JP2000349643A - Digital audio signal compression apparatus, digital audio signal compression method, and storage medium - Google Patents

Abstract

(57) [Summary] [PROBLEMS] To reduce the processing time required for compressing an audio signal. SOLUTION: In a digital audio signal compression apparatus for compressing a digital audio signal, a mask-to-noise ratio improvement index value is obtained in advance from a mask-to-noise ratio in each subband, and the mask-to-noise ratio improvement index value is obtained until the above-mentioned mask-to-noise ratio improvement index value is exceeded. By performing the mask-to-noise ratio calculation for each predetermined sub-band, the minimum mask-to-noise ratio calculation process can be omitted, and the time required for the calculation process can be reduced by omitting the minimum mask-to-noise ratio calculation process. Be able to shorten.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital audio signal compression apparatus, a digital audio signal compression method, and a storage medium, and more particularly to a bit allocation operation for compressing a digital audio signal.

[0002]

2. Description of the Related Art There is an MPEG audio compression method described in ISO / IEC1172-3 as a method for compressing a digital audio signal. FIG. 2 is a block diagram showing a schematic configuration of the MPEG audio compression device.

In FIG. 2, 1 is a digital audio input terminal, 2 is a sub-band encoding unit that divides a digital audio signal from the digital audio input terminal 1 into predetermined time intervals, and maps them to a plurality of frequency bands. An FFT operation unit that performs FFT analysis on the digital audio signal from the digital audio input terminal 1 during a period corresponding to the subband encoding unit 2.

[0004] Reference numeral 4 denotes a mask amount calculator for converting the result of the FFT calculator 3 into sound pressure and calculating a mask amount. Reference numeral 5 denotes a mask amount calculated by the mask amount calculator 4, which is assigned to each subband. A bit allocation operation unit for estimating the number of bits, and a data quantization unit 6 for quantizing a subband signal output from the subband encoding unit 2 in accordance with the number of bits allocated by the bit allocation operation unit 5 It is.

[0005] Reference numeral 7 denotes a transmission format forming unit for formatting the quantized data output from the data quantization unit 6 and the bit allocation information from the bit allocation calculation unit 5 into a predetermined transmission format. End 8.

The compression operation of the audio signal compression device will be described below. The data input to the digital audio input terminal 1 is input to the subband encoding unit 2 and the FFT operation unit 3.

First, the subband encoder 2 maps an input signal into band division data in units of 750 Hz at predetermined time intervals (referred to as audio frames) by an overlap polyphase filter bank (OPFB). . This data is data to be encoded later. On the other hand, the FFT operation unit 3 performs FFT analysis in order to analyze the frequency distribution of the sample data.

[0008] The result is input to the mask amount calculation unit 4. In the mask amount calculation unit 4, the result of the FFT analysis is converted into a sound pressure value [dB]. Then, from the calculated sound pressure value, a masker component that interferes with a sound pressure component on another frequency axis is extracted, and a mask amount L exerted on a nearby frequency position i is extracted.
Calculate Tm (i).

The above-described masker extraction and mask amount calculation processing are performed over the entire band, and finally the mask amount at the frequency position i is given by the sum of the static hearing limit value and the mask amounts by a plurality of maskers.

The mask amount calculated by the mask amount calculator 4 is output to the bit allocation calculator 5. The operation of the bit allocation calculation unit 5 will be described with reference to the flowchart of FIG. 7 and the conceptual diagram of FIG. Here, for the sake of simplicity, it is assumed that the total number of subbands is “8” and the number of bits Balc that can be allocated is 3000 bits, as shown in FIG.

FIG. 8 (a)
, The minimum value of the mask amount is detected for each sub-band,
The minimum mask amount LTmin [sb] is set (step S700 in FIG. 7). Here, the suffix sb is a function of the subband number.

The maximum value of the sub-band data is defined as the maximum sound pressure value L [sb], and the difference between the maximum sound pressure value L [sb] and the minimum mask amount LTmin [sb] is calculated as shown in FIG.
The signal-to-mask ratio shown in (b) is calculated (step S701 in FIG. 7).

Further, the signal-to-mask ratio [sb],
By taking the difference from the signal-to-noise ratio SNR (step), an MNR (mask-to-noise ratio) as shown in FIG. 8C is calculated (step S702 in FIG. 7). This SNR [st
ep] is a predetermined quantization step and SNR
FIG. 6 is a table quantitatively showing a relationship with (signal-to-noise ratio), and has a relationship as shown in FIG.

In the calculation of the MNR, no quantization step is applied, that is, since the step in FIG. 6 is calculated as "0", the MNR is substantially equal to -SM.
R [sb] (1) FIG. 8C shows an example of the above MNR value.
Shown in

When the MNR calculation is completed for all the subbands, the maximum allocable bit number Balc of the compressed sample data other than the header information permitted in the compression format is set (step S703 in FIG. 7).

This Balc is "30", based on the above assumption.
00 ”. Then, the MNR [sb] calculated earlier
Is searched for the minimum value (step S70 in FIG. 7).
4). Then, the search result is represented by MNRmin [SB].
And Further, the quantization step of the subband corresponding to MNRmin is changed to a value one step higher (step S705 in FIG. 7). This corresponds to reducing the quantization step.

For example, in the example of FIG. 8C, the minimum value is the MNR of the sub-band SB1, that is, 90 [dB].
Therefore, step [SB], which is currently “0”, is improved by one step, and the quantization step for the subband SB1 is reduced.

In response to the reduction in the quantization step, an assigned bit number Buse indicating the number of bits used by the quantization step is calculated (step S70 in FIG. 7).
6). Buse is a function of the quantization step and the number of bits used per subband, and as shown in FIG. 6, the number of samples per subband (36 samples)
And the number of allocated bits. Therefore,
In the example, 72 bits are allocated to SB1.

The number Ba of bits that can be allocated is
lc is compared with the number of allocated bits Buse (Step S707 in FIG. 7). If the Buse does not reach the Balc, it is determined that bit allocation is possible, and the buffer holds the bit allocation amount. Update the contents of ALC [SB] (Step S7 in FIG. 7)
08).

Then, a new SNR [step [S
B]] value to calculate the mask-to-noise ratio again,
[SB] The value is updated (step S709 in FIG. 7).

The hatched portion in FIG. 8C indicates the SNR [step
[SB]], and the MNR value of SB1 is the difference. That is, by making the quantization step one step higher than that of FIG. 6 as “2”, the SNR becomes “7 dB”.
As a result, a new MNR “83 dB” is derived from the SNR (7 dB) one signal-to-mask ratio (90 db). Thereafter, the process returns to step S703 in FIG. 7, and a loop process for searching for the minimum MNR is performed.

In the example shown in FIG. 8, SB-8 "-8"
7 dB "is determined as MNRmin, and the processing described above is repeated. Then, step S707 in FIG.
When it is determined that bit allocation is impossible, that is, when the number of allocable bits Balc <the number of allocated bits Buse, the loop processing ends, and the bit allocation calculation processing ends.

In the example of FIG. 8, as shown in (d), when the bit allocation ALC and NMR are determined, the Buse
Becomes "2988 bits" and cannot be allocated any more, and the bit allocation amount calculation ends in step S707 in FIG.

The bit allocation operation processing unit 5
Based on ALC [sb], which is a buffer holding the bit allocation amount resulting from
Then, the sub-band data is quantized, and the transmission format forming unit 7 formats the quantized data together with side information such as bit allocation information in a predetermined style, and outputs it to the compressed data output terminal 8. Such a compression process
Generally, it is realized by software processing using a DSP or the like.

[0025]

However, according to the above-mentioned prior art, when calculating the bit allocation amount, every time the quantization step of one sub-band is reduced by one step, the minimum MNR is reduced from all the sub-bands. Was searched again.

For example, ISO / IEC1172-3, a so-called MPEG audio compression layer 2,384
In kbps, approximately 250 quantization steps can be assigned to all the target subbands. Therefore, when this search process requires 300 steps, the search process alone requires a “75k process”.
When a DSP operating at 0 MHz is used, “24 m
A processing time of about "2.5 ms" was required for the audio frame "s".

Therefore, conventionally, there has been a problem that it takes an enormous amount of processing time to compress the audio signal. SUMMARY OF THE INVENTION The present invention has been made in view of the above-described problems, and has as its object to reduce the processing time required for compressing an audio signal.

[0028]

A digital audio signal compressor according to the present invention is a digital audio signal compressor for compressing a digital audio signal, wherein a mask-to-noise ratio improvement index value is obtained from a mask-to-noise ratio in each subband. The bit allocation process is performed for each sub-band until the calculated mask-to-noise ratio improvement index value is exceeded. Another feature of the digital audio signal compression device of the present invention is that input means for inputting a digital audio signal,
Orthogonal transform means for orthogonally transforming the digital audio signal input via the input means; and a mask amount calculating means for calculating a mask amount based on a sound masking effect from a power spectrum as a result of the orthogonal transform means. The ratio between the power spectrum or the mapped signal obtained by mapping the digital audio signal on the frequency axis by a method different from that of the orthogonal transform means and the mask amount obtained by the mask amount calculating means is predetermined. Signal-to-mask ratio calculating means for calculating in units of frequency width, signal-to-noise ratio defining means for defining an S / N ratio for a predetermined bit allocation amount, and signal-to-noise ratio each time the predetermined bit allocation amount is changed Updating the result of the defining means and calculating the mask-to-noise ratio with the result of the signal-to-mask ratio calculating means; A digital audio signal compression apparatus comprising: a return operation unit; and a bit allocation unit that determines a bit allocation number of compressed data in the predetermined frequency width unit based on a result of the repetition operation unit. A bit allocation target value setting means for setting a common predetermined bit allocation target value for each unit of width, and selecting a value at which a result of the signal-to-noise ratio defining means is a minimum value larger than the bit allocation target value. And a bit allocation operation means for performing a bit allocation operation using the result of the selection means. Another feature of the digital audio signal compression device of the present invention is that input means for inputting a digital audio signal and orthogonally transform the input digital audio signal into a predetermined period unit of an audio frame. Orthogonal transform means, and a mask amount for calculating a mask amount for each subband, which is a predetermined frequency band unit, based on a sound masking effect from a power spectrum of the digital audio signal obtained by the orthogonal transform means Calculating means for calculating the difference between the power spectrum or the mapped signal obtained by mapping the digital audio signal on the frequency axis by a method different from the orthogonal transform means and the mask amount obtained by the mask amount calculating means; Calculate each audio frame above Signal-to-mask ratio calculating means for obtaining a signal-to-mask ratio in the memory, signal-to-noise ratio storing means for storing a signal-to-noise ratio for a predetermined quantization step as a table, and the signal-to-noise ratio storing means Mask-to-noise ratio calculating means for performing a difference calculation between the signal-to-noise ratio, which is the output of the signal, and the signal-to-mask ratio obtained by the signal-to-mask ratio calculating means, for each subband to obtain a mask-to-noise ratio; The mask-to-noise ratio obtained by the mask-to-noise ratio calculation means, the signal-to-noise ratio, and the bits of the compressed data for each sub-band using bits that can be assigned to the audio frame during audio signal compression. In a digital audio signal compression device having bit allocation means for deciding the number of allocations, the digital audio signal compression means A mask-to-noise ratio average value calculating means for calculating an average value of the mask-to-noise ratio in a predetermined sub-band, wherein the bit allocating means includes the mask-to-noise ratio and the mask-to-noise ratio for each of the predetermined sub-bands. It is characterized in that the number of bits of compressed data for each subband is determined by using a ratio average value, the signal-to-noise ratio, and bits that can be allocated to the audio frame during audio signal compression. Another feature of the digital audio signal compression apparatus of the present invention is that, for one of the sub-bands, the mask-to-noise ratio calculating means performs the quantization step until the mask-to-noise ratio average value is exceeded. It is characterized by having a repetitive mask-to-noise ratio calculating means for updating the mask-to-noise ratio so as to make it finer, and thereafter repeating the same calculation for the other sub-bands. Another feature of the digital audio signal compression apparatus of the present invention is that, for one of the subbands, the mask-to-noise ratio calculating means sets the signal-to-noise ratio exceeding the mask-to-noise ratio average value. There is provided a signal-to-noise ratio search means for searching using the signal-to-noise ratio search means, and thereafter, the same operation is repeated for other subbands. Another feature of the digital audio signal compression apparatus of the present invention is that, for the predetermined subband, after the mask-to-noise ratio exceeds the mask-to-noise ratio average value, the mask for each of the subbands The number of bits of the compressed data for each sub-band is determined by using the noise-to-noise ratio, the average value of the mask-to-noise ratio, the signal-to-noise ratio, and the bits that can be allocated to the audio frame during audio signal compression. It is characterized by the decision. Another feature of the digital audio signal compression device of the present invention is that input means for inputting a digital audio signal and orthogonally transform the input digital audio signal into a predetermined period unit of an audio frame. Orthogonal transform means, and a mask amount for calculating a mask amount for each subband, which is a predetermined frequency band unit, based on a sound masking effect from a power spectrum of the digital audio signal obtained by the orthogonal transform means Calculating means for calculating the difference between the power spectrum or the mapped signal obtained by mapping the digital audio signal on the frequency axis by a method different from the orthogonal transform means and the mask amount obtained by the mask amount calculating means; Calculated for each audio frame Signal-to-mask ratio calculating means for obtaining a signal-to-mask ratio in the memory, signal-to-noise ratio storing means for storing a signal-to-noise ratio for a predetermined quantization step as a table, and the signal-to-noise ratio storing means Mask-to-noise ratio calculating means for performing a difference calculation between the signal-to-noise ratio, which is the output of the signal, and the signal-to-mask ratio obtained by the signal-to-mask ratio calculating means, for each subband to obtain a mask-to-noise ratio;
Bit allocation of compressed data for each sub-band by using the mask-to-noise ratio and the signal-to-noise ratio obtained by the mask-to-noise ratio calculating means, and the bits that can be allocated to the audio frame during audio signal compression In a digital audio signal compression apparatus having bit allocation means for determining the number, the number of bits that can be allocated to the audio frame, the quantization step, and the relationship between the signal-to-noise ratio and the relationship between the signal-to-noise ratio in the audio compression in advance Mask-to-noise ratio improvement amount calculating means for calculating an assigned mask-to-noise ratio improvement amount;
In the sub-band, a mask-to-noise ratio sum calculating means for calculating a total value of the mask-to-noise ratio obtained by the mask-to-noise ratio calculating means, and a mask pair which is a result of the mask-to-noise ratio improvement amount calculating means. A target value calculating means for calculating the sum of the noise ratio improvement amount and the mask-to-noise ratio total obtained by the mask-to-noise ratio total calculating means, and dividing by the predetermined number of sub-bands to obtain a bit allocation target value. It is characterized by having. Another feature of the digital audio signal compression apparatus of the present invention is that, for one of the sub-bands, the mask-to-noise ratio calculating means sets the signal-to-noise ratio exceeding the bit allocation target value to the signal. It is characterized by having signal-to-noise ratio search means for searching using the noise-to-noise ratio search means, and thereafter repeating the same calculation for other subbands. Another feature of the digital audio signal compression apparatus of the present invention is that the mask-to-noise ratio calculation means makes the quantization step finer for one of the sub-bands until the bit allocation target value is exceeded. And a repetitive mask-to-noise ratio calculating means for updating the mask-to-noise ratio so that the same calculation is repeated for the other subbands. Another feature of the digital audio signal compression device of the present invention is that, for the predetermined subband, after the mask-to-noise ratio exceeds the bit allocation target value, The mask-to-noise ratio, the mask-to-noise ratio average value, the signal-to-noise ratio, and the number of bits to be assigned to the compressed data for each sub-band are determined by using bits that can be assigned to the audio frame during audio signal compression. It is characterized by: Another feature of the digital audio signal compression device of the present invention is that the bit allocation means performs bit allocation for all subbands requiring bit allocation.

According to the digital audio signal compression method of the present invention, in the digital audio signal compression method for compressing a digital audio signal, a mask-to-noise ratio improvement index value is obtained from a mask-to-noise ratio in each subband, and the obtained mask pair It is characterized in that bit allocation processing is performed for each of the above subbands until the noise ratio improvement index value is exceeded. Further, other features of the digital audio signal compression method of the present invention include an input process of inputting a digital audio signal, an orthogonal transform process of orthogonally transforming the input digital audio signal, and an orthogonal transform process of the orthogonal transform process. A mask amount calculation process for calculating a mask amount based on a sound masking effect from a power spectrum as a result, and the power spectrum or the digital audio signal is mapped to a frequency axis by a method different from the orthogonal transformation process. A signal-to-mask ratio calculation process for calculating the ratio between any of the mapping signals and the mask amount obtained by the mask amount calculation process in a predetermined frequency width unit, and defining an SN ratio for a predetermined bit allocation amount Signal-to-noise ratio definition processing, and every time the predetermined bit allocation amount is changed. Updating the result of the signal-to-noise ratio defining process and performing a mask-to-noise ratio calculation with the result of the signal-to-mask ratio calculating process; and In a digital audio signal compression method for performing a bit allocation process for determining a bit allocation number of compressed data in a unit of a number of bits, a bit allocation for setting a predetermined predetermined bit allocation target value common to the unit of a predetermined frequency width A target value setting process, a selection process of selecting a value at which a result of the signal-to-noise ratio definition process is a minimum value larger than a bit allocation target value obtained by the bit allocation target value setting process, It is characterized in that a bit allocation operation is performed using the result. Other features of the digital audio signal compression method of the present invention include an input process of inputting a digital audio signal, and an orthogonal transformation of the input digital audio signal into a predetermined period unit of an audio frame. An orthogonal transform process, and a mask amount for calculating a mask amount for each subband, which is a predetermined frequency width unit, based on a sound masking effect from the power spectrum of the digital audio signal obtained by the orthogonal transform process Calculating the difference between one of the power spectrum or the mapped signal obtained by mapping the digital audio signal on the frequency axis by a method different from the orthogonal transform processing and the mask amount obtained by the mask amount calculating process, Calculate each audio frame above Signal-to-mask ratio calculation processing for obtaining a signal-to-mask ratio in the memory, signal-to-noise ratio storage processing for storing a signal-to-noise ratio for a predetermined quantization step as a table, and the signal-to-noise ratio storage processing A mask-to-noise ratio calculation process of obtaining a mask-to-noise ratio by performing a difference calculation between the signal-to-noise ratio and the signal-to-mask ratio obtained by the signal-to-mask ratio calculation process for each subband, The mask-to-noise ratio and the signal-to-noise ratio obtained by the mask-to-noise ratio calculation processing, and the bits of the compressed data for each sub-band using bits that can be assigned to the audio frame during audio signal compression. In a digital audio signal compression method for performing bit allocation processing for determining the number of allocations, a mask obtained by a mask-to-noise ratio calculation processing. For click-to-noise ratio, it has a mask-to-noise ratio average value calculation process that calculates the average value of the predetermined sub-band,
In the bit allocation process, the mask-to-noise ratio, the mask-to-noise ratio average value, and the signal-to-noise ratio for each of the predetermined sub-bands, using the bits that can be assigned to the audio frame during audio signal compression, It is characterized in that the number of bits of compressed data for each subband is determined. Another feature of the digital audio signal compression method of the present invention is that, for one of the subbands, the quantization step is performed in the mask-to-noise ratio calculation process until the mask-to-noise ratio average value is exceeded. It is characterized in that it has a repetitive mask-to-noise ratio calculation process for updating the mask-to-noise ratio so as to make it finer, and then repeats the same calculation for other subbands. Another feature of the digital audio signal compression method of the present invention is that, for one of the sub-bands, the signal-to-noise ratio exceeding the mask-to-noise ratio average value in the mask-to-noise ratio calculation processing. It has a signal-to-noise ratio search process for searching using the above-described signal-to-noise ratio search process, and thereafter repeats the same calculation for other subbands. Another feature of the digital audio signal compression method of the present invention is that, after the mask-to-noise ratio exceeds the mask-to-noise ratio average value for the predetermined subband, the mask pair for each subband is set. Using the noise ratio, the average value of the mask-to-noise ratio, the signal-to-noise ratio, and the bits that can be allocated to the audio frame when compressing the audio signal, determine the number of bits to be allocated to the compressed data for each sub-band. It is characterized by doing. Other features of the digital audio signal compression method of the present invention include an input process of inputting a digital audio signal, and an orthogonal transformation of the input digital audio signal into a predetermined period unit of an audio frame. An orthogonal transform process, and a mask amount for calculating a mask amount for each subband, which is a predetermined frequency width unit, based on a sound masking effect from the power spectrum of the digital audio signal obtained by the orthogonal transform process Calculating the difference between one of the power spectrum or the mapped signal obtained by mapping the digital audio signal on the frequency axis by a method different from the orthogonal transform processing and the mask amount obtained by the mask amount calculating process, Calculated for each audio frame Signal-to-mask ratio calculation processing for obtaining a signal-to-mask ratio in the memory, signal-to-noise ratio storage processing for storing a signal-to-noise ratio for a predetermined quantization step as a table, and signal-to-noise ratio storage processing A mask-to-noise ratio calculation process of obtaining a mask-to-noise ratio by performing a difference calculation between the signal-to-noise ratio and the signal-to-mask ratio obtained by the signal-to-mask ratio calculation process for each subband, Bit allocation of compressed data for each sub-band using the mask-to-noise ratio obtained by the mask-to-noise ratio calculation process, the signal-to-noise ratio, and bits that can be allocated to the audio frame during audio signal compression In a digital audio signal compression method for performing a bit allocation process for determining the number of audio signals, the above-described audio file The number of possible allocated bits to over arm and the quantization step,
A mask-to-noise ratio improvement amount calculation process for calculating a mask-to-noise ratio improvement amount assigned to the audio frame from the signal-to-noise ratio relationship; and a mask pair obtained by the mask-to-noise ratio calculation process in the subband. Mask-to-noise ratio sum calculation processing for calculating the sum of the noise ratios;
The sum of the mask-to-noise ratio improvement amount obtained as a result of the mask-to-noise ratio improvement amount calculation process and the mask-to-noise ratio total sum obtained by the mask-to-noise ratio total calculation process is obtained, and divided by the number of subbands. And performs a target value calculation process for obtaining a bit allocation target value. Another feature of the digital audio signal compression method according to the present invention is that the sub-band 1
And a signal-to-noise ratio search process for searching the signal-to-noise ratio exceeding the bit allocation target value using the signal-to-noise ratio search process in the mask-to-noise ratio calculation process. It is characterized in that the same calculation is repeated for the band. Another feature of the digital audio signal compression method of the present invention is that, in one of the sub-bands, the quantization step is finely performed in the mask-to-noise ratio calculation process until the bit allocation target value is exceeded. And a repetitive mask-to-noise ratio calculation process for updating the mask-to-noise ratio so that the same calculation is repeated for the other subbands. Another feature of the digital audio signal compression method of the present invention is that, for the predetermined subband, after the mask-to-noise ratio exceeds the bit allocation target value,
The mask-to-noise ratio, the mask-to-noise ratio average value, the signal-to-noise ratio, and the bits that can be assigned to the audio frame at the time of audio signal compression for each of the subbands. It is characterized in that the number of allocated bits is determined. Another feature of the digital audio signal compression method of the present invention is that, in the bit allocation processing, bit allocation is performed for all subbands requiring bit allocation.

[0030] The storage medium of the present invention is characterized in that a program for causing a computer to function as each means constituting the audio signal compression device is stored so as to be readable from the computer. Another feature of the storage medium of the present invention is that a program for causing a computer to execute the procedure of the audio signal compression method is stored so as to be readable from the computer.

[0031]

Since the present invention has the above technical means, in a digital audio signal compression method for compressing a digital audio signal, a mask-to-noise ratio improvement index value is obtained from a mask-to-noise ratio in each subband. Until the mask-to-noise ratio improvement index value is exceeded, the bit allocation process is performed for each of the subbands, thereby making it possible to omit the minimum mask-to-noise ratio calculation process.

[0032]

DESCRIPTION OF THE PREFERRED EMBODIMENTS [First Embodiment] FIG. 1 is a flowchart relating to a bit allocation operation according to the present embodiment. FIG. 2 is a block diagram, and FIG. 3 is an operation conceptual diagram of the first embodiment. Hereinafter, the operation of the present embodiment will be described with reference to FIGS.

In FIG. 2, 1 is a digital audio input terminal, 2 is a sub-band encoding unit that divides the digital audio signal from the digital audio input terminal 1 into predetermined time intervals, and maps them to a plurality of frequency bands. An FFT operation unit that performs FFT analysis on the digital audio signal from the digital audio input terminal 1 during a period corresponding to the subband encoding unit 2.

Reference numeral 4 denotes a mask amount calculation unit for converting the result of the FFT calculation unit 3 into sound pressure and calculating a mask amount. Reference numeral 5 designates a mask amount calculated by the mask amount calculation unit 4, which is assigned to each subband. A bit allocation operation unit for estimating the number of bits, and a data quantization unit 6 for quantizing a subband signal output from the subband encoding unit 2 in accordance with the number of bits allocated by the bit allocation operation unit 5 It is.

Reference numeral 7 denotes a transmission format forming unit for formatting the quantized data output from the data quantization unit 6 and the bit allocation information from the bit allocation calculating unit 5 into a predetermined transmission format. Data output end. Hereinafter, the compression operation will be described.

The data input to the digital audio input terminal 1 is supplied to a sub-band encoder 2 and an FFT processor 3
Is input to First, in the subband encoding unit 2, an overlap polyphase filter bank (OPFB)
Thus, the input signal is mapped to band division data in units of 750 Hz at predetermined time intervals (referred to as audio frames). This data is data to be encoded later.

On the other hand, the FFT operation unit 3 performs an FFT analysis in order to analyze the frequency distribution of the sample data.
This result is input to the mask amount calculation unit 4. In the mask amount calculation unit 4, the result of the FFT analysis is a sound pressure value [dB].
Is converted to Then, a masker component that interferes with a sound pressure component on another frequency axis is extracted from the calculated sound pressure value, and a mask amount LTm (i) that affects a nearby frequency position i is calculated.

The above-described masker extraction and mask amount calculation processing are performed over the entire band, and finally the mask amount at the frequency position i is given by the sum of the static hearing limit value and the mask amounts by a plurality of maskers. The mask amount calculated by the mask amount calculator 4 is output to the bit allocation calculator 5.

FIG. 8 (a)
, The minimum value of the mask amount is detected for each sub-band, and the minimum mask amount LTmin [s
b] (Step S100 in FIG. 1). Here, the suffix [sb] is a function of the subband number.

The maximum value of the sub-band data is defined as the maximum sound pressure value L [sb], and the mask-to-noise ratio (MNR) is calculated from the difference between the maximum sound pressure value L [sb] and the minimum mask amount LTmin [sb]. (Step S10 in FIG. 1)
1). It is assumed that the calculation of the mask-to-noise ratio is completed for all the sub-bands and the mask-to-noise ratio of each sub-band has the value shown in FIG.

From this NMR, the NMR average value MNRave-53 [dB] of all the sub-bands is calculated (step S102 in FIG. 1). Then, the maximum allocable bit number Balc (3000) of the compressed sample data other than the header information permitted in the compression format is set (FIG. 1).
Step S103).

Then, the loop counter sb indicating the sub-band is reset (step S10 in FIG. 1).
4) MNR [sb] is compared with MNRave (step S105 in FIG. 1), and if MNR [sb] is small, the quantization step of the subband is changed to a value one step higher (step S105). Step S106 in FIG. 1). this is,
This corresponds to reducing the quantization step.

In response to the reduction in the quantization step, an assigned bit number Buse indicating the number of bits used by the quantization step is calculated (step S10 in FIG. 1).
7). Then, the contents of ALC [SB], which is the buffer holding the bit allocation amount, are updated (step S108 in FIG. 1), and the mask-to-noise ratio is calculated again using the new SNR [step [SB]] value. , MNR [SB] value (step S109 in FIG. 1).

Thereafter, the process returns to step S105 in FIG. 1, and the MNR value of the sub-band to be processed is M
The quantization step is reduced until it becomes larger than NRave. For example, in the example of FIG. 3A, since the sb counter is “0”, the MNR [SB0] (-85d
B) is compared with MNRave (53 dB).

As a result, MNR [SB0] becomes MNRav
e, the quantization step for the sub-band SB0 is reduced. Thereby, step “2” in FIG. 6 is selected, and the SNR is updated to “7 dB”. That is, the MNR of SB0 is “−83 dB”.

However, since the MNRave has not been reached, the quantization step of SB0 is further updated from step "2" to step "3", the SNR is changed from "7 dB" to "16 dB", and the MNR is recalculated. .

Such a loop process is performed by setting the MNR to the MNR.
Repeat until ave or more. As a result, in SB0, in step "5", SNR (32 dB) -SMR (8
5 dB) = MNR (-53 dB), and the process exits the loop at the branch of step S105 in FIG.

The thick shaded portion in FIG. 3A indicates the SNR [step [SB0]] value "53 dB" obtained by this loop processing, and the MNR value of SB0 is the difference. .

When it is determined in step S105 of FIG. 1 that the MNR value of the subband to be processed is larger than MNRave, the subband to be processed is determined by the predetermined maximum number of subbands sbmax. Is determined (step S110 in FIG. 1).

If the result of this determination is that sb <sbmax, the sb counter value is incremented (step S111 in FIG. 1), and the MNR processing of the next subband SB1 (steps S105 to S109 in FIG. 1) is performed as described above. repeat.

At the time when the MNR process (steps S105 to S109 in FIG. 1) is performed on all the subbands,
(B) An SNR value as shown by a thick shaded portion is obtained, and an outlined portion is an MNR value corresponding to the current bit allocation amount.

Thereafter, the same processing as in the conventional example is performed. That is, the minimum value is searched for in the MNR [sb1 calculated so far (step S112 in FIG. 1). The search result is set to MNRmin [SB]. Further, MNRm
The quantization step of the subband corresponding to in is changed to a value one step higher (step S113 in FIG. 1). And
The number of allocated bits B indicating the number of bits used by the quantization step, corresponding to the reduced quantization step.
Use is calculated (step S114 in FIG. 1).

Next, a numerical comparison is made between the number of allocable bits Balc and the number of allocated bits Buse (FIG. 1).
If Bus does not reach Balc, it is determined that bit allocation is possible, and the contents of ALC [SB], which is a buffer holding the bit allocation amount, are updated (step S116 in FIG. 1).

Then, a new SNR [step [S
B]] value to calculate the mask-to-noise ratio again and calculate the MNR
[SB] The value is updated (step S116 in FIG. 1). Thereafter, the process returns to step S112 in FIG. 1, and a loop process for searching for the minimum MNR is performed again. This loop process is performed until it is determined in step S115 in FIG. 1 that bit allocation is impossible, that is, until Balc <Buse.

At the time when Balc <Buse, FIG.
The bit allocation calculation process is completed in a state where the MNR value is improved by the amount indicated by the thin oblique line in (d) and the MNR variation between subbands is small.

Then, the bit allocation operation processing section 5
Based on ALC [sb], which is a buffer holding the bit allocation amount resulting from
Then, the sub-band data is quantized, and the transmission format forming unit 7 formats the quantized data together with side information such as bit allocation information in a predetermined style, and outputs it to the compressed data output terminal 8.

By adopting a configuration in which the quantization step is uniquely reduced to the MNR average value between the subbands as described above, the minimum NMRmin is searched every time the quantization step is changed even by one step. Optimal bit allocation calculation can be performed without taking steps.

[Second Embodiment] FIG. 4 is a flowchart for explaining a procedure of a bit allocation operation according to a second embodiment of the present invention. Hereinafter, the operation of the second exemplary embodiment of the present invention will be described with reference to the block diagram of FIG. 2 and the operation conceptual diagram of FIG. Here, the description of the entire compression processing is equivalent to that of the first embodiment, and thus the description thereof will be omitted.

FIG. 5A shows the bit allocation operation section 5.
, The minimum value of the mask amount is detected for each sub-band,
The minimum mask amount LTmin [sb] is set (step S400 in FIG. 4). Here, the suffix [sb] is a function of the subband number.

The maximum value of the sub-band data is defined as the maximum sound pressure value L [sb], and the MNR is calculated from the difference between the maximum sound pressure value L [sb] and the minimum mask amount LTmin [sb].
(Mask to noise ratio) is calculated (Step S40 in FIG. 4)
1).

Next, an MNR improvement amount MNRup determined from the maximum allocable bit number Balc of the compressed sample data other than the header information permitted in the compression format is set (step S402 in FIG. 4).

The MNRup value is a decibel amount that can be improved by the number of bits of Balc, that is, the MNRup value of one subband.
This is a value calculated from the number of bits necessary for improving the NR by “1 dB” and the decibel amount that can improve the MNR with the number of bits given by the allocatable bit number Balc, and can be converted from, for example, the SNR table in FIG.

That is, from the relationship between the SNR and the number of allocated bits shown in FIG.
Paying attention to "dB", the SNR that can be improved per bit allocation is SNR (98 dB) / the maximum bit allocation per subband (576 bits) = the maximum improvement (0.17 dB) per bit. From this, the SNR that can be improved by the number of assignable bits Balc is 1
Improvement per bit (0.17 dB) x Balc (3
000) = MNRup (510 dB).

On the other hand, in all subbands, the MNR
When the calculation is completed, the sum MNRall of the MNR values of all subbands is calculated (step S403 in FIG. 4), and the MNRup value is distributed to each subband, so that the NMR of each subband becomes constant. ALCtgt = (MNR
up tens MNRall) / SBSB: Obtained from the number of subbands requiring bit allocation (step S40 in FIG. 4)
4). In the example of FIG. 5, MNRall is "-424 dB".
Therefore, ALCtgt is 10 dB.

Then, the loop counter sb indicating the sub-band is reset (step S40 in FIG. 4).
5), the solution of the difference calculation between ALCtgt and MNR [sb] is stored in MNRtmp (step S406 in FIG. 4),
The SNR table is searched as shown in FIG.
The step value is determined so that [step] becomes the minimum value larger than MNRtmp (step S 407 in FIG. 4).
-Step S410).

The relationship that SNR [step] is the minimum value larger than MNRtmp is established.
8) or when the step value exceeds a predetermined quantization step, for example, in the case of FIG. 6, when the number of steps is 18 or more (step S409 in FIG. 4), step =
As step 17, the step is stored in ALC [SB0] which is a buffer for holding the bit allocation amount (FIG. 4).
Step S411).

For example, in the example shown in FIG. 5, since the sb counter is currently "0", the process for "SB0" is performed first.

MNR [SB0] (-85 dB) and ALC
MNRtmp (95 dB) is derived from tgt (10 dB), a step value 16 is determined from the MNRtmp (95 dB) using an SNR table, and this is stored in ALC [SB0], which is a buffer that holds the bit allocation amount.

Therefore, M shown in FIG.
The NR improvement is realized by one MNRtmp calculation. Then, the same processing is performed for the next subband SB1, but in this case, the MNRtmp becomes “100 dB”, and
Since the maximum value of the quantization step is exceeded, FIG.
As shown in (b), the step does not reach ALCtgt (10 dB), and the step has the maximum value “16” in step S404 in FIG.
The process exits the loop processing of SB1 under the condition that

Such processing is repeatedly performed to determine whether the number of subbands currently being processed is the predetermined maximum number of subbands sbmax (step S412 in FIG. 4), and if sb <sbmax. For example, the sb counter value is incremented (step S41 in FIG. 4).
3), the next sub-band bit allocation operation (FIG. 4)
Steps S406 to S410) are repeated as described above for all subbands.

By performing the above-described processing, an optimal bit allocation operation can be performed without performing a quantization step such as searching for the minimum NMRmin. Furthermore, since the allocation target value including the number of bits that can be allocated, Balc, as an element is set, an optimal bit allocation operation can be performed by one operation for each subband, so that the processing can be speeded up.

(Other Embodiments of the Present Invention) The present invention is applied to a system composed of a plurality of devices (for example, a host computer, an interface device, a reader, a printer, etc.) and is applied to an apparatus composed of one device. You may.

Further, in order to realize the functions of the above-described embodiments, a device connected to the above-mentioned various devices or a computer in a system is realized so as to operate various devices so as to realize the functions of the above-described embodiments. The present invention also includes programs implemented by supplying the program code of the software described above and operating the various devices according to programs stored in a computer (CPU or MPU) of the system or apparatus.

In this case, the program code of the software implements the functions of the above-described embodiment, and the program code itself and means for supplying the program code to the computer are provided.
For example, a storage medium storing such a program code constitutes the present invention. As a storage medium for storing such a program code, for example, a floppy disk, hard disk, optical disk, magneto-optical disk, CD-ROM,
A magnetic tape, a nonvolatile memory card, a ROM, or the like can be used.

When the computer executes the supplied program code, not only the functions of the above-described embodiment are realized, but also the OS (operating system) or other operating system running on the computer. Needless to say, even when the functions of the above-described embodiments are realized in cooperation with application software or the like, such program codes are included in the embodiments of the present invention.

Further, after the supplied program code is stored in a memory provided in a function expansion board of a computer or a function expansion unit connected to the computer, the function expansion board or the function expansion unit is specified based on the instruction of the program code. It is needless to say that the present invention also includes a case where the CPU or the like provided in the first embodiment performs part or all of the actual processing, and the processing realizes the functions of the above-described embodiments.

[0077]

According to the present invention, a mask-to-noise ratio improvement index value is obtained in advance from the mask-to-noise ratio in each sub-band, and the mask-to-noise ratio calculation is performed uniquely for each of the sub-bands. Since the bit allocation process is performed until the ratio improvement index value is exceeded, the time required for the calculation process can be reduced by omitting the minimum mask-to-noise ratio calculation process.

According to another feature of the present invention, it is possible to perform appropriate bit allocation in a single bit allocation calculation process for each subband, and it is possible to greatly reduce the calculation processing time.

Further, according to another feature of the present invention, since the allocation target value including the number of allocable bits Balc as an element is set, the optimum bit allocation operation is performed by one operation for each subband. It is possible to increase the processing speed.

[Brief description of the drawings]

FIG. 1 is a flowchart of a bit allocation method according to a first embodiment of the present invention.

FIG. 2 is a block diagram of digital audio signal compression.

FIG. 3 is an operation conceptual diagram of the first embodiment of the present invention.

FIG. 4 is a flowchart of a bit allocation method according to a second embodiment of the present invention.

FIG. 5 is an operation conceptual diagram according to a second embodiment of the present invention.

FIG. 6 is a diagram illustrating an example of a table of quantization steps and a signal-to-noise ratio.

FIG. 7 is a flowchart of a conventional bit allocation method.

FIG. 8 is a diagram for explaining the operation of a bit allocation operation unit.

[Explanation of symbols]

 Reference Signs List 1 digital audio input terminal 2 subband encoding unit 3 FFT operation unit 4 mask amount operation unit 5 bit allocation operation unit 6 data quantization unit 7 transmission format formation unit 8 compressed data output terminal

Claims (24)

[Claims] 1. A digital audio signal compression apparatus for compressing a digital audio signal, wherein a mask-to-noise ratio improvement index value is obtained from a mask-to-noise ratio in each subband until the calculated mask-to-noise ratio improvement index value is exceeded. A digital audio signal compression device, wherein a bit allocation process is performed for each subband. 2. An input means for inputting a digital audio signal, an orthogonal transform means for orthogonally transforming the digital audio signal input via the input means, and a sound spectrum obtained from a power spectrum as a result of the orthogonal transform means. A mask amount calculating means for calculating a mask amount based on a masking effect; and either the power spectrum or a mapped signal obtained by mapping the digital audio signal on a frequency axis by a method different from that of the orthogonal transform means; A signal-to-mask ratio calculating means for calculating a ratio of the mask amount obtained by the means in a predetermined frequency range unit; a signal-to-noise ratio defining means for defining an SN ratio for a predetermined bit allocation amount; The result of the signal-to-noise ratio definition means is updated each time the predetermined bit allocation amount is changed, and And a bit for determining the bit allocation number of the compressed data in the predetermined frequency width unit based on the result of the iterative operation means. A digital audio signal compression device having an allocation unit, wherein the bit allocation target value setting unit sets a common predetermined bit allocation target value in the predetermined frequency band unit, and the signal to noise ratio defining unit. Selecting means for selecting a value that results in a minimum value greater than the bit allocation target value, and bit allocation calculating means for performing a bit allocation operation using the result of the selecting means. Digital audio signal compression device. 3. An input means for inputting a digital audio signal, an orthogonal transform means for orthogonally transforming the input digital audio signal in a predetermined period unit of an audio frame, and a digital signal obtained by the orthogonal transform means. Mask amount calculating means for calculating a mask amount for each sub-band, which is a predetermined frequency range unit, based on a sound masking effect from a power spectrum of an audio signal, and the power spectrum or the digital audio signal The difference between one of the mapped signals mapped on the frequency axis by a method different from the orthogonal transformation means and the mask amount obtained by the mask amount calculation means is calculated for each subband, and the signal-to-mask ratio in the audio frame is calculated. The signal-to-mask ratio calculation means to be determined and a predetermined Signal-to-noise ratio storage means for storing a signal-to-noise ratio for a given quantization step as a table, and signal-to-noise ratio and signal-to-mask ratio calculation means output from the signal-to-noise ratio storage means. A mask-to-noise ratio calculating means for calculating a difference between the signal-to-mask ratio obtained by the above for each sub-band to obtain a mask-to-noise ratio; and a mask-to-noise ratio obtained by the mask-to-noise ratio calculating means. In a digital audio signal compression apparatus having a signal-to-noise ratio and bit allocation means for determining the number of allocated bits of compressed data for each sub-band using bits that can be allocated to the audio frame during audio signal compression, An average value in a predetermined sub-band is calculated for the mask-to-noise ratio obtained by the mask-to-noise ratio calculating means. A mask-to-noise ratio average value calculating means, wherein the bit allocating means performs the mask-to-noise ratio, the mask-to-noise ratio average value, the signal-to-noise ratio, and the audio-signal compression for each of the predetermined subbands. A digital audio signal compression device, wherein a bit allocation number of compressed data for each subband is determined using bits that can be allocated to the audio frame. 4. The mask-to-noise ratio of one of the sub-bands is updated by the mask-to-noise ratio calculating means so that the quantization step is refined until the mask-to-noise ratio average value is exceeded. 4. The digital audio signal compression apparatus according to claim 3, further comprising a repetition mask-to-noise ratio calculation unit, and thereafter repeating the same calculation for the other subbands. 5. A signal pair for one of the sub-bands, wherein the signal-to-noise ratio search means uses the signal-to-noise ratio search means to search for the signal-to-noise ratio exceeding the mask-to-noise ratio average value in the mask-to-noise ratio calculation means. 4. The digital audio signal compression device according to claim 3, further comprising a noise ratio search unit, wherein the same operation is repeated for other subbands thereafter. 6. The mask-to-noise ratio for each of the sub-bands and the mask-to-noise ratio for each of the sub-bands after the mask-to-noise ratio exceeds the mask-to-noise ratio average value. A digital audio signal compression apparatus for determining the number of bits to be allocated to the compressed data for each sub-band by using the signal-to-noise ratio and bits that can be allocated to the audio frame when compressing the audio signal. 7. An input means for inputting a digital audio signal, an orthogonal transform means for orthogonally transforming the input digital audio signal in a predetermined period unit of an audio frame, and a digital signal obtained by the orthogonal transform means. Mask amount calculation means for calculating a mask amount for each sub-band, which is a predetermined frequency range unit, based on a sound masking effect from a power spectrum of an audio signal, and the power spectrum or the digital audio signal The difference between one of the mapped signals mapped on the frequency axis by a method different from the orthogonal transform means and the mask amount obtained by the mask amount calculating means is calculated for each of the subbands, and the signal-to-mask ratio in the audio frame is obtained. Signal-to-mask ratio calculation means and predetermined Signal-to-noise ratio storage means for storing a signal-to-noise ratio for a given quantization step as a table; and signal-to-noise ratio and signal-to-mask ratio calculation means output from the signal-to-noise ratio storage means. A mask-to-noise ratio calculating means for performing a difference calculation with respect to the signal-to-mask ratio obtained for each sub-band to obtain a mask-to-noise ratio; and a mask-to-noise ratio obtained by the mask-to-noise ratio calculating means. In a digital audio signal compression apparatus having a signal-to-noise ratio and bit allocation means for determining a bit allocation number of compressed data for each sub-band by using bits that can be allocated to the audio frame at the time of audio signal compression, In audio compression, the number of bits that can be allocated to the audio frame and the quantization step, Mask-to-noise ratio improvement amount calculating means for calculating a mask-to-noise ratio improvement amount assigned to the audio frame from a signal-to-noise ratio relationship, and a mask pair obtained by the mask-to-noise ratio calculation means in the subband. A mask-to-noise ratio sum calculating means for calculating the sum of the noise ratios; a mask-to-noise ratio improvement amount as a result of the mask-to-noise ratio improvement amount calculating means; and a mask obtained by the mask-to-noise ratio sum calculating means. A digital audio signal compression device comprising a target value calculating means for obtaining a sum with a noise-to-noise ratio sum and dividing by the predetermined number of subbands to obtain a bit allocation target value. 8. A signal-to-noise ratio for one of the sub-bands, wherein the mask-to-noise ratio calculating means searches for the signal-to-noise ratio exceeding the bit allocation target value using the signal-to-noise ratio searching means. 8. The digital audio signal compression apparatus according to claim 7, further comprising a search unit, wherein the same operation is repeated for other subbands thereafter. 9. A repetitive mask for updating the mask-to-noise ratio for one of the sub-bands so that the mask-to-noise ratio calculating means refines the quantization step until the bit allocation target value is exceeded. 8. The digital audio signal compression apparatus according to claim 7, further comprising noise-to-noise ratio calculation means, and thereafter repeating the same calculation for other subbands. 10. The mask-to-noise ratio, the mask-to-noise ratio average value, and the mask-to-noise ratio for each of the predetermined sub-bands, after the mask-to-noise ratio exceeds the bit allocation target value. Signal to noise ratio and
10. The digital audio signal compression apparatus according to claim 8, wherein the number of bits allocated to the compressed data for each sub-band is determined by using bits that can be allocated to the audio frame when compressing the audio signal. 11. A digital audio signal compression method for compressing a digital audio signal, comprising: obtaining a mask-to-noise ratio improvement index value from a mask-to-noise ratio in each subband until the mask-to-noise ratio improvement index value exceeds the calculated mask-to-noise ratio improvement index value. A digital audio signal compression method, wherein a bit allocation process is performed for each sub-band. 12. An input process for inputting a digital audio signal, an orthogonal transform process for orthogonally transforming the input digital audio signal, and a mask based on a speech masking effect from a power spectrum which is a result of the orthogonal transform process. A mask amount calculation process for calculating the amount, and either the power spectrum or the mapped signal obtained by mapping the digital audio signal on the frequency axis by a method different from the orthogonal transformation process, and the mask amount calculation process. A signal-to-mask ratio calculation process for calculating a ratio with the mask amount in a predetermined frequency range unit; a signal-to-noise ratio definition process for defining an SN ratio for a predetermined bit allocation amount; Each time the amount is changed, the result of the signal-to-noise ratio definition process is updated, and the signal-to-mask A repetition calculation process for calculating the mask-to-noise ratio with the result of the ratio calculation process; and a bit allocation process for determining the bit allocation number of the compressed data in the predetermined frequency width unit based on the result of the repetition calculation process. In the digital audio signal compression method to be performed, a bit allocation target value setting process for setting a common predetermined bit allocation target value in the predetermined frequency range unit, and a result of the signal-to-noise ratio defining process are as described above. A digital processing unit for performing a selection process of selecting a value that is smaller than a bit allocation target value obtained by a bit allocation target value setting process, and performing a bit allocation operation using a result of the selection process. Audio signal compression method. 13. An input process for inputting a digital audio signal, an orthogonal transform process for orthogonally transforming the input digital audio signal in a predetermined period unit of an audio frame, and a digital signal obtained by the orthogonal transform process. A mask amount calculation process for calculating a mask amount for each sub-band which is a predetermined frequency range unit based on a sound masking effect from a power spectrum of an audio signal, and converting the power spectrum or the digital audio signal to The difference between one of the mapped signals mapped on the frequency axis by a method different from the orthogonal transformation process and the mask amount obtained by the mask amount calculation process is calculated for each subband, and the signal-to-mask ratio in the audio frame is calculated. The required signal-to-mask ratio calculation A signal-to-noise ratio storage process for storing a signal-to-noise ratio for a quantization step being set as a table, and a calculation of the signal-to-noise ratio and the signal-to-mask ratio output from the signal-to-noise ratio storage process A mask-to-noise ratio calculation process for calculating a difference between the signal-to-mask ratio obtained by the processing for each sub-band to obtain a mask-to-noise ratio; and a mask-to-noise ratio obtained by the mask-to-noise ratio calculation process. In the digital audio signal compression method, the signal-to-noise ratio and a bit allocation process for determining a bit allocation number of compressed data for each sub-band by using bits that can be allocated to the audio frame during audio signal compression. Calculating an average value in a predetermined sub-band of the mask-to-noise ratio obtained by the mask-to-noise ratio calculation processing. In the bit allocation process, the mask-to-noise ratio, the mask-to-noise ratio average value, the signal-to-noise ratio, and the audio-to-noise ratio compression process are performed in the bit allocation process. Determining a bit allocation number of compressed data for each of said subbands by using bits that can be allocated to said audio frame. 14. The mask-to-noise ratio for one of the sub-bands is updated in the mask-to-noise ratio calculation process such that the quantization step is refined until the average value exceeds the mask-to-noise ratio. 14. The digital audio signal compression method according to claim 13, further comprising a repetitive mask-to-noise ratio calculation process, and thereafter repeating the same calculation for other subbands. 15. A signal pair for searching one of the subbands for the signal-to-noise ratio exceeding the mask-to-noise ratio average value in the mask-to-noise ratio calculation process using the signal-to-noise ratio search process. 14. The digital audio signal compression method according to claim 13, further comprising a noise ratio search process, and thereafter repeating the same operation for other subbands. 16. After the mask-to-noise ratio exceeds the mask-to-noise ratio average value for the predetermined sub-band, the mask-to-noise ratio for each sub-band, the mask-to-noise ratio average value, A digital audio signal compression method, comprising: determining a bit allocation number of compressed data for each sub-band by using the signal-to-noise ratio and bits that can be allocated to the audio frame during audio signal compression. 17. An input process for inputting a digital audio signal, an orthogonal transform process for orthogonally transforming the input digital audio signal in a predetermined period unit of an audio frame, and a digital signal obtained by the orthogonal transform process. A mask amount calculation process for calculating a mask amount for each sub-band which is a predetermined frequency width unit based on a sound masking effect from a power spectrum of an audio signal, and the power spectrum or the digital audio signal A difference between one of the mapped signals mapped on the frequency axis by a method different from the orthogonal transformation process and the mask amount obtained by the mask amount calculation process is calculated for each of the subbands, and a signal-to-mask ratio in the audio frame is obtained. Signal-to-mask ratio calculation processing and preset A signal-to-noise ratio storing process for storing a signal-to-noise ratio for a quantization step being set as a table; and calculating the signal-to-noise ratio and the signal-to-mask ratio output from the signal-to-noise ratio storing process. A mask-to-noise ratio calculation process for calculating a difference between the signal-to-mask ratio obtained by the processing for each sub-band to obtain a mask-to-noise ratio; and a mask-to-noise ratio obtained by the mask-to-noise ratio calculation process. In the digital audio signal compression method, the signal-to-noise ratio and a bit allocation process of determining a bit allocation number of compressed data for each of the sub-bands using bits that can be allocated to the audio frame during audio signal compression, The number of bits that can be allocated to the audio frame and the quantization step during audio compression in advance; A mask-to-noise ratio improvement amount calculation process for calculating a mask-to-noise ratio improvement amount assigned to the audio frame from a relationship between the signal-to-noise ratio, and a mask-to-noise ratio obtained by the mask-to-noise ratio calculation process in the subband. A mask-to-noise ratio sum calculation process for calculating the sum of the ratios; a mask-to-noise ratio improvement amount as a result of the mask-to-noise ratio improvement amount calculation process; and a mask pair obtained by the mask-to-noise ratio sum calculation process. A digital audio signal compression method comprising: obtaining a sum with a noise ratio sum; performing a division by the number of sub-bands; and performing a target value calculation process for obtaining a bit allocation target value. 18. A signal-to-noise ratio for searching one of the sub-bands using the signal-to-noise ratio search process to find the signal-to-noise ratio exceeding the bit allocation target value in the mask-to-noise ratio calculation process. 18. The digital audio signal compression method according to claim 17, further comprising a search process, and thereafter repeating the same operation for other subbands. 19. A repetitive mask for updating the mask-to-noise ratio for one of the sub-bands so as to refine the quantization step until the bit-allocation target value is exceeded in the mask-to-noise ratio calculation process. 18. The digital audio signal compression method according to claim 17, further comprising a noise-to-noise ratio calculation process, and thereafter repeating the same calculation for other subbands. 20. For the predetermined sub-band, after the mask-to-noise ratio exceeds the bit allocation target value, the mask-to-noise ratio, the mask-to-noise ratio average value, and the signal pair for each of the sub-bands. 19. The noise ratio and the number of bits to be allocated to the compressed data for each sub-band are determined by using bits that can be allocated to the audio frame when compressing the audio signal.
Or a digital audio signal compression method according to item 19. 21. A storage medium storing a program for causing a computer to function as each means constituting the audio signal compression apparatus according to claim 1 so as to be readable from the computer. 22. A storage medium storing a program for causing a computer to execute the procedure of the audio signal compression method according to claim 11 so as to be readable from the computer. 23. The digital audio signal compression apparatus according to claim 2, wherein said bit allocation means performs bit allocation for all subbands requiring bit allocation. 24. The bit allocation process according to claim 11, wherein bit allocation is performed for all subbands requiring bit allocation.
3. A digital audio signal compression method according to claim 1.