JPH09200056A - Audio signal coding method and audio signal coder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a reversible boding system in which an audio signal is compression-coded with high efficiency and the coded signal is completely decoded. SOLUTION: A frequency are information generating section A applies band split to a digital audio signal and encodes the resulting signal by using the quantized sample value though bit distribution depending on an information amount based on the energy obtained from each band for information of a frequency region, a time region information generating section B applies inverse quantization to the sampled value after quantization fed from the frequency are information generating section A and encodes the received signal by using a residual signal between the original signal and the signal after band synthesis for re-conversion to the time region signal for the time region information. In this case, in order to limit the amplitude of the residual signal to be a prescribed amplitude or below, a pulse generated by a pulse generator 12 is added to the input signal by an input signal synthesis section 13 to apply compression- coding to the residual signal. Then reversible compression enable coding is conducted by using a multiplexer section 9 to multiplex the signal from the frequency area information generating section A.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a coding method and a coding apparatus for compressing a digital audio signal or the like, and more particularly to a digital audio signal (original signal) before coding.
The present invention relates to an audio signal coding method and an apparatus therefor capable of reproducing a decoded signal obtained by decompressing the compressed coded signal in a state without information loss (Loss Less).

[0002]

2. Description of the Related Art Generally, a sound signal simply digitally converted has some continuity or redundancy unless it is random data such as white noise, and some lossless compression methods (Loss Less compression method) is known. The fact that adjacent samples on this time axis have correlation (continuity) can be easily understood by observing the acoustic signal. In addition, the amplitude distribution of the acoustic signal generally has a bias, and the digitized signal is
There is regularity in the occurrence probability that each bit is "0" or "1". When this bias increases, the redundancy of the acoustic signal also increases.

The present inventor has used the time domain correction for the frequency conversion coding method, which has been conventionally considered unsuitable as the lossless coding method, and thereby efficiently and completely retains the reversibility of the acoustic signal. The encoding system and the decoding system thereof were applied for in Japanese Patent Application No. 7-2111220.

In the acoustic signal encoding device according to this application,
In order to use the deviation of the frequency distribution of the acoustic signal, the frequency domain information generator divides the digital acoustic signal into bands, and then based on the energy value obtained for each band, quantum distribution is performed by bit allocation according to the amount of information. The encoded sample value is encoded as frequency domain information, and the time domain information generation unit dequantizes the quantized sample value supplied from the frequency domain information generation unit and reconverts it into a time domain signal. In order to encode the residual signal of the signal after the band synthesis and the original signal as time domain information,
Reversible compression encoding was performed by multiplexing both of these in a multiplexing unit.

Then, in the acoustic signal decoding device, the supplied multiplexed signal is separated into frequency domain information and residual signal, and the frequency domain information is band-combined and then corrected by the residual signal. I was getting the same digital audio signal as the original signal.

Since the decoded signal completely matches the original signal, the band synthesizing unit in the encoding device and the band synthesizing unit in the decoding device have a common configuration, and the calculation accuracy and the rounding process are the same method. I was trying to take it. As a result, the difference value between the band-combined signal and the original signal in the encoding device is obtained, and this is encoded and transmitted as information in the time domain, so that the decoding device corrects the time domain in the signal after the band combination. The added signal and the decoded signal become exactly the same as the original signal.

The lossless compression coding method and the decoding method for multiplexing and coding the information of both the frequency domain and the time domain of the acoustic signal are the spectral distribution of the signal by treating the acoustic signal in the frequency domain. The redundancy caused by the deviation of can be effectively reduced regardless of the type of signal (audio sources of various genres), and the residual signal, which is a time domain signal, can be used for correction to ensure the original. It is possible to restore the signal. In addition, this method can perform completely lossless compression without strictly calculating the quantization accuracy of the frequency domain signal, as compared with the case of encoding in the frequency domain alone.

Although different from the reversible coding system, as an acoustic signal processing system incorporating a feedback structure, noise components such as mechanical vibration noise, for example, noise of an air conditioning fan or engine sound of an automobile is actively reduced. There are active noise control methods. These analyze the own signal (noise component) and cancel the acoustic signal treated as a noise component by superimposing the antiphase component on the original signal. A method incorporating this feedback structure is disclosed in, for example, Japanese Patent Application Laid-Open No. 5-80777.

Similarly, as an acoustic signal processing method incorporating a feedback structure, the difference between the decoded signal and the original signal is fed back to the original signal, and the encoding is repeated, thereby causing the quantization caused by the psychoacoustic model. A high-efficiency coding method aimed at reducing noise is disclosed in Japanese Patent Laid-Open No. 18501/1991.
No. 7.

[0010]

DISCLOSURE OF INVENTION Problems to be Solved by the Invention The lossless encoding method and the decoding method of the acoustic signal, which the present inventor applied for in Japanese Patent Application No. 7-211220, are grouped in encoding the residual signal in the time domain. In the means for determining the bit allocation for each, the bit allocation of the residual signal has been determined by using the minimum number of bits capable of expressing the maximum amplitude value of the samples in the group as a guide.

Here, the time domain residual signal to be encoded is caused by a calculation error at the time of band synthesis. According to the probability statistics theory, this calculation error is ± Δ / in one multiplication with a quantization step width Δ, as shown in FIG.
It is evenly distributed in the range of 2. When the addition of the data including the calculation error is repeated, the calculation error included in the final data generated has a normal distribution in the range of ± n × Δ / 2 as shown in FIG. Will be formed. This calculation error is independent of the input signal, and the total information amount of the residual signal is proportional to the total calculation amount. In informatics, there is a limit value for the coded information amount derived from Shannon's theorem. In this case, the limit value of the coded information amount in the time domain information coding is the total information amount of the residual signal. .

For example, if each sample in the group has substantially the same amplitude and the difference between the average amplitude value of the residual signal and the maximum allowable amplitude value determined by the bit allocation amount is small, the total encoded information amount Indicates that effective coding is performed by approaching the total information amount of the residual signal. on the other hand,
If there is a residual signal with a significantly larger amplitude than the other samples in the group (a group of residual signals with a large standard deviation), many other high-order bits for which this sample has no information to represent this signal. Since it must be owned, the amount of coded information is much larger than the total amount of information of the residual signal, and the amount of information in the time domain increases, resulting in inefficient coding.

As the number of additions increases as in the band synthesizing filter calculation, the residual signal group has such a large standard deviation that many high-order bits without information are held and the coding efficiency is lowered.

Further, the method of positively canceling the noise component by introducing the feedback structure is to remove the noise component which is annoying to the auditory sense, and is intended for a specific frequency band, which causes a calculation error. As described above, it is difficult to apply it to a noise signal that is evenly distributed in the band. Furthermore, it has not been possible to reduce the total amount of noise components themselves caused by the calculation error. Therefore, when the noise component due to the calculation error is encoded to transmit the acoustic signal, it is not possible to reduce the encoded information amount even if the conventional active noise control method is applied as it is.

Since the method of suppressing the quantization error caused by the psychoacoustic model by incorporating the feedback structure is the lossy compression, the calculation error amplitude value distribution which is the purpose of the present application (described later) is The intention is different from flattening. Moreover, this method uses the residual signal to adjust the original signal to prevent significant quantization errors. This method always has a quantization error as long as it uses a psychoacoustic model, and by dispersing this quantization error in a band that cannot be perceived by the sense of hearing, it is comparable to a CD (compact disc) that is not compressed by the psychoacoustic model. It is intended to provide an acoustic signal that Therefore, since the psychoacoustic model is improved by the feedback of the quantization error, Japanese Patent Application No. 7-21 which aims at the reversible encoding.
Even if it is used for the lossless encoding method of the audio signal No. 1220, it is difficult to improve the effect.

In the lossless coding method for acoustic signals of Japanese Patent Application No. 7-211220, in order to improve the coding efficiency, the average amplitude value of the intra-group residual signal is determined by the bit allocation of the block. It is preferable to reduce the bit allocation for each group closer to the maximum allowable amplitude value. That is, it is possible to improve the coding quality by utilizing the fact that the bit allocation can be reduced if the maximum amplitude value decreases even if the average amplitude value of the residual signals in each group is large.

Since the total amount of information of the residual component generated by the calculation error which is independent of the input signal is a certain amount of information according to the probability theory regardless of the type of the input signal, the present application It was not possible to reduce the amount of coded information even if the conventional feedback structure was adopted for the acoustic signal coding method that a person previously applied for.

Therefore, according to the present invention, in order to reduce the amount of information in the time domain, the residual signal is analyzed in order to average the residual signal for each block, and a pulse signal based on this is generated to generate the original signal. , And re-encoding,
An object of the present invention is to provide an audio signal encoding method that reduces the bit allocation in the time domain, has a small amount of encoded total information, and that maintains efficient and complete reversibility.

[0019]

As means for achieving the above object, a digital acoustic signal of frequency domain information obtained by band-dividing and encoding a digital acoustic signal of time domain information supplied as an original signal is provided. Step 1 and this first
Generating a residual signal as time-domain information between the digital acoustic signal obtained by band-synthesizing the digital acoustic signal of the frequency domain information encoded in the step of (1) and the digital acoustic signal of the time domain information supplied as the original signal. In the second step, a pulse signal for averaging the information amount of each sample is generated in the frame forming the residual signal of the time domain information coded in the second step, and the original signal is generated. And a third step of adding, wherein the residual signal and the digital acoustic signal of the frequency domain information are multiplexed and output, and an audio signal encoding method, and a time supplied as an original signal. An input signal synthesizing unit for temporarily recording the digital acoustic signal of the domain information, a band division filter for band-dividing the digital acoustic signal of the time domain information supplied from the input signal synthesizing unit, Quantizer for encoding the digital acoustic signal of the frequency domain information supplied from the band division filter, dequantizer for dequantizing the signal supplied from this quantizer, and dequantizer A band synthesizing filter for band-synthesizing the supplied signals to re-convert it into a signal of time domain information, and a signal of the time domain information supplied from this band synthesizing filter and a digital sound of the time domain information supplied as the original signal. And a residual calculation unit for generating a residual signal with the signal, and a pulse signal for averaging the information content of each sample in the frame constituting the residual signal supplied from the residual calculation unit. A pulse generator for outputting to the input signal synthesizer, and a multiplexer for multiplexing the residual signal supplied from the residual calculator and the signal supplied from the quantizer. Special It is intended to provide an acoustic signal encoding apparatus according to.

[0020]

BEST MODE FOR CARRYING OUT THE INVENTION The error component due to rounding in the calculation process cannot be predicted because it includes an uncertain factor. The error component in the calculation process is equal to or lower than the calculation precision in order to guarantee a certain predetermined calculation precision in the multiplication, assuming that the input data and the values of the coefficients required for the calculation do not include an error. It occurs when information is rounded up to the top. In the previous application, this calculation error was coded as correction information in the time domain, but this time domain portion was not information-compressed, and redundancy was left in this portion.

In order to efficiently code the information in the time domain, the amplitude values of the blocked intra-group samples may be averaged and the bit allocation may be performed in accordance with this average value. If the sample group of the residual signal in a certain block (frame) is the sample group shown in FIG. 13A, at least 7 bits are required to express the maximum value Max of this sample group. Therefore, the bit allocation of this residual signal group must be 7 bits, but if the Max value of this residual signal (hereinafter referred to as the Max signal) is reduced to a value that can be represented by 6 bits, M
The bit allocation of the group of residual signals including the ax signal can be reduced to 1 bit by setting 6 bits.

As a method of reducing the Max signal, a pulse signal is added to the position of the Max signal.
That is, in the present invention, the amplitude value of the Max signal is adjusted and the sign-inverted pulse signal is added to the original signal, and frequency conversion and inverse conversion are performed again. In this case, the output signal is as shown in FIG.
A new residual signal as shown in FIG. 13 (C) is output by superimposing the pulse signal shown in FIG. 13 (B) on the residual signal shown in FIG. 3 (A).

This is because when a pulse signal as shown in FIG. 14 (A) is expanded in the frequency domain and then further expanded in the time domain (depending on the frequency conversion method and calculation accuracy), FIG. 14 (B) As shown in FIG. 13, since a minute pulse signal (noise signal) caused by a calculation error due to dispersion or quantization appears slightly at the sampling points around the position of the pulse signal as shown in FIG.
13A is obtained by adding the pulse signal to the position of the Max signal of the residual signal shown in FIG. 13A and then performing frequency conversion and inverse conversion.
The residual signal shown in (C) is obtained. Since these noise signals shown in FIG. 14B are extremely small signals as compared with the pulse signals, they hardly affect the signals at other sampling points.

In the residual signal group (frame), not only the residual signal shown in FIG. 13 (A), in which only one sample is projected, but a plurality of samples are included in signals in other groups. There are cases where it is larger than that. However, in that case, if the pulse signal is superimposed on the Max signal a plurality of times, a more averaged residual signal can be obtained in the target group. In addition, the number of bit allocations that can be reduced from the information amount of the residual signal in the group is calculated in advance, and the original signal adjustment pulse group composed of a plurality of pulse signals is superposed to obtain the residual signal averaged at one time. May be obtained. In the embodiment described below, the case where the layers are overlapped a plurality of times has been described.

As described above, by grasping the residual signal in the time domain, generating an appropriate pulse signal, superimposing this on the original signal, and encoding again, the efficiency of the time domain information is improved. Information compression can be achieved. By applying this to the acoustic signal coding method previously filed by the applicant, the coding efficiency can be further improved.

[0026]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of an acoustic signal coding method and an acoustic signal coding device of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of an audio signal encoding apparatus of the present invention, and FIG. 6 is a flow chart showing its operation. An embodiment of the acoustic signal coding method will be described at the same time with reference to the drawings.

The acoustic signal coding apparatus shown in FIG. 1 includes a pulse generator 12, an input signal synthesizer 13, a frequency domain information generator A, a time domain information generator B and a multiplexer (multiplexer).
9. The frequency domain information generation unit A includes a band division filter 1, a maximum value selection unit 2, a bit allocation unit 3, and a quantizer 4, and the time domain information generation unit B includes an inverse quantizer 5 and a band. It is composed of a synthesis filter 6, a delay unit 7, and a residual calculation unit 8. Further, the input digital audio signal is formed of blocks in which a frame is a processing unit. In this embodiment, the acoustic signal forming one frame is 512 samples per channel, and the band division width is 32 bands.

The least significant bit precision is fixed to an arbitrary bit width, but here, fixed quantization precision of 16 bits is used, and information below this is rounded to the upper bit (the 16th bit which is the least significant bit). reflect. The bit allocation information in the frequency domain represents 4 bits to represent a code 0 to 16 bits, and the time domain bit allocation represents 3 bits to represent a code 0 to 8 bits.

Here, the bit allocation width in the time domain depends on the characteristics of the filter, and is left to the number of multiplications during the product-sum calculation for band synthesis. If the worst value of the accumulated error in this operation is within the time allocation bit width, the bit allocation width in the frequency domain guarantees the final quantization accuracy. In this embodiment, since encoding is performed with 3 bits, 16-bit quantization accuracy is guaranteed. If the bit allocation width is represented by 4 bits, it is possible to perform reversible compression of the input acoustic signal having a quantization accuracy of 24 bits. However, in this case, since the time domain information increases and the coding quality deteriorates, it is necessary to efficiently distribute the information amount with the frequency domain information.

A 4-bit frequency domain bit allocation table is shown in Table 1, and a 3-bit time domain bit allocation table is shown in Table 2. It should be noted that the code 0 indicates that there is no distribution, and if there is distribution, it is set between the code 2 and the code maximum value including the sign bit.

[0031]

[Table 1]

[0032]

[Table 2]

Next, the operation of the acoustic signal coding apparatus shown in FIG. 1 will be described. The input digital audio signal (original signal) is supplied to the delay device 7 described later and also to the input signal synthesis unit 13. Then, the input signal synthesizer 1
The original signal is temporarily recorded at 3 and then output to the band division filter 1. The band division filter 1 develops the input signal, which is time domain information, into frequency domain information (step 101). Here, the sub-band filter is equally divided into 32 bands, and the band width W thereof is set as in the following equation. The precision of the output subband data is 16 bits, which is the least significant bit precision as described above.

W = (sampling frequency × 0.5) / 32 (Hz)

As the band division filter 1 and the band synthesis filter 6 described later, for example, a filter for performing an orthogonal transformation such as DCT, a subband filter using the principle of the filter, a signal which is decomposed into a base waveform is used. The wavelet transform to be expressed, and further Fourier transform, which is a typical method of frequency transform, can be used. And
In the present invention, since the information component in the time domain is also used, it is not necessary to return the signal after the band division synthesis to the complete original signal, so any frequency conversion method may be used. In the present embodiment, a 512-tap subband filter (polyphase filter) will be used for concrete description. Note that the delay is generated by 480 samples through the band division synthesis.

The acoustic signal of the frequency domain information divided into 32 equal bands by the band division filter 1 is supplied to the maximum value selection unit 2 and the quantizer 4. In the maximum value selection unit 2, for each of 32 bands existing in one frame, 16 (5
The absolute value comparison of certain sub-band data (amplitude value) or energy value is performed (12/32), and the maximum value S is selected and output (step 102).

Then, the maximum value S of the sub-band data output from the maximum value selection unit 2 is supplied to the bit allocation unit 3. The bit allocation unit 3 refers to the maximum value S of each band in the frequency domain bit allocation table shown in Table 1 and
The number of bits allocated to each of the two bands is determined (step 103). The bit allocation here is performed corresponding to the minimum number of bits required to represent the maximum value sample in 2's complement notation, as shown in Table 1.

In the quantizer 4, based on the bit allocation information supplied from the bit allocation unit 3, the band division filter 1
The number of bits of each subband data of the acoustic signal of the frequency domain information supplied by the above is reduced (step 104). The reduction here is (16-the number of allocated bits) reductions of the upper bits that are the same as the sign bits, excluding the sign bits. Table 3 shows an example of reducing the number of bits of the subband data.

[0039]

[Table 3]

As shown in Table 3, when the sub-band data represented by a decimal number is 6, 31, 84, ..., 12, 54, the maximum absolute value of these is −94, and the bit allocation information is (Number of allocated bits) is 8 bits from Table 1. Therefore, of the 16-bit subband data, the data (8 bits) obtained by combining the lower 7 bits and the most significant bit (sign bit) as a sign bit is used as the quantized data, and the upper 2nd to 9th 8 bits are used. Is a reduction bit. This is a 1-bit sign bit part
It is equivalent to reducing the bit leaving.

Therefore, in the inverse quantization, it is only necessary to fill the reduced number of high-order bits with the designated number of sign bits according to the bit allocation information. The subband data and the bit allocation information quantized in this way are output to a multiplexer (multiplexing unit) 9 and, at the same time, inversely quantized in order to retransform the subband data into the time domain. Output to the container 5.

The inverse quantization in the inverse quantizer 5 is performed by adding the same code as the sign bit to the upper bits (16-the number of allocated bits) (step 10).
5). Therefore, in such quantization and dequantization, no constraint relating to the operation, such as rounding off, is performed during the operation, so that the quantization error does not occur.

The sub-band data aligned to the least significant bit precision (16 bits in this embodiment) by the inverse quantizer 5 is supplied to the band synthesizing filter 6 and converted into a signal of time domain information. It is converted (step 106).
In the band synthesizing process in the band synthesizing filter 6, it is necessary that the calculation precision, the filter coefficient precision, the calculating process, and the rounding process for the data of the output stage are completely matched with the band synthesizing filter 6 of the decoding device described later.

Generally, D is used for digital audio signal processing.
SP (Digital Signal Processor) is often used.
Therefore, the band synthesizing process in the band synthesizing filter 6 will be described by taking the DSP as an example. FIG. 3 shows the internal operation block configuration of the DSP used for the band synthesis filter 6. The DSP used here is, for example, a fixed point of 16 bits × 16 bits, and the internal calculation precision and the data width inside the memory are 16 bits.

In the band division synthesis with the sub-band filter or the like, the conversion is mainly performed by the sum of products operation. Therefore, the band synthesizing filter 6 is composed of a multiplier, an adder, and a register of an input / output stage, and is used as a memory for various coefficients and a memory for intermediate data (operation memory) and an input / output bus required in the arithmetic process. Connected.

In the figure, input data (subband data) and filter coefficient data corresponding to this input data are sequentially input to the multiplier 21 via a 16-bit input data bus. As the input data, the data output from the dequantizer 5 is stored in a memory (not shown) with 16-bit precision, and the necessary data is the multiplier 21.
It is supplied to us from time to time. Also, the filter coefficient data is
Similarly, it is stored in advance in 16-bit precision in various coefficient memories (not shown). Then, the multiplier 21 multiplies the input data by the filter coefficient data and outputs it as 32-bit data.

The output of the multiplier 21 is 32 bits + α
Is supplied to the adder 23. This α indicates a high-order extension bit. A 32-bit bus is used for the product-sum operation, and a 16-bit bus is used for other operations. Then, the output from the adder 23 is temporarily accumulated in the register 22, and the output of this register 22 and the output of the next multiplier 21 are added by the adder 23 and stored in the register 22 for the next calculation. The state of being accumulated is repeated until there is no input data.

When the primary product-sum operation is completed, the output from the adder 23 is supplied to the register 24 as it is as 32-bit precision data, rounded to 16-bit precision data, and output to an operation memory (not shown). To do. This operation memory stores 16-bit precision data that has been rounded from 32 bits and outputs it to the multiplier 21 for performing the quadratic product sum operation. And in the same way,
When the quadratic product sum operation is performed, 16-bit precision data (acoustic signal of time domain information) is output from the register 24 to the residual calculation unit 8 as data output.

Since the multiplier 21 outputs 32-bit precision data for each 16-bit precision data input, no arithmetic error occurs. Then, when the data is stored in each memory, rounding processing with 16-bit precision is performed. However, in the adder 23, an integer range (upper extension bits) of a sufficient number of bits is prepared for overflow or underflow during the product-sum operation. Has been secured. Therefore, the rounding process is performed only when storing it in the memory or when obtaining the final output value, and therefore, no extra calculation error is accumulated.

The band division synthesis operation as described above is performed, and the acoustic signal (post-decoding signal) of 16-bit precision time domain information output from the band synthesis filter 6 in FIG. 1 is input to the delay unit 7. The digital acoustic signal (original signal) that has been generated is supplied to the residual calculation unit 8 and the calculation error that has occurred in the band division synthesis in the band division filter 1 and the band synthesis filter 6 is calculated to obtain a residual signal. Output (Step 1
07). The residual signal is treated as a block of 24 samples here. Then, the minimum number of bits required to express the maximum value in this block is set as the block bit width, and this is encoded with 3 bits. Table 2 shows the relationship between the residual signal and the block bit width. In Table 2, the number shown as the bit allocation is the block bit width, and the numerical value is determined by the maximum value M of the absolute value of the residual signal in the block.

Since the time axis of the decoded signal that has passed through the band division filter 1 and the band synthesis filter 6 is delayed due to the delay peculiar to the filter, the original signal is passed through the delay unit 7 to the residual calculation unit 8. By supplying (Step 1
08), the time axis of the decoded signal is aligned.

The delay peculiar to this filter will be briefly described with reference to FIG. When a digital acoustic signal (original signal) of time domain information as shown in FIG. 3A is supplied to the band-splitting filter 1, the original signal is accumulated in the filter bank while shifting by 32 samples, and 32 subbands at any time. A sample is generated ((B) in the figure). Then, in the band synthesis filter 6, the sub-band samples are accumulated in the filter bank while shifting by 32 samples,
32 output signals (decoded signals converted to the time axis) are generated at any time ((C) in the same figure). At this time, the delay caused by a series of band division synthesis is 480 samples, and the decoded signal outputs the same data with a delay of 480 samples with respect to the original signal. Therefore, in the delay device 7, 48
The sample is delayed by 0 sample to absorb the delay peculiar to the filter ((D) in the figure), and the residual signal can be calculated in the residual calculation section 8.

The residual signal output from the residual calculator 8 is output to the pulse generator 12. Then, a pulse is output from the pulse generator 12 as needed,
The input signal synthesizer 13 adds the original signal (step 112). Here, the configuration of the pulse generator 12 is shown in FIG. 8 and will be described below. The residual signal output from the residual calculating unit 8 is supplied to the intra-group residual signal average information amount calculating unit 31 and the allocatable range residual signal detecting unit 33 in the pulse generator 12.

Then, the in-group residual signal average information amount calculating section 31 obtains an average amplitude value (residual signal average information amount) in a predetermined group and outputs it to the target bit allocation calculating section 32. . Note that, here, the average value is obtained based on the absolute value of the residual signal. Further, the actual amount of information includes code bits, and the calculated average amount of information is one bit smaller than the actual amount, so this must be taken into consideration when calculating the bit allocation value.

The target bit allocation calculator 32 performs a process for optimal bit allocation (step 110).
If all the samples of the individual residual signals can be averaged, then the minimum value of the allocated bits for representing the intra-group residual signal can represent the residual signal average information content supplied. It is the number of bits. However, in practice, the individual residual signal samples are very difficult to average because they result from arithmetic errors that have no regularity. Therefore, when the residual signal average information amount is close to the lower limit value of this allocated bit, a sample of the residual signal that cannot be represented by the allocated bit number may occur. Therefore, it is better to determine the target bit allocation value in a state in which some margin is included with respect to the minimum value of the allocation bit.

The determination of the target allocation bit will be described with reference to FIG. The actual residual signal average amplitude values shown by (a) and (b) in the figure are both represented by 7 bits. However, if the threshold value for calculating the target bit allocation value is set as shown in FIG.
In the residual signal block having the residual signal average amplitude value shown in (b), the target bit allocation is also 7 bits,
Since the residual signal block having the average residual signal amplitude value shown in (a) exceeds the threshold value for calculating the target bit allocation value, the target bit allocation is 8 bits. Note that this margin (threshold value for calculating the target bit allocation value) is set at the center of the minimum value and the maximum value that can be represented by the MSB bits in the allocation bits in FIG. It can be adjusted according to the quantization accuracy and the number of arithmetic processing steps.

Since the residual signal is averaged as the number of times of feedback is increased, the margin can be set lower and the number of allocated bits can be further reduced. However, an increase in the number of times of feedback increases the number of calculation steps, resulting in a decrease in processing speed. Therefore, in practice, based on the standard deviation of the calculation error caused by the frequency conversion method used, there is a tradeoff between coding efficiency and processing speed. The margin will be set.

The target bit allocation value output from the target bit allocation calculating unit 32 is the allocatable range residual signal detecting unit 3
3 is supplied. The allocatable range residual signal detection unit 33 selects a sample of the residual signal that cannot be represented by the allocated number of bits from the group of residual signals output from the residual calculation unit 8 and generates a pulse signal. It is output to the unit 34 (step 111 → Y). At this time, the sample number of the residual signal is output as a pulse signal generation sample point, and the difference between the amplitude value of the residual signal and the representable range by the target bit allocation value is simultaneously obtained and output. Note that the residual signal sample that can be represented by the number of allocated bits does not basically need to generate a pulse signal.
Although it does not have to be output to the pulse signal generation unit 34, when the pulse signal is very close to the threshold value for calculating the target bit allocation value, the target bit is added when the pulse signal is added to the sample of the residual signal in the neighborhood. Since the threshold for calculating the assigned value may be exceeded, the pulse signal generation unit 3 also in this case.
It is possible to reduce the number of times of returning by outputting to 4.

The pulse signal generator 34 determines the amplitude value of the pulse signal to be superimposed on the sample of the residual signal which requires the pulse signal, and outputs the pulse signal having the amplitude value to the input signal synthesizer 13. To do. At this time, a pulse signal is added to the sample of the residual signal having an amplitude value larger (or near the threshold value) for calculating the target bit allocation value to make the target bit allocation value. In order to do so, it is necessary to properly determine the amplitude value of the pulse signal.

Here, an example of a reference for determining the amplitude value of the pulse signal will be described with reference to FIG. The absolute value of the amplitude value of the pulse signal is the amplitude value of the residual signal minus the value of 3/4 of the representable range of the target bit allocation value, and the amplitude value with the opposite sign to the residual signal. To do. The value of 3/4 corresponds to the center of the value expressed by the target bit allocation value, and the noise signal (see FIG. 14) generated when the feedback is repeated affects the sample of other residual signals. This is the smallest value.

Then, in the input signal synthesizing section 13, the pulse signal sent from the pulse generator 12 is added to the sample of the specific residual signal of the temporarily recorded residual signal group, and the band division filter 1 is added. Output to. In the same manner, the process is repeated until there are no more residual signal samples to which the pulse signal is added, and when the amplitude values of all the residual signal samples become values represented by the target bit allocation value (step 111 → N), and outputs the signals from the quantizer 4 and the residual calculator 8 to the multiplexer 9.

It should be noted that the feedback causes a slight change in the frequency-converted signal, and the data output from the band division filter 1 is bit-allocated in the frequency domain initially determined by the bit-allocation unit 3. There is a possibility of exceeding. In this case, the bit allocation by the bit allocation unit 3 is not increased, but is replaced with the maximum value or the minimum value in the initially determined bit allocation. Then, the change in the frequency domain is limited to the sample value, and the bit allocation is performed to the end with the allocation calculated at the beginning. In this way, by fixing the amount of information in the frequency domain, it is possible to reduce time domain information without being affected by bit allocation in the frequency domain. However, conversely, when the bit allocation in the frequency domain decreases, the bit allocation in the bit allocation unit 3 may be changed to reduce the information amount in the frequency domain. In this case, more efficient encoding can be performed.

In the multiplexer 9, the sub-band data (digital acoustic signal of frequency domain information) with the reduced number of bits supplied from the quantizer 4 and the residual signal supplied from the residual calculator 8 are obtained. On the other hand, a frame synchronization word, various modes and auxiliary information, auxiliary information of frequency domain signals (frequency domain side information), further auxiliary information of time domain signals (time domain side information), etc. are added. Arrange as shown to generate a bitstream (step 10
9). Since the residual signal is included in the same frame after being multiplexed in this way, the amount of delay required for decoding by correcting the signal that has been band-synthesized and converted into the time domain at the time of decoding is reduced. be able to.

Next, FIG. 2 shows a block diagram of an embodiment of an acoustic signal decoding apparatus for decoding an acoustic signal encoded by the acoustic signal encoding apparatus of the present invention, and FIG. 7 shows an operation flowchart thereof. . The inverse quantizer 5 and the band synthesizing filter 6 shown in the acoustic signal coding apparatus of FIG. 1 and the inverse quantizer 5 and the band synthesizing filter 6 shown in FIG. ing.

The bit stream coded by the audio signal coding apparatus is supplied to the demultiplexer 10 to decode the synchronization word, the mode, the auxiliary information, etc., and further, the frequency domain signal and the time domain signal. Separated (step 20)
1). The signals in the frequency domain after separation are supplied to an inverse quantizer 5 which is a pre-processing for band synthesis. The signal in the time domain is supplied to the residual correction unit 11 to correct the signal that has passed through the band synthesis filter 6 later.

As described above, the inverse quantizer 5 is completely the same as the inverse quantizer 5 and the band synthesizing filter 6 used in the acoustic signal coding apparatus together with the band synthesizing filter 6, and its The operation is also the same. Also here, the same code as the sign bit is placed in the higher order (16-number of allocated bits).
Individually added and output (step 202). The signal output from the inverse quantizer 5 is supplied to the band synthesis filter 6, and the frequency domain signal is converted into the time domain signal (step 203). Since the process from the inverse quantizer 5 to the band synthesizing filter 6 is exactly the same as in the acoustic signal encoding device, the error with the original signal generated during the encoding process is the acoustic signal as a signal in the time domain. If corrected by the residual component supplied from the encoding device, the signal after band synthesis returns to the same signal as the original signal.

Therefore, the residual correction unit 11 adds the time domain correction signal (residual signal) supplied from the demultiplexer 10 to the time domain band synthesized signal supplied from the band synthesis filter 6. Thus, the original signal is restored (step 204). Since the delay involved in the band synthesizing process in the band synthesizing filter 6 is adjusted on the acoustic signal encoding device side, the residual signal in the time domain is the residual error correcting unit in the state where the time axis coincides with the post-band synthesizing signal. 12 are supplied. That is, the residual signal is generated from the signal that has passed through the band division filter 1 and the band synthesis filter 6 and the original signal delayed in accordance with this signal on the acoustic signal encoding device side. Although delayed by 480 samples with respect to the signal, since the signal in the frequency domain does not pass through the band synthesizing filter 6, the signal is multiplexed with a small delay amount. Therefore, on the acoustic signal decoding device side, the delay amount after the signal in the frequency domain has passed through the band synthesis filter 6 is the same as the residual signal, and the time axes match.

As a result, the digital audio signal (original signal) input to the audio signal encoding device of the present invention is transmitted or stored as a data-compressed signal as an encoded signal, and this audio signal decoding device It can be output as a decoded signal that completely matches the original signal.

[0069]

The acoustic signal coding method and the acoustic signal coding device of the present invention perform efficient compression coding by making use of the characteristics of the acoustic signal by the frequency domain information, and then carry out this frequency domain information. Since the residual signal, which is time domain information, is appropriately and minimally corrected, the original signal can be surely decoded. Further, in the present invention, since the residual signal which is time domain information is also compression-encoded, there is an effect that more efficient compression-encoding is possible.

[Brief description of drawings]

FIG. 1 is a configuration diagram showing an embodiment of an audio signal encoding device of the present invention.

FIG. 2 is a configuration diagram showing an embodiment of an audio signal decoding device.

FIG. 3 is a configuration diagram showing an internal operation block configuration of a DSP used for a band synthesis filter.

FIG. 4 is a diagram for explaining a delay amount specific to a filter.

FIG. 5 is a configuration diagram illustrating a generation example of a bitstream.

FIG. 6 is a flowchart showing an operation example of the acoustic signal encoding device of the present invention.

FIG. 7 is a flowchart showing an operation example of the acoustic signal decoding device.

FIG. 8 is a configuration diagram showing an embodiment of a pulse generator that constitutes the audio signal encoding device of the present invention.

FIG. 9 is a diagram for explaining a method of calculating a target bit allocation value.

FIG. 10 is a diagram for explaining determination of an amplitude value of a pulse signal.

FIG. 11 is a graph showing a probability distribution of N sample calculation errors.

FIG. 12 is a graph showing a probability distribution of calculation errors when the number of additions is n.

FIG. 13 is a graph for explaining the change in the residual signal due to the addition of the pulse signal.

FIG. 14 is a graph for explaining changes in a pulse signal after frequency conversion and inverse conversion.

[Explanation of symbols]

 1 band division filter 2 maximum value selection unit 3 bit allocation unit 4 quantizer 5 inverse quantizer 6 band synthesis filter 7 delay device 8 residual calculation unit 9 multiplexing unit (multiplexer) 10 demultiplexer 11 residual error Correction unit 12 Pulse generator 13 Input signal synthesis unit 31 Intra-group residual signal average information amount calculation unit 32 Target bit allocation calculation unit 33 Outside allocatable range residual signal detection unit 34 Pulse signal generation unit A Frequency domain information generation unit B Time domain information generator

Claims (6)

[Claims] 1. A first step of band-dividing and encoding a digital acoustic signal of time domain information supplied as an original signal to obtain a digital acoustic signal of frequency domain information, and encoding in the first step. A second step of generating, as time domain information, a residual signal between the digital sound signal obtained by band-synthesizing the digital sound signal of the frequency domain information and the digital sound signal of the time domain information supplied as the original signal; Third step of generating a pulse signal for averaging the information amount of each sample in a frame forming a residual signal of time domain information encoded in the second step and adding the pulse signal to the original signal An audio signal encoding method comprising: multiplexing the residual signal and the digital audio signal of the frequency domain information and outputting the multiplexed signal. 2. The residual signal is compressed to the number of bits of a target bit allocation value calculated from the average value of the information amount of each sample in a frame forming the residual signal. 1. The audio signal encoding method according to 1. 3. The first step to the third step are repeated until the maximum value of the information amount of each sample within a frame forming the residual signal becomes a predetermined value or less. Alternatively, the acoustic signal encoding method according to claim 2. 4. In the third step, a plurality of pulse signals for averaging the information amount of each sample are simultaneously generated in a frame forming the residual signal to form the residual signal. The acoustic signal coding method according to claim 1, wherein the information amount of the sample is set to a predetermined value or less. 5. An input signal synthesizing section for temporarily recording a digital acoustic signal of time domain information supplied as an original signal, and a band division for band-dividing the digital acoustic signal of time domain information supplied from this input signal synthesizing section. A filter, a quantizer for encoding the digital acoustic signal of the frequency domain information supplied by the band division filter, an inverse quantizer for inverse quantizing the signal supplied by the quantizer, and an inverse quantizer A band synthesizing filter for band-synthesizing the signals supplied from the quantizer and re-converting into a signal of time domain information, and a signal of the time domain information supplied from this band synthesizing filter and the time domain supplied as the original signal. A residual calculation unit for generating a residual signal with the digital audio signal of the information, and information of each sample in the frame constituting the residual signal supplied from the residual calculation unit. A pulse generator that generates a pulse signal for averaging the amount and outputs the pulse signal to the input signal combining unit, the residual signal supplied from the residual calculating unit, and the signal supplied from the quantizer. And a multiplexing unit that multiplexes the audio signal encoding apparatus. 6. An information amount equal to or more than a predetermined permissible information amount defined below a maximum representation value that can be represented by a target bit allocation value calculated from an average value of information amounts of respective samples in a frame forming the residual signal. The pulse generator generates a sign-inverted pulse signal having an amount of information for making the amount of information of this sample equal to or less than the predetermined permissible amount of information for the sample of the residual signal having The acoustic signal encoding device according to claim 5.