered the improvement in compression ratio. When a bit error occurs, the decoded signal after the bit error has completely different properties from the original signal due to error propagation, and it is almost impossible to supplement the original signal. SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned problems, and has as its object the purpose of aligning bit strings having the same energy level (by reconstructing each digit of a bit) to generate a signal, in particular, of importance. Increase the stochastic concentration of high bits (increase the bias of the probability of occurrence), create an effective codebook, improve the information compression rate, and at least approximate the signal source to bit errors. To make it reproducible.

[0004]

The present invention provides the following (1) to (6) in order to solve the above-mentioned problems. (1) A PCM signal in which a plurality of samples each consisting of a plurality of bits are consecutively reconstructed into a bit sequence grouped by the number of quantization bit digits, and the bit sequence grouped by the number of quantization bit digits is converted into a predetermined number of bits. The code after being subdivided into is compressed and encoded by variable-length encoding, and an encoded sequence grouped by the number of quantization bit digits is generated,
A PCM signal encoding method, characterized in that coded sequences grouped by the number of quantization bit digits are multiplexed in accordance with the priority order of bit importance by adding quantization bit number information. (2) For a coded signal obtained by multiplexing coded sequences grouped according to the number of quantized bit digits in accordance with the priority order of bit importance, for each coded sequence grouped according to the number of bit digits The variable-length coding is solved to obtain bit strings grouped by the number of quantization bit digits, and grouped by the number of quantization bit digits according to the quantization bit number information described in the encoded signal. A PC that reconstructs a bit sequence into a bit sequence for each sample and that has multiple samples
A method for decoding a PCM signal, comprising decoding to an M signal. (3) A plurality of frames are formed by dividing a PCM signal in which a plurality of samples each consisting of a plurality of bits are connected for each of a predetermined number of samples, and a bit string grouped by the number of quantization bit digits for each frame is formed. A code obtained by subdividing the bit string grouped by the number of quantization bit digits into a predetermined number of bits is compression-encoded by variable-length coding to generate an encoded string grouped by the number of quantization bit digits. Then, for each frame, the coded sequence grouped by the number of quantization bit digits, multiplexed according to the priority order of the importance of the bit by adding the quantization bit number information,
A PCM signal encoding method, characterized in that a coded sequence multiplexed for each frame is connected to form a bit stream. (4) The bit stream generated by the PCM signal encoding method described in (3) is divided into a multiplexed coded sequence for each frame, and the coded signal of the multiplexed coded sequence for each frame is divided. On the other hand, the variable length coding is solved for each of the encoded sequences grouped by the number of bit digits, and a bit sequence grouped by the number of quantized bit digits is obtained, and the quantized bits described in the encoded signal are obtained. According to the digit number information, a bit string grouped by the quantization bit digit number for each frame is reconstructed into a bit string for each sample, and the obtained samples are connected to connect a plurality of samples.
A method for decoding a PCM signal, comprising decoding to an M signal. (5) In the PCM signal encoding method described in the above (3), before compression encoding of a bit string, all samples having one of positive and negative logics are positive or negative logic. A PCM signal encoding method, characterized in that all bits except for a sign bit indicating whether the bit is inverted are bit-inverted. (6) The bit stream generated by the PCM signal encoding method described in (5) is divided into a multiplexed coded sequence for each frame, and the coded signal of the multiplexed coded sequence for each frame is divided. The variable length coding is solved for each encoded sequence grouped by the number of bit digits to obtain a bit sequence grouped by the number of quantized bit digits, and the quantization described in the encoded signal is performed. According to the bit number information,
The bit strings grouped according to the number of quantization bit digits for each frame are reconstructed into bit strings for each sample, and each of the bit strings for each sample has a sign bit indicating logic designated to be bit-inverted during encoding. All bits except the sign bit of the bit sequence of the sample are bit-inverted, and the obtained samples are connected to decode a PCM signal in which a plurality of samples are connected.
CM signal decoding method.

[0005]

DESCRIPTION OF THE PREFERRED EMBODIMENTS PCM signal encoding method and PCM signal decoding method according to a first embodiment of the present invention, respectively.
A signal encoding device and a decoding device will be described with reference to the drawings.
The variable length coding is roughly classified into a dynamic method and a static method, but in this embodiment, the description will be made using the static coding to simplify the explanation. The encoding apparatus and the decoding apparatus have a plurality of common dictionaries created in advance, select an optimal codebook among them, and perform encoding and decoding by transmitting the information. Such an approach requires that the source of the signal be modeled in advance, but by providing the dictionary from scratch, the overhead of transmitting the codebook itself is eliminated, and the booklength is one.
It can be expected to speed up code search because it is shorter than two cases.

First, the PCM signal encoding device shown in FIG. 1 will be described. This PCM signal encoding device is a PCM data rearrangement control unit 2 which is a means for aligning a PCM signal group consisting of a plurality of bits into a bit string having the same energy level (a means for reconstructing into a bit string grouped by the number of quantization bit digits). , A bit-by-bit sorting memory 3,
A variable-length encoding unit 4 for applying a variable-length encoding to a code obtained by subdividing a newly reconstructed bit string into a predetermined number of bits and compressing the amount of information; Means for performing a multiplexing process for each bit digit in accordance with the important bit priority array (here, from the upper digit to the lower digit of the quantization bit digit) in order to prevent the (A means for performing multiplexing in the order of the number of digits).

FIG. 2 shows a bit arrangement of a multi-bit PCM signal which is an original signal coming into the apparatus shown in FIG. Here, the bit order is MSB first. Let the number of quantization bits be n bits and the total number of samples be m.

[0008] The PCM signal encoding device first decomposes the original signal into bits, and rearranges them according to the number of bit digits.
FIG. 3 shows the rearranged bit arrangement. Subsequently, variable length coding is performed on the newly generated bit string. After performing this for all the bit strings, the generated encoded data is output after adding codebook selection information and bit string end code or sample length information.

The bit-by-bit rearrangement memory 3 shown in FIG.
Has the same number of pointers as the number of quantization bits in the memory area. Assuming that the original signal shown in FIG. 2 is used, an area of at least (n × m) bits is secured in a certain space on the memory, and the positions of the n bit string pointers are specified in advance. The distance between the pointers is equal to or greater than the number of original signal samples to be encoded. n-bit PCM
Each sample of the signal is decomposed into n-bit data, and the data is written at the position of the pointer on each memory in the order of the samples. The pointer address is set to 1 each time data is written
Will increase. After the rearrangement of all samples is completed, data is sequentially read from the bit string on the MSB side, and variable length coding is performed.

In practice, the rearrangement of bits is undesirable because it takes a lot of man-hours, and the PCM data rearrangement control unit for generating addresses virtually arranges the bit array as shown in FIG. A bit string can be generated at high speed, and the input PCM memory 1 in which an input multi-bit PCM signal is stored can be used as it is. From the address specified by the PCM data rearrangement control unit, necessary bit strings are extracted by bit-shifting the bit mask and bit according to the order.

FIG. 4 shows an embodiment of an encoding apparatus using the input PCM memory 1 as a memory for rearranging bits. Memory 1 has n words and m words (ADR)
In this example, zigzag scanning is performed to read one bit at a time, thereby forming a virtual bit string according to the number of bit digits. 1 and 4 both have a configuration in which the variable length coding unit includes a reference codebook in this example.

FIG. 5 shows an example of the multiplexing method. The upper bit string is arranged with priority over the lower bit string. Quantized bit digit number information and the like are added to the header. Usually, it is considered that some distortion factor exists in a transmission path or a recording medium through which an encoded signal is passed or stored. Therefore, a serious error is prevented by providing a bit error detection word and an error correction code as additional information in at least the upper bit string encoded data.

FIG. 6 shows a conventional general multiplexing method for variable-length coding and a reproduced signal when an error occurs. The reproduced signal is represented by a continuous analog signal so that it can be easily grasped visually. In the conventional method (encoded data for each sample), if an error occurs at point A in the figure, the error propagates thereafter and it is impossible to decode the original signal. To avoid this, auxiliary information such as an error detection word and an error correction code covering the entire area is required. However, the compression rate of the encoded data is reduced by these pieces of auxiliary information.

On the other hand, in the encoding method of the present invention,
The important part of the signal source is arranged first (the coded sequence grouped by the number of quantization bits is multiplexed in accordance with the priority order of the importance of the bits, that is, multiplexed in order from the most significant bit). Auxiliary information such as an error detection word and an error correction code is added. Therefore, even if an error occurs at the point A, the approximate state of the original signal can be restored by the data of the high-order bit portion which has been decoded before and has high importance. Normally, the compression ratio of the variable-length code on the upper bit side with a small transition is better than the compression ratio on the lower bit side. Therefore, it is expected that the ratio of the variable-length code amount on the upper bit side in the entire encoded data is relatively small. This has the effect of reducing the amount of auxiliary information such as the error detection word and the error correction code in a proportional relationship.

When an error occurs at the same point A in the coded sequence of the present invention as in the conventional example, information before point A can be decoded, and several bits from the upper bits are distributed over the entire signal source. Can be reproduced. Although this restored signal includes a quantization error due to truncation of the lower bits of the original signal, it can reproduce the approximate shape of the original signal.
The lower-order bit data having an error to be complemented may be filled with zeros, or if the quantization noise component interferes with the restored data, the lower bit data may be complemented with a white noise component like dither.

Further, by setting the data area to be protected according to the state of the transmission path and the recording medium, it is possible to balance the compression ratio and the quality after decoding.

Next, the decoding device will be described. FIG. 8 shows the configuration of the decoding device. First, the header information is extracted by the header decoding unit 22 from the encoded data shown in FIG. Subsequently, the variable length decoding unit 23 performs variable length decoding in order from the upper bit coded sequence. A space corresponding to the number of samples is secured in the decoded signal memory 24 in advance, and as soon as decoding is completed, a bit string having a code length uniquely determined by a code book is written at a position indicated by the number of bit digits. This bit digit number is based on the quantized bit digit number information analyzed from the header information. The same processing is repeated until the least significant bit coded sequence is reproduced to the original number of quantization bits. The PCM data reproduction ADR control unit 25 reproduces a multi-bit PCM signal in which a plurality of samples are connected.

Here, consider a case where an error occurs in the encoded signal for some reason. As described above, it is assumed that the coded sequence of several bits from the upper bit is protected. When decoding a variable-length code normally, if the total number of samples is given, the coded data should be used up at the end of decoding, and if a data end code is given, this code must be used correctly for each bit string. It should be detectable. When a state other than the above occurs, or when an error is detected in the protection section, the processing shifts to an error process.

If the correction can be made by the error correction code in the protection section, the subsequent lower bit string is also decoded continuously. If the correction is impossible, the upper bit string decoded normally is output as a restoration signal, and the lower bits are filled with 0 or random data. If an error is recognized at the end of the restoration, the upper bit sequence up to the protection section is output as a restoration signal, and the lower bits are filled with 0 or random data. By employing such error processing, it is possible to provide a corrected restoration signal over the entire signal source even if an error occurs.

With the above-described method, it is possible to restore the original signal at the time of error, which is difficult with conventional variable-length coding, and it is optimal for each signal energy level (for each number of quantization bit digits). Information compression can be effectively performed by selecting a codebook and performing variable-length coding. Note that the error detection code and the error correction code multiplexed for each coded data of the bit string according to the number of quantization bit digits have an intensity weighted stepwise according to the importance of the bit string according to the quantization bit number. You may do so.

Next, a description will be given of a second embodiment of the PCM signal encoding method and its decoding method. In the encoding device according to the first embodiment, the required storage capacity of the bit-by-bit rearrangement memory for storing the bit sequence before encoding increases in proportion to the amount of information of the incoming PCM signal. The size of the storage capacity may directly lead to an increase in the cost of the entire apparatus and an increase in the circuit scale. In addition, the large amount of information can increase the size of the codebook itself, reduce search speed,
In some cases, it is difficult to follow the change in the signal, which may hinder codebook optimization.

Therefore, the second embodiment has been made to improve these points. The encoding apparatus using the second embodiment shown in FIG. 9 includes an input unit before the means (PCM data rearrangement control unit 2 and per-bit rearrangement memory 33) for reconstructing a bit string for each number of quantized bit digits. It is provided with a frame composing section 37 for forming a frame by dividing an incoming multi-bit PCM signal by a predetermined number of samples. This encoding device forms a bit string grouped by the number of quantization bit digits for each frame. And
Codes obtained by subdividing the bit strings grouped by the number of quantization bit digits into a predetermined number of bits are compression-encoded by variable-length coding to generate an encoded sequence grouped by the number of quantization bit digits. Then, for each frame,
The coded sequences grouped by the number of quantized bit digits are multiplexed according to the priority order of bit importance by adding the quantized bit digit information, and the coded sequences multiplexed for each frame are connected. A bit stream. FIG. 10 shows a configuration diagram of a bit stream generated by this encoding device.

The bit-by-bit rearrangement memory 33 in FIG.
May secure a capacity determined by a predetermined number of samples constituting one frame. (In the first embodiment, the total number of samples of the incoming multi-bit PCM signal is shown in FIG.
Requires a capacity of m samples. )

In addition, since the number of samples is limited, it is possible to select a more optimal codebook in the variable length encoder 4, and it is possible to flexibly cope with any signal source and to reduce the cost by reducing the storage capacity. And the circuit scale can be reduced. Also, the insertion of the synchronization signal or the block end signal (see FIG. 10) completes the encoded signal sequence, so that the influence of the encoding error can be kept in the frame.

The decoding method of the bit stream of the coded signal output from the coding apparatus is almost the same as that of the first embodiment except that the decoding is performed on a frame basis. That is, the bit stream is divided into multiplexed coded sequences for each frame. For the coded signal of the multiplexed coded sequence for each frame, variable-length coding is solved for each coded sequence grouped by the number of bit digits, and the bit sequence grouped by the number of quantized bit digits is obtained. obtain. Then, according to the quantization bit number information described in the coded signal, the bit strings grouped by the quantization bit number for each frame are reconstructed into bit strings for each sample. Then, the obtained samples are connected, and decoded into a PCM signal in which a plurality of samples are connected.

As described above, in the second embodiment, the circuit scale can be made smaller than in the first embodiment, and the influence of bit errors during decoding can be limited to a limited number of samples.

Next, a third embodiment of the PCM signal encoding method and its decoding method will be described. FIG. 13 is a diagram illustrating a PCM signal encoding device using the PCM signal encoding method according to the third embodiment. FIG. 14 is a diagram illustrating a PCM signal decoding device using the PCM signal decoding method according to the third embodiment. FIG. In the third embodiment, the original signal is a PCM having both positive and negative values.
This is an embodiment in the case of a signal.

When the PCM samples of the original signal as shown in FIG. 11 are variable-length coded according to the number of bit digits, even in the upper bit sequence, 0s and 1s appear with almost no bias,
It is expected that the appearance probability will be flat regardless of the number of bits for the code length. If the signal sources fluctuating in the positive and negative directions are simply grouped by the number of digits and subjected to variable length coding, the improvement in coding efficiency is small.

Therefore, as shown in FIG. 13, a positive / negative determining unit and a bit inverting unit 38 are provided before the variable length coding unit 4 to determine whether to be positive or negative before variable length coding. ) Of the sample is a sign bit (a bit indicating whether the signal is of positive or negative logic, and in the case of 2's complement expression, the MSB is a sign bit). The degree of concentration is increased to a code of 1. FIG. 12 shows an example in which a sample having a negative value is bit-inverted.

The compression ratio of the coded sequence generated by the variable length coding unit 4 is “0” by the above-mentioned bit inversion processing.
Significant improvement is achieved by increasing the probability of occurrence of codes.
In particular, the increase in the probability of appearance of “0” in the upper bit string of the second and third columns of the number of quantization bit digits is remarkable. With this method, the compression ratio of the small amplitude signal group is considerably improved.

The PCM signal encoding apparatus shown in FIG. 13 first forms a plurality of frames by dividing a PCM signal in which a plurality of samples each consisting of a plurality of bits are successively divided into a predetermined number of samples. For all samples having one of the positive and negative logics, all bits except the sign bit indicating which one is the positive or negative logic are bit-inverted. A bit string grouped by the number of quantization bit digits is formed for each frame. Then, a code obtained by subdividing the bit string grouped by the number of quantization bit digits into a predetermined number of bits is compression-coded by variable-length coding to generate an encoded sequence grouped by the number of quantization bit digits. I do. Thereafter, for each frame, the coded sequence grouped by the number of quantization bit digits is multiplexed according to the priority order of bit importance by adding quantization bit number information, and multiplexed for each frame. The coded sequences are connected to form a bit stream.

Next, the decoding apparatus shown in FIG. 14 will be described. First, header information is extracted by the header decoding unit 22 from the encoded data shown in FIG. Subsequently, variable length decoding is performed by the variable length decoding unit 23 from the upper bit coded sequence. Since the sign bit is the most significant bit, it is decoded first from the packed signal sequence according to the upper bit priority arrangement. The decoded sign bit is sent to a sign determining unit and a bit inverting unit 46 via the frame unit decoded signal memory 44.
If the sample has a logical bit designated to perform bit inversion on the lower bit string to be decoded next, the bit inversion process is performed after decoding. After that, the bit strings are rearranged in the same manner as in the apparatus shown in FIG. 8, and the original bit strings in sample units are reconstructed from the bit strings according to the number of quantization bit digits. Then, decoding to the original PCM signal via the output signal memory 47 is completed.

As described above, in the third embodiment, as in the second embodiment, the circuit scale can be made smaller than that in the first embodiment, and the influence of bit errors during decoding can be reduced by a limited number of samples. Keep in units. Further, the third embodiment can improve the data compression ratio by giving a bias to the appearance of codes even in signal sources having both positive and negative values.

[0034]

As described above, according to the present invention, a group is formed for a multi-bit PCM signal according to the energy (quantization bit digit number) of the bit, and the reconstructed bit sequence is group-by-group. By performing variable-length encoding having a structure in which an optimal codebook is selected, it is possible to perform effective information compression while grasping the properties of a signal to be encoded. Also, by preferentially multiplexing and protecting the upper bits, it is possible to restore the original signal that was difficult at the time of error. Further, in the case where a plurality of frames are formed by dividing each of a predetermined number of samples and processing is performed on a frame basis, the circuit scale can be reduced in addition to the above effects, and the influence of bit errors during decoding is limited. The number of samples taken. Also, for all samples that have one of the positive and negative logics,
When all bits except for the sign bit indicating positive or negative logic are processed by bit inversion, in addition to the above-described effects, a bias is imposed on the probability of appearance of the sign even in a signal source having both positive and negative values. Thereby, the data compression ratio can be further improved.

[Brief description of the drawings]

FIG. 1 illustrates a PM signal using a first embodiment of a PCM signal encoding method.
It is a block diagram which shows a CM signal encoding apparatus.

FIG. 2 is a bit arrangement diagram of a multi-bit PCM signal before encoding.

FIG. 3 is a bit arrangement diagram of a PCM signal divided and reconfigured according to energy rank (quantization bit digit number).

FIG. 4 is a diagram showing P using the first embodiment of the PCM signal encoding method;
FIG. 13 is a configuration diagram illustrating another example of a CM signal encoding device.

FIG. 5 is a configuration diagram of an encoded signal multiplexed according to the first embodiment of the PCM signal encoding method.

FIG. 6 is a configuration diagram of a coded signal multiplexed by a conventional variable length coder and a restored signal diagram when an error occurs.

FIG. 7 is a configuration diagram of an encoded signal multiplexed by the first embodiment of the PCM signal encoding method and a restored signal when an error occurs.

FIG. 8 is a diagram showing P using the first embodiment of the PCM signal decoding method;
It is a block diagram showing a CM signal decoding device.

FIG. 9 is a diagram showing P using a second embodiment of the PCM signal encoding method.
It is a block diagram which shows a CM signal encoding apparatus.

FIG. 10 is a configuration diagram of a bit stream generated by a second embodiment of the PCM signal encoding method.

FIG. 11 is a diagram showing a binary notation of a PCM signal as an original signal.

FIG. 12 is a diagram illustrating a binary notation of a PCM signal whose bits are inverted from the positive / negative determination result.

FIG. 13 is a configuration diagram showing a PCM signal encoding device using a third embodiment of the PCM signal encoding method.

FIG. 14 is a configuration diagram illustrating a PCM signal decoding apparatus using a third embodiment of the PCM signal decoding method.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 input PCM memory 2 PCM data rearrangement control unit 3 per-bit rearrangement memory 4 variable-length encoding unit 5 encoded signal memory 6 multiplexing unit

Claims (6)

[Claims] A PCM signal in which a plurality of samples each consisting of a plurality of bits are connected is reconstructed into a bit string grouped by the number of quantization bit digits, and a bit string grouped by the number of quantization bit digits is converted into a predetermined bit string. The code after being subdivided into the number of bits is compression-encoded by variable-length coding to generate an encoded sequence grouped by the number of quantized bit digits, and an encoded sequence grouped by the number of quantized bit digits Signal multiplexing in accordance with the priority order of bit importance by adding quantization bit number information. 2. A coded sequence obtained by multiplexing coded sequences grouped according to the number of quantized bit digits in accordance with the priority order of bit importance, and coding sequences coded according to the number of bit digits. Solve the variable length coding for each, obtain a bit sequence grouped by the number of quantization bit digits, and group by the number of quantization bit digits according to the quantization bit number information described in the encoded signal. A PCM signal decoding method, which reconstructs the obtained bit string into a bit string for each sample and decodes the bit string into a PCM signal in which a plurality of samples are connected. 3. A bit sequence formed by dividing a PCM signal in which a plurality of samples each consisting of a plurality of bits are consecutively divided into a predetermined number of samples to form a plurality of frames, and grouping each frame by quantization digit number. Is formed, and a code string obtained by subdividing the bit string grouped by the number of quantization bit digits into a predetermined number of bits is compression-coded by variable length coding, and the encoded string grouped by the number of quantization bit digits is formed. For each frame, the coded sequence grouped by the number of quantization bit digits is multiplexed according to the priority order of bit importance by adding quantization bit number information. A PCM signal encoding method characterized in that multiplexed encoded sequences are connected to form a bit stream. 4. A bit stream generated by the PCM signal encoding method according to claim 3 is divided into a multiplexed coded sequence for each frame, and the multiplexed coded sequence for each frame is encoded. The signal is subjected to variable-length coding for each of the encoded sequences grouped by the number of bit digits, to obtain a bit sequence grouped by the number of quantized bit digits, and the quantization described in the encoded signal. According to the bit number information, the bit string grouped by the quantization bit number for each frame is reconstructed into a bit string for each sample, and the obtained samples are connected to form a PCM signal in which a plurality of samples are connected. A PCM signal decoding method, characterized by decoding. 5. A PCM signal encoding method according to claim 3, wherein all samples having one of the positive and negative logics are subjected to either positive or negative logic at a stage before the compression encoding of the bit string. A PCM signal encoding method characterized in that all bits except a sign bit indicating whether there is a bit are inverted. 6. A method for dividing a bit stream generated by the PCM signal encoding method according to claim 5 into a multiplexed coded sequence for each frame, and encoding the multiplexed coded sequence for each frame. The signal is subjected to variable-length coding for each coded sequence grouped according to the number of bit digits, to obtain a bit sequence grouped according to the number of quantized bit digits, and a quantum sequence described in the coded signal is obtained. According to the digitized bit digit information, the bit strings grouped by the quantized bit digit number for each frame are reconstructed into bit strings for each sample, and bit inversion is designated at the time of encoding in the bit string for each sample. All bits except the sign bit of the bit string of the sample having the sign bit indicating logic are bit-inverted, and the obtained samples are connected to connect a plurality of samples. A PCM signal decoding method, which decodes a PCM signal into a decoded PCM signal.