JP2652371B2 - Audio coding method - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a speech encoding method.

[Prior Art] For example, when a voice signal is transmitted through a high-speed digital line or when a voice signal is digitally processed such as storing and synthesizing a voice signal for a voice response device, the voice signal is converted into a digital signal by some method. Need to be converted to

Basically, an audio signal is an analog signal having a frequency band of 0.3 to 3.4 KHz. In order to convert this into a digital signal, for example, an analog / digital converter having a sampling frequency of 8 KHz and a resolution of 8 bits may be used. (PCM (P
ulse Code Modulation). Then, to return this digital signal to the original audio signal, the sampling frequency
What is necessary is to convert to an analog signal with a digital / analog converter having a resolution of 8 bits at 8 KHz, and to further shape the waveform through a low-pass filter. At this time, the higher the resolution of the analog / digital converter and the digital / analog converter (that is, the bit width of the PCM code), the higher the quality of the reproduced sound.

By the way, such a PCM-encoded audio signal is 1
The bit rate per second (data rate; hereinafter, referred to as bit rate) is 64 Kbps, and transmitting such a high bit rate audio signal requires a very high-speed transmission path, and accumulates such an audio signal. For this purpose, a memory having a huge storage capacity is required. Therefore, various proposals have conventionally been made for reducing the bit rate of the audio signal.

One of them is a differential PCM coding method for forming a difference between PCM codes adjacent in time series. This differential PCM coding method uses redundancy based on the correlation of the audio waveform, and since the value change between adjacent samples is often included in a limited dynamic range. The number of bits per sample can be reduced. Adaptive differential PCM which further advanced this differential PCM coding method
The adaptive differential PCM system, which is one of the encoding systems and recommended by CCITT (International Telegraph and Telephone Consultative Committee), achieves a bit rate of 32 Kbps.

In addition, APC-AB (Adaptive Prediction Coding with APC-AB) using non-locality and linear predictability of speech signals
daptive Bit Allocation) method or LSP (Line Spectrum Pair) method using a voice analysis / synthesis method.

However, such an adaptive difference PCM method, APC-AB
The scheme and the LSP scheme have a disadvantage that encoding and decoding processes are very complicated, and a device for realizing them is very expensive.

On the other hand, there is a quasi-instantaneous companding method as one of high-quality PCM audio transmission methods for broadcasting satellites. This semi-instantaneous companding method uses PCM
The coded audio data is divided into blocks of a predetermined number in a time series, and scale data representing the most significant digit corresponding to the maximum value of the signal absolute value in each block is identified. The bit number data is shaped into code data, and the encoding process is relatively simple, and the bit number of one sample can be easily reduced. However, such a quasi-instantaneous drawing method is not efficient enough.

Therefore, as a method of improving the efficiency of the quasi-instantaneous companding method, a `` combination of the differential PCM method and the quasi-instantaneous companding method '' can be considered. The missing bit at the time causes a transmission error, and the error is accumulated in the integrator on the receiving side to make reception impossible.

 Next, this point will be described.

Now, consider encoding a speech signal as shown in FIG. 11 (a) by an encoding method in which quasi-instantaneous companding is applied to the differential PCM method. First, the audio signal is sampled at a sampling frequency of, for example, 8 KHz for differential PCM encoding to form a difference value between samples. here,
The difference value between adjacent samples is represented by signed 8-bit data, that is, 8-bit data in 2's complement representation.
The condition for the quasi-instantaneous companding is that one block is composed of eight samples, and the transmission data is three bits per sample. The scale data is 3 bits.

Here, it is assumed that the difference values of the eight samples # 1 to # 8 are obtained as shown in FIG. In this block, the sample having the largest absolute value of the difference value is the sample ♯1, and therefore, the scale position at this time.
POS is bit 4 which is the most significant digit of the bit pattern of sample # 1, and the value of scale position POS is (100) ₂
become.

Thereby, the transmission bit (transmission data;
That is, the code data) is data of three bits from bit 5 (sign bit) representing the code value, that is, data of bits 5, 4, and 3, which is one digit higher than the scale position POS.

Therefore, in this block, first the scale position
The transmission data (code data) configured by POS followed by the transmission bits of samples # 1 to # 8 sequentially and sequentially arranged is as shown in FIG.

When decoding such code data, first, the code data for one block is decomposed into 3 bits, and the scale position POS is identified from the value of the first 3 bits. Then, when expanding the following 3-bit code data to 8-bit data, the MSB (most significant bit) of the code data is made to coincide with the one digit higher than the scale position POS, and each of the digits higher than the MSB is set. The value of the sign bit is set in the digit, and 0 is set in each digit lower than the LSB (least significant bit).

As a result, the data shown in FIG. 3C is obtained as decoded data. When this data is compared with the data before encoding, it is found that information (DC component) of lower digits than the transmission bit is missing in the decoded data (see FIG. 11 (b)). That is, missing bits are generated in the information.

When an audio signal is reproduced based on such decoded data, as shown by a dashed line in FIG. 11 (c), errors corresponding to missing bits are accumulated and a negative DC shift is generated. The waveform becomes lower right than the original waveform shown, and as a result, information cannot be transmitted properly.

One solution to such a problem is to use a differential companding PCM (DC-P
CM) ”(Takahashi et al., IEICE Transactions '84 / 10 Vol.J67
−B No. 10) has been proposed.

However, this method is effective when the difference data of about 15 bits is compressed to about 8 bits, and cannot be applied to a low bit rate coding method in which the difference data of 8 bits is compressed to about 3 bits.

In other words, in the case of such a low bit rate, the scale position may fluctuate greatly between blocks, for example, when the amplitude of the audio waveform greatly changes between blocks, and therefore, the accumulated error signal May be larger than the valid data to be transmitted. In such a case, data to be transmitted is buried in the error signal, and proper data transmission cannot be realized.

[Purpose] The present invention has been made in order to solve the above-described disadvantages of the related art, and has as its object to provide a speech encoding method capable of reproducing high-quality speech with a low bit rate and simple processing. The purpose is.

[Configuration] The present invention forms a difference value between adjacent samples in PCM-encoded audio data, and divides the difference value into a predetermined number of blocks in a time-series manner. Identifies the scale data representing the most significant digit corresponding to the maximum value of the absolute value of the difference value in the above, shapes the data of a predetermined number of bits including the most significant digit into code data, and compresses the audio data in the block unit The least significant bit of the difference value is 1
Is calculated, a second temporary difference value obtained by subtracting 1 from the least significant bit of the difference value is calculated, and first temporary code data based on the difference value is calculated, Calculating second temporary code data based on the temporary difference value and third temporary code data based on the second temporary difference value, and calculating a first temporary decoded value based on the first temporary code data;
A second temporary decoded value based on the second temporary code data and a third temporary decoded value based on the third temporary code data are calculated, and the first temporary decoded value and the second temporary decoded value are calculated. A decoded value and a temporary decoded value that takes a value closest to the value of the original signal in the sample among the third temporary decoded values are determined, and code data corresponding to the determined temporary decoded value is determined in the sample. In addition to the selection as the code data, the difference value of the first sample of the subsequent block is calculated based on the decoded value of the last sample of the block. Further, the difference value may limit a maximum value of the absolute value.

Also, a difference value between adjacent samples of the PCM-encoded audio data is formed with respect to the immediately preceding sample, and the difference value is divided in time series into blocks of a predetermined number, and the absolute value of the difference value in each block is calculated. An audio encoding method for identifying scale data representing the most significant digit corresponding to the maximum value, shaping data of a predetermined number of bits including the most significant digit into code data, and compressing the audio data in block units And forming a plurality of code data sequences obtained by shaping the code data based on the most significant digit and a plurality of scale data respectively corresponding to the digits adjacent to the most significant digit, and forming the plurality of code data sequences. And a first temporary difference value obtained by adding 1 to the least significant bit of the difference value, and 1 from the least significant bit of the difference value. Calculating a second temporary difference value, the first temporary code data based on the difference value, and a second temporary code data based on the first temporary difference value, the second
Calculates third temporary code data based on the temporary difference value of
A first temporary decoded value based on the first temporary code data, a second temporary decoded value based on the second temporary code data, and a third temporary decoded value based on the third temporary code data And calculating a temporary decoded value that takes a value closest to the value of the original signal in the sample among the first temporary decoded value, the second temporary decoded value, and the third temporary decoded value. And correcting the plurality of code data sequences so as to select code data corresponding to the determined temporary decoded value as code data in the sample,
The code data sequence having the smallest error in the block is selected as the code data sequence of the block from among the corrected code data sequences, and is selected in the block when calculating the difference value of the first sample of the subsequent block. This is based on the decoded value of the last sample of the encoded data sequence.

Further, the error power within each block of the code data sequence can be used as the intra-block error. Further, as the intra-block error, a sum of absolute values of errors in each sample in each block of the code data sequence can be used. Further, the difference value is:
It is advisable to limit the maximum value of the absolute value.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

FIG. 1 shows a speech encoding apparatus according to one embodiment of the present invention. In this embodiment, the quasi-instantaneous companding is applied to the differential PCM coding method, and the block-compressed differential data which is obtained when the quasi-instantaneous companding is performed is sequentially decoded and compared with the original signal. Is corrected for each sample point so as to obtain differential data having a small error within the number of compressed bits. Further, in this embodiment, an audio signal is sampled at a sampling frequency of 8 KHz, and a difference value between samples is represented by 8-bit data in a two's complement representation, and one block for quasi-instantaneous companding is constituted by eight samples. Transmission data is 1
3 bits per sample and 3 scale data
Bit.

In FIG. 1, an input audio signal SS is band-limited by a low-pass filter 1 and then applied to an analog / digital converter 2 to be converted into an 8-bit digital signal DS. This analog / digital converter 2 performs linear quantization at a sampling frequency of 8 KHz.

The digital signal DS is stored in a buffer 3 having a storage capacity of 8 samples forming one block.
Is added to the plus input terminal of an adder / subtracter 4 for forming difference data.

9-bit difference data output from the adder / subtracter 4
DDs is converted into 8-bit difference data DD by the maximum value limiting circuit 5. The reason for limiting the maximum value of the difference data in this way is as follows.

In other words, when the difference data DDs are directly compressed in the quasi-instant manner, a large scale value corresponding to the block containing the large difference data is suddenly set, and therefore, the data following the quasi-instantaneous compression follows other small difference data. The sex worsens. As a result, the restored audio signal gives an auditory jerky feeling. Therefore, by limiting the maximum value of the difference data to a certain value, such an auditory problem can be solved.

The difference data DD is applied to a buffer 6 having a storage capacity of 8 samples, a scale value setting unit 7 for setting a scale value for quasi-instantaneous compression, and one input terminal of an adder 8.

The output of the adder 8 is applied to a register 9. The output of the register 9 is applied to a minus input terminal of the adder / subtractor 4, another input terminal of the adder 8, and a buffer 10 having a storage capacity of 8 samples. ing.

In this way, the integrated value of the difference data DD limited to 8 bits by the maximum value limiting circuit 5 is formed by the adder 8, and this data is used as the data of the immediately preceding sample for forming the difference data DDs. ing.

The data stored in the buffer 6 is added to a quasi-instantaneous compression unit 11 that performs quasi-instantaneous compression encoding for each sample.

The scale value setting unit 7 identifies the one having the largest absolute value among the eight consecutive samples of the difference data DD output from the maximum value limiting circuit 5, determines the most significant digit of the bit pattern, and determines the bit. The position is output as 3-bit scale data DK. The scale data DK is the quasi-instantaneous compression unit 11 and the compressed difference data DC output from the quasi-instantaneous compression unit 11.
, An input terminal of a multiplexer 13 for shaping data of one block into a predetermined signal format, and for decompressing the optimized compressed difference data. Are added to the quasi-instantaneous extension unit 14 of the first embodiment.

The data stored in the buffer 10 is decoded data formed by integrating the difference data DD output from the register 9, and is used as an original audio signal to be encoded (hereinafter referred to as an original signal) by an optimization processing unit. Added to 12

The quasi-instantaneous compression unit 11 calculates the difference data DD of 8 bits added from the buffer 6 for each sample.
Extracts 3-bit data with the MSB as the first digit higher than the scale position represented by the scale data DK added from the scale value setting unit 7 and compresses it.
Output to the optimization processing unit 12 as DC.

The optimization processing unit 12 is configured to add 1
The compressed difference data DC for the block is sequentially decoded and buffered.
By comparing with the original signal added from 10, the compressed difference data DC is corrected for each sample point so that the difference data has a small error within the number of compressed bits, and the corrected difference data DCo is used as the optimized difference data DCo. It is added to the input terminal and the quasi-instantaneous extension unit 14.

As shown in FIG. 2, the multiplexer 13 arranges the scale data DK output from the scale value setting unit 7 at the head, and subsequently places the optimized difference data DC of each sample.
The signal constituted by sequentially arranging o is formed as encoded data DL for one block, and is output to the next-stage device (for example, a data transmission device or a data storage device).

The quasi-instantaneous decompression unit 14 places one digit higher than the digit indicated by the scale data DK added from the scale value setting unit 7.
The 3-bit optimized difference data DCo output from the optimization processing unit 12 is arranged so that the MSBs match,
The value of the code data of the optimized difference data DCo is arranged in the upper digit, and 0 is arranged in the lower digit to form 8-bit decoded data DE. Added to fifteen.

The integrated value 15 is obtained by multiplying the decoded data DE output from the quasi-instantaneous decompression unit 14 and forming the decoded data SD when the encoded data DL is actually decoded and restored.
SD is output to the register 9. The register 9 captures the restored data SD immediately before the processing of one block is completed and the processing of the next block is started.

As a result, the accumulation of errors due to missing bits specific to quasi-instantaneous companding can be eliminated when forming the first sample data of the next block, thereby forming more accurate encoded data DL. be able to.

As described above, in the present embodiment, the compressed difference data is controlled by the optimization processing unit 12 so as to follow the original audio signal.
In addition to correcting DC, the restored data formed by the integration unit 15
Since the accumulated error in the block is reflected in the next block by SD, it is possible to more accurately realize a low bit rate speech encoding process by quasi-instantaneous companding.

Next, the optimization processing of the compressed difference data executed by the optimization processing unit 12 will be described.

Now, consider encoding a speech waveform as shown in FIG. 3 (a). When each of the differences of samples # 1, # 2, and # 3 is formed by 8 bits with reference to sample # 0, sample # 1 has the largest absolute value of the difference in this case. Therefore, the 3-bit compressed difference data at this time is formed based on this sample # 1, and the scale value is the digit position of the most significant digit of the bit pattern.

Now, when the compressed difference data is represented by 3 bits,
The data that can be represented has a quantization width that is one bit lower than the scale value. Therefore, the value of each sample # 1, # 2, and # 3 is replaced with data that can be represented by this quantization width. You. For example, the compressed difference data of the sample # 1 is a value P12 (= (010) ₂ ; lower than the actual value P11 of the data that can be expressed, where LSB is one bit lower digit than the scale position; the same applies hereinafter). become.

By the way, of the data that can be expressed by this quantization width,
Data corresponding to the value P13 (= (011) ₂ ) which is one larger than P12 is closer to the actual value P11 of the sample # 1.
Therefore, if this value P13 is used as the compressed difference data of sample # 1, the error of the decoded audio signal (decoded value) can be reduced. That is, the error of the decoded value at this time can be suppressed to a half of the quantization width of the compressed difference data at the maximum.

Similarly, when considering samples # 2 and # 3, the decoded value is the value of the signal before encoding (the value P21,
In sample # 3, the compressed difference data closest to the value P31) may be selected.

That is, in this case, for sample # 2, the value P2
Since the decoded value based on the value P23 that is larger than the value P21 is closer to the value P21 than the decoded value based on the value P22 that is smaller than 1, the decoded values of the sample # 1 that are the values P13 and P23 are closer to the value P21. (= (110) ₂ ) is set as the compressed difference data. For sample # 3, since the value P31 matches the decoded value that can be represented by the compressed difference data, the difference (= (001) ₂ ) between the value P23, which is the decoded value of sample # 2, and the value P31 is represented by the compressed difference data. Set to.

In this way, it is possible to form compressed difference data with improved tracking of the original audio signal. FIGS. 4 (a) and 4 (b) show an example of an optimized difference bit routine which is a process for that.

First, the compressed difference data d (DC) is input from the quasi-instantaneous compression unit 11 (process 101), and the value of the compressed difference data d is a positive maximum that can be represented by the number of compressed bits (3 bits in this case). It is determined whether the value is larger than the value MAX (= (011) ₂ ) or smaller than the negative maximum value MIN (= (100) ₂ ) (decisions 102 and 103). If the result of the decision 102 is YES, the compressed difference data is determined. The value MAX is substituted for d (process 104), and when the result of the judgment 103 becomes YES, the value MI is added to the compressed difference data d.
N is substituted (process 105).

Next, a value dm smaller by LSB (= (001) ₂ ) than the compressed difference data d and a value dp larger by LSB than the compressed difference data d are formed (processing 106 and 107), and the value dm becomes smaller than the value MIN. , The value MIN is substituted for the value dm (decision 108, processing
109) If the value dp is larger than the value MAX, the value MAX is substituted for the value dp (decision 110, processing 111).

When the values dp, dm are formed in this way, 8-bit values dd, ddp, ddm when the values d, dp, dm are quasi-instantaneously expanded based on the scale data DK are calculated (process 112). And
The decoded values da ₀ of the data one sample before are added to the values dd, ddp, ddm, respectively, to form local decoded values da, dap, dam corresponding to the respective values d, dp, dm (process 113). Note has realized local decoding processing in the processing 112 and 113, in order that stores the decoded value da ₀ of the previous sample.

Next, the value dai of the original signal corresponding to the sample is read from the buffer 10, and the absolute values Da, Dp, Dm of the difference between the value dai of the original signal and each of the local decoded values da, dap, dam are calculated. (Process 114), the value dai of the original signal is
It is checked whether it is greater than (decision 115).

When the result of this determination 115 is YES, it is checked whether or not the absolute value Da is greater than the absolute value Dp (determination 116). When the result of this determination 116 is YES, the value dp is substituted for the value d ( Process 117) Return to process 106, and the result of judgment 116 is NO
, The value d is output as the optimized difference data DCo (process 118).

When the result of the determination 115 is NO, it is checked whether or not the absolute value Da is greater than the absolute value Dp (determination 119). When the result of the determination 119 is YES, the value dm is substituted for the value d ( Process 120) Return to process 106, and the result of judgment 119 is NO
When it becomes, the process 118 is executed.

That is, as shown on the left side of FIG. 3B, when the value dai of the original signal is closer to the decoded value dap than the decoded value da,
The compressed difference data d is updated to a value dp larger by LSB, and the value is output as optimized difference data DCo. Conversely, when the value dai of the original signal is closer to the decoded value dam than the decoded value da, the compressed difference data d is updated to a value dm smaller by LSB, and the value is output as the optimized difference data DCo. If an optimum value cannot be obtained in one process, this process is repeatedly executed.

In this way, the operation of adding and subtracting the LSB of the compressed bit to and from the compressed difference data obtained by the quasi-instantaneous companding is repeated for each sample point, so that the difference between the decoded value and the original signal is reduced. Has been corrected.

The data stored in the buffer 10 as an original signal may be output data of the buffer 3 before being input to the maximum value limiting circuit 5.

Therefore, for example, consider encoding a speech signal as shown in FIG. 5 (a) (same as FIG. 11 (a)) by quasi-instantaneous compression using such an optimized difference bit routine.

First, the difference values of the eight samples # 1 to # 8 of the audio signal are obtained as shown in FIG. 6 (a). Therefore, the scale position POS at this time is the most significant digit of the bit pattern of the sample # 1. And the value of the scale position POS is (100) ₂ . As a result, the compressed difference data DC formed by the quasi-instantaneous compression unit 11 is shown in FIG. 6 (b) for each of the samples # 1 to # 8.
It becomes what was shown in.

The compressed difference data DC is corrected to the optimized difference data DCo as shown in FIG. 6C for each of the samples # 1 to # 8 by the above-described optimized difference bit routine.

As a result, in the speech decoding device described later, the values of the samples # 1 to # 8 of the optimized difference data DCo are expanded to 8-bit difference data as shown in FIG. Based on the difference data, the audio signal is decoded as shown by the solid line in FIG. 5 (c). Here, the reproduced audio signal when the above-described compressed difference data DC is used as it is as the encoded data is shown by a dashed line in FIG. 5 (c).

As can be seen from these comparisons, the optimized difference data DC
The audio signal reproduced on the basis of o is clinged to the audio signal before encoding indicated by the broken line in FIG. 5 (c), and the followability to the audio signal before encoding is greatly improved. .

Note that, other than the above-described optimization difference bit routine, the above-described optimization processing of the compressed difference data can be realized. For example, after determining the scale value, for each sample, sequentially select the decoded value that is closest to the sampled value among the decoded values that can be taken with the quantization width corresponding to the scale value, and calculate the compression difference corresponding to the decoded value. By forming the data, the compressed difference data can be optimized.

FIG. 7 shows an example of a speech decoding apparatus according to one embodiment of the present invention. This audio decoding device decodes the encoded data DL encoded by the above-described audio encoding device and outputs an audio signal.

In the figure, coded data DL output from a preceding device (not shown) such as a data receiving device or a data storage device is applied to a demultiplexer 21, and for each block, the first three bits are scale values. It is identified as SC and is applied to the scale value input terminal of the quasi-instantaneous decompression unit 22, and the other code data (compression difference data) is applied to the code data input terminal of the quasi-instantaneous decompression unit 22.

The quasi-instantaneous decompression unit 22 distinguishes the added code data by 3 bits, arranges the 3-bit data at a bit position corresponding to the input scale data SC in the 8-bit data, and places the higher-order data above the code data. The contents of the sign bit are placed in the digits, and 0 is placed in the lower digits to expand the data into 8-bit data (see FIG. 6D).

The integrator 23 integrates the sequentially input 8-bit data to form a signal value in each sample of the audio signal, and outputs the signal value to the digital / analog converter 24.

The digital / analog converter 24 converts the received signal value to 8K
Analog signal (level signal) corresponding to the conversion frequency of Hz
And outputs this to the low-pass filter 25. The analog signal is shaped by the low-pass filter 25 and then output as a reproduced audio signal to a next-stage device (for example, an audio output device).

Thus, the configuration of the audio decoding device for decoding the encoded data according to the present invention is very simple. Therefore, for example, this speech decoding device can be realized using a general-purpose 8-bit microprocessor, and the cost can be kept extremely small.

By the way, in the quasi-instantaneous companding, since the scale position is set based on the bit pattern in which the absolute value of the signal value is maximum in one block, for example, a sample that suddenly takes a large value in that block In such a case, since a scale value corresponding to the large value is set, the ability of the encoded data to follow another small value is deteriorated, which may cause an inconvenience that it is not perceptible.

Next, another embodiment according to the present invention which can solve such a problem will be described.

Therefore, in order to solve the above-mentioned inconvenience, it is only necessary to form encoded data having the least error in the block, and for that purpose, an appropriate scale position (value) may be set.

Therefore, in the present embodiment, the quasi-instantaneous compression is performed based on the scale position when the quasi-instantaneous compression is performed based on the initially set scale position and when the digit one digit higher than the scale position is set as the scale position. When the quasi-instantaneous compression is performed on the basis of the scale position when one digit lower order than the first scale position is set as the scale position, the decoded value of each sample and the audio signal before encoding are An error with the value is calculated, and an evaluation value at each scale position is formed based on the error. Based on the evaluation value, the best one at those scale positions is selected. As the evaluation value, the sum of the absolute value of the difference between the original speech signal and the decoded value in each sample, the sum of the squared values of the difference, or the square root (error power) can be used.

That is, for example, when there is an audio signal as shown in FIG. 8 (a), each sample # 1 to # 8 of this audio signal
It is assumed that the scale value is determined based on the sample # 1 having the largest difference value among them, and the state is as shown in FIG.

On the other hand, when the scale value is reduced by one, the quantization width of the sample value is reduced by one step, and therefore, FIG.
When the scale value is increased by one, the quantization width of the sample value is increased by one step, so that FIG.
State.

Further, when the optimization process is performed by the above-described optimization difference bit routine in each state, the sign bit in each sample is as shown in the following table.

Here, SC ₀ represents the scale value in the case of FIG.
C _-1 is the scale value of the case of FIG one less than SC ₀ (b), SC ₁ shows the scale values in the case of FIG one greater than SC ₀ (C), respectively.

When the scale value is changed in this manner, naturally, the sum or error power of the difference between the original audio signal and the decoded value is changed to one of the scale values corresponding to the change in the audio signal in the block. The smallest one has the best followability to the audio signal in the block.

For example, as viewed in inventors statistically According to experiments, is approximately 60% of total blocks is the error power is the smallest of those of the scale value SC _0, the error power of one of the scale value SC _-1 is most the smaller is about 30% of total blocks, the error power of one of the scale value SC ₁ of the smallest is about 10% of total blocks.

In this way, by selecting the scale value to be used for each block, the sound quality of the reproduced audio signal (audible)
Is improved.

FIG. 9 shows an example of the speech encoding apparatus according to this embodiment. In this figure, the same parts as those in FIG. 1 are denoted by the same reference numerals, and the corresponding parts are denoted by the same reference numerals with suffixes, and the description thereof is omitted. This device also
Operates in units of one block of quasi-instantaneous compression.

In the figure, a scale value setting unit 7a determines a scale value in a block composed of continuous eight samples of difference data DD processed from the maximum value limiting circuit 5, and also sets a value larger by one than the scale value and 1 The next smaller value is formed and output as scale data DK ₀ , DK ₁ , DK _-1 respectively. These scale data DK ₀ , DK ₁ , DK _-1 correspond to the above-mentioned scale values SC ₀ , SC ₁ , SC _-1 respectively.

The scale data DK ₀ is converted to a quasi-instantaneous compression unit ₁₁₀ , an optimization process
12 ₀ are added to the quasi-instantaneous expansion unit 14 ₀ and single input terminal of the selector 31, the scale data DK ₁ is quasi-instantaneous compression unit 11 _1, the optimization processing unit 12 _1, quasi-instantaneous extension portion 14 of the _first and Sekukuta 31 The scale data DK- ₁ applied to the other input terminal is
11 _-1 , an optimization processing unit 12 _-1 , a quasi-instantaneous decompression unit _14-1, and a selector 31.

Quasi instantaneous compression unit 11 ₀ forms a compressed difference data DC ₀ from the output data of the buffer 6 on the basis of the scale data DK _0, the optimization processing unit 12 ₀ is the optimized differential bit routine described above to the compressed difference data DC ₀ Apply and optimize difference data DCo ₀
The optimized difference data DCo ₀ is generated by the quasi-instantaneous decompression unit 14.
₀ and one input terminal of the selector 32 are added. Quasi instantaneous expansion unit 14 ₀ is extended to optimize differential data DCo ₀ input to the data DE ₀ 8-bit data on the basis of the scale data DK _0, the data DE ₀ is decoded is accumulated by the integrator 15 ₀ , The decoded value SD ₀ is applied to one input terminal of the comparison unit 33 and one input terminal of the selector 34.

The quasi-instantaneous compression unit 11 ₁ forms compressed difference data DC ₁ from the output data of the buffer 6 based on the scale data DK ₁ , and the optimization processing unit 12 ₁ adds the above-mentioned optimized difference bit to the compressed difference data DC _1. by applying the routines to form an optimized difference data DCo _1, this optimization difference data DCo ₁ is applied to one input terminal of the quasi-instantaneous decompression unit 14 ₁ and the selector 32. Data D 8-bit data quasi instantaneous decompression unit 14 ₁ optimization difference data DCo ₁ input as a reference scale data DK ₁
Extends E _1, the data DE ₁ is decoded is accumulated by the integrator 15 _1, the decoded value SD ₁ is applied to one input terminal and one input terminal of the selector 34 of the comparator 33.

Similarly, the quasi-instantaneous compression unit 11 _-1 to form a compressed difference data DC _-1 from the output data of the buffer 6 on the basis of the scale data DK _-1, the optimization processing unit 12 _-1 to the compressed difference data DC _-1 forming an optimized differential data DCo _-1 by applying an optimization difference bit routine described above, the optimization difference data DCo _-1 is applied to one input terminal of the quasi-instantaneous expansion unit 14 ₁ and the selector 32. The quasi-instantaneous decompressor _14-1 receives the input optimized difference data
The DCo _-1 extends data DE _-1 of 8-bit data on the basis of the scale data DK _-1, the data DE _-1 integral part 15 _-1
And the decoded value SD- ₁ is compared by the comparator
One input terminal 33 and one input terminal of the selector 34 are added.

Also, the difference data DD output from the maximum value limiting circuit 5
Is stored in the buffer 10a, the output of the buffer 10a is sequentially integrated by the integrator 15a to form an uncompressed audio signal (that is, an original signal), and the data SDa corresponding thereto is subjected to optimization processing. Part 12 ₀ , 1
2 ₁ , 12 _-1 and the comparison unit 33.

As described above, the comparison unit 33 stores the data corresponding to the original signal.
SDa, scale value SC ₀ (scale data DK ₀₎ decoded value SD ₀ when decoding the optimization difference data DCo ₀ corresponding to the scale value SC ₁ (scale data DK ₁₎ optimizing the difference data corresponding to DCo ₁ The decoded value SD ₁ obtained when decoding is performed, and the decoded value SD ₋₁ obtained when the optimized difference data DCo ₋₁ corresponding to the scale value SC ₋₁ (scale data DK ₋₁ ) are decoded are obtained for each sample. Added.

Comparator 33, based on data SDa the decoded value _{SD 0, SD 1, SD -1} , to form their decoded value SD _0, SD _1, the SD _-1 error from the data SDa for each sample, The error power in one block is calculated for each of the decoded values SD ₀ , SD ₁ , SD ₋₁ , and the one that takes the minimum value is determined.

Then, when the error power for the decoded value SD ₀ becomes the minimum value, the selector 31 sets the scale data DK ₀
Is selected and output to the multiplexer 13, and the optimized difference data DCo ₀ is selected by the selector 32 and output to the multiplexer 13. The code value SD ₀ is selected by the selector 34 and is used as data to be taken into the register 9.

When the error power for the decoded value SD ₁ is minimized, the selector 31 selects the scale data DK ₁ and outputs it to the multiplexer 13, and the selector 32 selects the optimized difference data DCo ₁ and outputs it. The signal is output to the multiplexer 13. Further, the code value SD ₁ is selected by the selector 34 and is used as data to be taken into the register 9.

Similarly, when the error power of the decoded value SD- ₁ is minimized, the selector 31 selects the scale data DK- ₁ and outputs it to the multiplexer 13, and the selector 32 selects the optimized difference data DCo- ₁ . This is output to the multiplexer 13. The code value SD- ₁ is selected by the selector 34, and is used as data to be taken into the register 9.

Therefore, the coded data DL having the smallest error power in the block is output from the multiplexer 13.

FIG. 10 shows an example of the comparison routine executed by the comparison unit 33.

First, the data SDa and the decoded values SD ₀ , SD ₁ , and SD ₋₁ are input for each sample (processing 201), and the following expression is calculated based on the input data to obtain the decoded values SD ₀ and SD _1. , SD
Respectively calculates each corresponding error power RMS _0, RMS _1, RMS _-1 to _-1 (processing 202).

Here, k = 0, 1, -1 and j are sample numbers in the block, SDa _j is data in each sample in the block, and SD _kj is a code value SD _k in each sample in the block.

Then, it is determined which of the error powers RMS ₀ , RMS ₁ , RMS ₋₁ is the smallest (decisions 203, 204, 205), and the scale data DK ₀ , D corresponding to the minimum error powers RMS ₀ , RMS ₁ , RMS ₋₁
K ₁ , DK ₋₁ and optimized difference data DCo ₀ , DCo ₁ , DCo ₋₁ are selected (processing 206, 207, 208).

In this comparison routine, each decrypted value
As the evaluation value based on the error between SD ₀ , SD ₁ , SD _-1 and the data SDa, in addition to the above error power, for example, the data SDa and the decoded values SD ₀ , SD ₁ , SD _-1 in each sample are used. And the sum of the absolute values of the differences.

Also, the encoded data DL formed by this embodiment
The same audio decoding device as that shown in FIG. 7 can be used for decoding.

By the way, in each embodiment described above, the difference PC
Although the case where the quasi-instantaneous companding is applied to the M code has been described, the present invention can be similarly applied to the case where the quasi-instantaneous companding is applied to the PCM code.

It should be noted that the various constants in the above-described embodiments are merely examples for implementing the present invention, and the present invention is not limited to these.

[Effects] As described above, according to the present invention, a difference value between adjacent samples in PCM-encoded audio data is formed with respect to the immediately preceding sample, and the difference value is time-sequentially calculated for each predetermined number. By dividing the data into blocks, identifying scale data representing the most significant digit corresponding to the maximum absolute value of the difference value in each block, shaping data of a predetermined number of bits including the most significant digit into code data, In the audio encoding method for compressing audio data in block units, a first temporary difference value obtained by adding 1 to the least significant bit of the difference value and a second temporary difference value obtained by subtracting 1 from the least significant bit of the difference value
Are calculated, a first temporary code data based on the difference value, a second temporary code data based on the first temporary difference value, and a third temporary code data based on the second temporary difference value. Calculating temporary code data, based on the first temporary decode value based on the first temporary code data, the second temporary decode value based on the second temporary code data, and the third temporary code data; A third temporary decoded value is calculated, and a value closest to the value of the original signal in the sample among the first temporary decoded value, the second temporary decoded value, and the third temporary decoded value is calculated. The temporary decoding value to be taken is determined, the code data corresponding to the determined temporary decoding value is naturally selected as the code data in the sample, and the difference value of the first sample of the subsequent block is calculated when the difference value of the first sample of the subsequent block is calculated. sample Since the decoded value is used as a reference, a value closer to the original audio signal is selected as the compressed data by the quasi-instantaneous compression process at the time of encoding, and a high-quality sound can be reproduced at a low bit rate. The effect is obtained. Further, since the selection of the compressed data is performed at the stage of the encoding process on the encoder side, the decoding process on the decoder side is simplified, and as a result, the device configuration of the decoder can be simplified, and the The effect that the apparatus cost can be significantly reduced is also obtained.

Also, a difference value between adjacent samples of the PCM-encoded audio data is formed with respect to the immediately preceding sample, and the difference value is divided in time series into blocks of a predetermined number, and the absolute value of the difference value in each block is calculated. An audio encoding method for identifying scale data representing the most significant digit corresponding to the maximum value, shaping data of a predetermined number of bits including the most significant digit into code data, and compressing the audio data in block units And forming a plurality of code data sequences obtained by shaping the code data based on the most significant digit and a plurality of scale data respectively corresponding to the digits adjacent to the most significant digit, and forming the plurality of code data sequences. And a first temporary difference value obtained by adding 1 to the least significant bit of the difference value, and 1 from the least significant bit of the difference value. Calculating a second temporary difference value, the first temporary code data based on the difference value, and a second temporary code data based on the first temporary difference value, the second
Calculates third temporary code data based on the temporary difference value of
A first temporary decoded value based on the first temporary code data, a second temporary decoded value based on the second temporary code data, and a third temporary decoded value based on the third temporary code data And calculating a temporary decoding value that takes the value closest to the value of the original signal in the sample among the first temporary decoding value, the second temporary decoding value, and the third temporary decoding value. And correcting the plurality of code data sequences so as to select code data corresponding to the determined temporary decoded value as code data in the sample,
Among the corrected code data sequences, the code data sequence with the smallest error in the block is selected as the code data sequence of the block, and when calculating the difference value of the first sample of the subsequent block, the code data sequence is selected. Since the decoded value of the last sample of the coded data sequence is used as a reference, a coded data sequence with a smaller error from the original audio signal is selected as the compressed data by the quasi-instantaneous compression process during encoding. An effect is obtained that high-quality sound can be reproduced at a bit rate.
Also, at the stage of the encoding process on the encoder side, the selection of the compressed data is performed, so that the decoding process on the decoder side is simplified, and as a result, the device configuration of the decoder can be simplified,
The effect that the apparatus cost of the decoder can be significantly reduced is also obtained.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a speech encoding apparatus according to one embodiment of the present invention, FIG. 2 is a signal arrangement diagram illustrating a signal format of encoded data, and FIG. FIG. 4B is a waveform diagram for explaining the operation, and FIG. 4B is a signal arrangement diagram for explaining the operation of the optimized difference bit routine.
(B) is a flowchart showing an example of the optimization difference bit routine, FIG. 5 is a diagram for explaining the effect of the optimization processing, and FIG.
FIGS. 7A to 7D are signal arrangement diagrams showing the state of optimization;
FIG. 7 is a block diagram showing an example of a speech decoding apparatus according to an embodiment of the present invention. FIGS. 8 (a) to 8 (c) are illustrations of the principle of another embodiment of the present invention. FIG. 13 is a block diagram showing a speech encoding apparatus according to another embodiment of the present invention, and FIG.
FIG. 11 is a flowchart showing an example of the comparison routine, and FIG.
The figures are explanatory diagrams of the prior art, and FIGS. 12 (a) to 12 (c) are signal arrangement diagrams for explaining the state of encoding / decoding in the prior art. 1,25 low-pass filter, 2 analog / digital converter, 3, 6, 10, 10a buffer, 4 adder / subtractor, 5
…… Maximum value limiting circuit, 7,7a …… Scale value setting section, 8…
... adder, 9 ... register, 11,11 ₀ , 11 ₁ , 11 _-1 quasi-instantaneous compression unit, 12,12 ₀ , 12 ₁ , 12 _-1 ... optimization processing unit, 13 ...
Multiplexers, 14, 14 ₀ , 14 ₁ , 14 _-1 , 22... Quasi-instantaneous decompressor, 15, 15 ₀ , 15 ₁ , 15 _-1 , 15 a, 23... Integrator, 21. … Digital / analog converters 31, 32,
34 ... selector, 33 ... comparator.

Claims (6)

(57) [Claims] 1. A method of forming a difference value between an adjacent sample of PCM-encoded audio data and an immediately preceding sample, dividing the difference value into a predetermined number of blocks in a time series, and calculating a difference between each block. Identifies the scale data representing the most significant digit corresponding to the maximum value of the absolute value of the value, and forms a predetermined number of bits of data including the most significant digit into code data, and compresses the audio data in the block unit. In the encoding method, a first temporary difference value obtained by adding 1 to the least significant bit of the difference value and a second temporary difference value obtained by subtracting 1 from the least significant bit of the difference value are calculated. First temporary code data based on
Calculating second temporary code data based on the temporary difference value and third temporary code data based on the second temporary difference value, a first temporary decoded value based on the first temporary code data, Calculating a second temporary decoded value based on the second temporary code data and a third temporary decoded value based on the third temporary code data; and calculating the first temporary decoded value and the second temporary decoded value. A decoded value and a temporary decoded value that takes a value closest to the value of the original signal in the sample among the third temporary decoded values are determined, and code data corresponding to the determined temporary decoded value is determined in the sample. A speech encoding method which is selected as code data and uses a decoded value of a last sample of a subsequent block as a reference when calculating a difference value of a first sample of a subsequent block. 2. The speech encoding method according to claim 1, wherein a maximum value of an absolute value of said difference value is limited. 3. A PCM-encoded audio data, wherein a difference value between adjacent samples is formed between adjacent samples, and the difference value is divided into a predetermined number of blocks in a time series. Identifies the scale data representing the most significant digit corresponding to the maximum value of the absolute value of the value, and forms a predetermined number of bits of data including the most significant digit into code data, and compresses the audio data in the block unit. In the encoding method, a plurality of code data sequences are formed by shaping the code data based on the most significant digit and a plurality of scale data respectively corresponding to the digits adjacent to the most significant digit, and For the code data sequence, a first temporary difference value obtained by adding 1 to the least significant bit of the difference value, and a least significant bit of the difference value Is calculated, a first temporary code data based on the difference value, a second temporary code data based on the first temporary difference value, and a second temporary difference value are calculated. , And calculates a first temporary decoded value based on the first temporary code data, a second temporary decoded value based on the second temporary code data, and a third temporary decoded value based on the second temporary code data. Calculating a third temporary decoded value based on the temporary code data, and among the first temporary decoded value, the second temporary decoded value, and the third temporary decoded value, the value of the original signal in the sample; Is determined, and the code data corresponding to the determined temporary decode value is selected as code data in the sample, and the plurality of code data sequences are corrected. Corrected code data The code data sequence in which the error in the block is the smallest among the columns is selected as the code data sequence of the block, and when calculating the difference value of the first sample of the subsequent block, the last sample of the code data sequence selected in the block is used. A speech encoding method characterized by using a decoded value of a reference as a reference. 4. A speech encoding method according to claim 3, wherein said intra-block error is an error power in each block of the code data sequence. 5. A speech coding method according to claim 3, wherein said error in a block is a sum of absolute values of errors in each sample in a block of each code data sequence. . 6. The speech encoding method according to claim 3, wherein the maximum value of the absolute value of the difference value is limited.