JPH08179794A - Sub-band coding method and device - Google Patents

Abstract

PURPOSE: To restrain a pre-echo without deteriorating coding efficiency even in a digital audio signal having a part where an energy change is large when a sub-band is coded. CONSTITUTION: Amplitude of an input signal from a digital audio signal input terminal 9 is detected by a mode data output part 6, and a mode to group data obtained by dividing the input signal into bands by a frequency band dividing filter bank 2 according to a part where this amplitude suddenly changes is set, and is outputted. The data is grouped into a data group according to the mode by a scale factor calculating part 3. A quantizing bit is allocated with every this data group by a bit allocating part 4, and the data is quantized on the basis of the quantizing bit by a quantizing part 5.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a subband coding method and apparatus for dividing data of a digital audio signal into data of a plurality of frequency bands and coding the data.

[0002]

2. Description of the Related Art In recent years, in order to reduce the amount of data for digital audio signals, for example, a moving picture coding coordinating organization (moving picture image codin) has been developed.
G experts group (MPEG) standardized by M
A subband coding method is adopted in a coding algorithm such as PEG-1. This subband coding scheme is
This is a method of reducing the data of the redundant portion included in the stereo audio signal by utilizing the property of the stereo audio signal data being unevenly distributed in the frequency axis direction of the stereo audio signal.

Here, FIG. 1 is a block diagram schematically showing an example of a configuration for realizing the above-mentioned subband coding system.
6 is shown.

In the sub-band coding method, first, the digital audio signal sent from the input terminal 91 is divided into M frequency regions by the band-pass filter section 81 to obtain sub-band division data. The data obtained by each band pass filter unit 81 is quantized by the quantizing unit 82, and further, the bit stream accumulating unit 83.
A bitstream is formed and stored using the data quantized in. The data in the bit stream is inversely quantized by the inverse quantizer 84, and the synthesis filter 85 assembles the data divided into M frequency regions to reproduce a digital audio signal.

Further, as shown in FIG. 14, the configuration of the main part of a subband coding apparatus to which the above subband coding method is applied is such that sample data of an input digital audio signal is converted into 32 frequency bands, for example. Of each frequency band, that is, a frequency band division filter bank 61 for dividing into subbands, and a maximum value of amplitude of data obtained by dividing into each of the subbands, that is, a scale factor is calculated, and the amplitude of the above data is A scale factor calculation unit 62 that normalizes using the value of the factor, and a quantization level for each subband so that each block has a certain amount of data according to the value of the scale factor for each subband,
Alternatively, the bit allocation unit 63 that allocates the quantization bit
And a quantizer 64 that performs a quantization process on the data of each subband with assigned quantization bits, and a bitstream that rearranges the quantized data in a bitstream format to form a bitstream. And a forming portion 65. Further, the distance in the time axis direction of the subband, that is, the so-called code length is 320.
And

As shown in FIG. 17, the range L of the digital audio signal input from the input terminal 71 is one block, and the code length of each block is 320. That is, one block has 320 samples.

The frequency band division filter 61 divides the digital audio signal into 32 subbands, as shown in FIG. 18, for example. It should be noted that there are 10 pieces of data per block in each subband.

The scale factor calculator 62 sends the above data for each subband. For example, the data of the nth subband as shown in FIG. 19 is the 6th data,
That is, since the absolute value of the amplitude of Sn (6) becomes the maximum value, this scale factor is calculated. Furthermore, based on this scale factor, Sn (0) to Sn
The data up to (9) is normalized.

As shown in FIG. 20, the bit allocation unit 63 uses the uneven distribution of the data amount of each subband based on the scale factor of each subband so that the data amount becomes constant in block units. Quantize bits are assigned to each subband.

Further, as shown in FIG. 21, the bit stream forming unit 65 uses a bit stream format consisting of an auxiliary block 101, a quantized bit block 102, a scale factor block 103 and a sample data block 104, and a quantizing unit 64. The quantized data is rearranged to form a bitstream.

The bit stream obtained by the sub-band coding device is decoded by the sub-band decoding device.

As shown in FIG. 15, this subband decoding device comprises a bitstream decoding unit 66, a bit allocation decoding unit 67, an inverse quantization unit 68, and a synthesis filter bank 69. It is a thing.

In the subband decoding device, the input data from the bitstream input terminal 73 is supplied to the bit allocation decoding unit 67 via the bitstream decoding unit 66. The output data from the bit allocation decoding unit 67 is supplied to the synthesis filter bank 69 via the inverse quantization unit 68. The output data from the synthesis filter bank 69 is supplied to the digital audio signal output terminal 74.

Here, the bit stream decoding unit 66
Is to obtain quantized data by dividing the above bitstream in block units from the data of the auxiliary blocks of the bitstream shown in FIG. Further, the bit allocation decoding unit 67 decodes the quantized bits of each subband from the quantized bit block 102 and sends the quantized bits to the inverse quantization unit 68.
The dequantization unit 68 dequantizes the quantized data from the data in the data block 104 based on the quantized bits. The synthesis filter bank 69 synthesizes the dequantized data to reproduce a digital audio signal, and outputs the digital audio signal from the digital audio signal output terminal 74.

[0015]

By the way, when a signal having a portion where the amplitude changes sharply, that is, an attack portion 111, as shown in FIG. 22, is encoded, the energy distribution of the frequency spectrum is ,
It spreads from low to high frequencies. Therefore, the uneven distribution in the frequency axis direction becomes weak.

In the above subband coding apparatus, a scale factor corresponding to the maximum value of the amplitude is calculated for each subband, and the scale factor is used to normalize the data in the subband. . Therefore, even minute data on the front side of the attack portion is uniformly quantized at the quantization level corresponding to the data having a large amplitude after the attack portion.

Therefore, the efficiency of the encoding process by the above-mentioned sub-band encoding device is reduced, and the quantization noise generated by the above-mentioned quantization process is as shown in FIG. Spread to. For this reason, the audio signal obtained by performing the decoding process again after the encoding process has a signal with quantization noise, that is, a so-called pre-echo 112, in a part before the attack part as shown in FIG. It becomes a signal. Further, there is a problem that the pre-echo is easily perceived.

The present invention has been made in view of the above-mentioned circumstances, and even in a digital audio signal having a portion with a large energy change, the sub-band coding efficiency does not decrease and the pre-echo can be suppressed. It is an object to provide an encoding method and device.

[0019]

In order to solve the above-mentioned problems, the present invention is a subband coding method for dividing sample data of a digital audio signal into data of a plurality of frequency bands and coding the data. A band division step of dividing the sample data of the input digital audio signal into data of a plurality of frequency bands, and grouping the data divided into the frequency bands according to the amplitude of the input digital audio signal Grouping process,
And a quantization step of quantizing the data for each of the data groups obtained in the grouping step.

Further, according to the present invention, in the sub-band coding method, the maximum absolute value is detected in block units of a fixed number of samples for the sample data of the digital audio signal, and the grouping step is performed continuously in time. When the change amount of the maximum absolute value between the blocks is larger than a predetermined threshold value, the length in the time axis direction is shortened to group the data obtained in the band division step.

In the sub-band encoding method of the present invention, the maximum absolute value of the sample data of the digital audio signal is detected in each of the frequency bands,
In the grouping step, when the amount of change in the maximum absolute value between temporally continuous frequency bands is larger than a predetermined threshold value, the length in the time axis direction is shortened to obtain in the band dividing step. The data is grouped.

Further, the present invention is a sub-band encoding apparatus for dividing data of a digital audio signal into sample data of a plurality of frequency bands and encoding the same, wherein the data of the input digital audio signal is divided into a plurality of frequencies. Band division means for dividing into band data, amplitude detection means for detecting the amplitude of the input digital audio signal, and data divided by the band division means into amplitudes detected by the amplitude detection means. Based on, for example, a data group defined by the number of samples in the frequency axis direction and the number of samples in the time axis direction, the grouping means for grouping, and the data for each data group obtained by the grouping means And a quantizing means for quantizing.

Further, in the subband coding device according to the present invention, the amplitude detecting means detects the maximum absolute value of the sample data of the digital audio signal in block units of a fixed number of samples and is continuous in time. Detecting the amount of change in the maximum absolute value between the blocks, the grouping means shortens the length in the time axis direction by the band dividing means when the amount of change is larger than a predetermined threshold value. The obtained data is grouped.

Also, in the subband coding apparatus according to the present invention, the amplitude detecting means detects the maximum absolute value in each of the frequency bands of the sample data of the digital audio signal, and the temporally continuous frequencies. The change amount of the maximum absolute value between the bands is detected, and when the change amount is larger than a predetermined threshold, the grouping unit shortens the length in the time axis direction and is obtained by the band dividing unit. It is a grouping of data.

[0025]

According to the sub-band coding method having the band dividing step, the grouping step, and the quantizing step according to the present invention, the sample data of the digital audio signal input in the band dividing step is converted into the frequency band. And divided in the grouping step, and the data is grouped in a data group defined by the number of samples in the frequency axis direction and the number of samples in the time axis direction according to the amplitude of the digital audio signal. To be done. This data group is set so that when the frequency bandwidth of the digital audio signal changes abruptly, the time resolution in this band is increased. Increases efficiency.

In the subband coding method, when the amount of change in the maximum absolute value between temporally consecutive blocks is larger than a predetermined threshold in the grouping step, the length in the time axis direction is increased. When the data obtained in the band dividing step is grouped by shortening, the data obtained in the band dividing step is the maximum absolute value between the temporally consecutive blocks of the digital audio signal. When the amount of change is larger than a predetermined threshold value, the data group is shortened in the time axis direction. This data group is set so that when the frequency bandwidth of the digital audio signal changes abruptly, the time resolution in this band is increased. Increases efficiency.

In the sub-band coding method, when the amount of change in the maximum absolute value between the temporally consecutive frequency bands in the grouping step is larger than a predetermined threshold, the length in the time axis direction is increased. When the data obtained in the band dividing step is grouped by shortening, the data obtained in the band dividing step is the maximum absolute value between the temporally continuous frequency bands of the digital audio signal. When the change amount of is larger than a predetermined threshold value, the data group is shortened in the time axis direction. This data group is set so that when the frequency bandwidth of the digital audio signal changes abruptly, the time resolution in this band is increased. Increases efficiency.

According to the subband coding apparatus having the band dividing means, the amplitude detecting means, the grouping means, and the quantizing means according to the present invention, the band dividing means is provided for the input digital audio signal. The sample data is divided into each band divided into frequency bands, that is, subband data. The amplitude detecting means detects the amplitude of the input digital audio signal. Further, the grouping means groups the data based on the amplitude in a data group defined by, for example, the number of samples in the frequency axis direction and the number of samples in the time axis direction. This data group is set so that when the frequency bandwidth of the digital audio signal changes abruptly, the time resolution in this band is increased. Increases efficiency.

Further, in the sub-band coding device, the amplitude detecting means detects the maximum absolute value of the sample data of the digital audio signal in block units of a fixed number of samples, and also between consecutive blocks in time. The amount of change in the maximum absolute value is detected, and when the amount of change is larger than a predetermined threshold value, the grouping unit shortens the length in the time axis direction to obtain the data obtained by the band dividing unit. In the case of grouping, this data group is set so that the temporal resolution in this band is increased when the frequency bandwidth of the digital audio signal changes abruptly, and is performed by the quantizing means. The efficiency of the quantization process is improved.

In the sub-band coding device, the amplitude detecting means detects the maximum absolute value of the sample data of the digital audio signal in each of the frequency bands, and the maximum between the time-continuous frequency bands. Detecting the amount of change in absolute value, the grouping means,
When the amount of change is larger than a predetermined threshold value, when the data obtained by the band dividing means is grouped by shortening the length in the time axis direction, this data group is a frequency band of the digital audio signal. The time resolution in this band is set to be increased when the width changes abruptly, and the efficiency of the quantization processing performed by the quantization means is increased.

[0031]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An example of a subband coding method and a subband coding device according to the present invention will be described below in detail with reference to the drawings. Here, at a sampling frequency of 48 kHz,
A subband encoding device for sampled data will be taken as an example.

A first example of the above subband coding apparatus is as follows:
As shown in FIG. This sub-band encoding device includes a frequency band division filter bank 2 that divides input digital audio signal sample data into frequency bands, that is, sub-band data, and a mode data output unit 6 described later. Based on the mode data sent from the above, the above data is grouped, and the maximum value of the amplitude of the data of each data group, a scale factor calculation unit 3 for calculating a so-called scale factor, and each data group based on the above scale factor A bit allocation unit 4 for allocating a quantized bit for quantizing the data, and a quantizer 5 for quantizing the data for each data group according to the quantized bit. is there.

Further, the sub-band coding apparatus detects the maximum absolute value of the sample data of the digital audio signal in block units of a fixed number of samples, and changes in the maximum absolute value between consecutive blocks in terms of time. It also has a mode data output unit 6 as an amplitude detecting means for detecting the amount and outputting the mode data set based on the amount of change.

In the subband coding apparatus, the input data from the digital audio signal input terminal 9 is supplied to the block delay section 1 and the mode data output section 6.

The output data from the block delay unit 1 is supplied to the scale factor calculation unit 3 via the frequency band division filter bank 2. The output data from the scale factor calculation unit 3 is stored in the bit allocation unit 4
Is supplied to the quantizer 5. The output from the quantizing unit 5 is supplied to the bitstream forming unit 7.

The output data from the mode data output unit 6 is supplied to the scale factor calculation unit 3 and the bit stream formation unit 7. Further, the mode data output unit 6 is supplied with a threshold value for determining the mode data from the control unit 8.

Here, the block delay unit 1 divides the digital audio signal sent from the audio processor or the like via the digital audio signal input terminal 9 into blocks, and the data of the digital audio signal in blocks is divided into frequency bands. Send to bank 2.

The frequency band division filter bank 2 is shown in FIG.
As shown in FIG. 7, the data of the digital audio signal in the block unit is 0.7 from DC to 24 kHz band.
It is divided into data of 32 frequency bands, that is, subbands (Sub0 to Sub31) in steps of 5 kHz. Here, the vertical axis represents frequency and the horizontal axis represents time. In addition, the data divided into subbands in the frequency band filter bank 2 are present in 10 pieces per subband in the time axis direction. That is, the total number of data is 320. Hereinafter, in the sample data, for example, the 0th subband (Sub0) of DC-0.75 kHz in the time axis direction will be represented as S0 (0). Further, the frequency band division filter bank 2 sends all 320 pieces of data to the scale factor calculation unit 3, as shown in FIG.

Further, the mode data output unit 6 detects the sample data of the digital audio signal in block units, detects the maximum absolute value of the sample data of each block, and further detects the maximum absolute value of the blocks continuous in time. The change amount is detected by taking the ratio of the values. Further, the mode data output unit 6 compares the threshold value sent from the control unit 8 with the ratio of the maximum absolute value (hereinafter, this operation is referred to as attack detection), and the ratio of the maximum absolute value is larger than the threshold value. , I.e., if the attack portion is detected, the mode data of the first mode, and if the ratio of the maximum absolute value is smaller than the threshold value, that is, if the attack portion is not detected, the mode data of the second mode, It is sent to the scale factor calculation unit 3 and the bitstream formation unit 7.

The scale factor calculator 3 groups the 320 pieces of data sent from the frequency band division filter bank 2 in units of 10 based on the mode data sent from the mode data output section 6. That is, the above 320 pieces of data are reorganized and divided into 32 data groups.

Here, the grouping mode in the scale factor calculation unit 3 will be described. The first mode is adopted when the attack portion is detected,
Further, as shown in FIG. 4, for example, S0 (0), S1
(0), S2 (0), S3 (0), S0 (1), S1
(1), S2 (1), S3 (1), S0 (2), S1
(2) is a mode in which data having similar amplitudes are grouped in units of 10 so as to form the first data group. Also, when viewed as a whole of the above data, as shown in FIG. It is designed to be finely divided in the axial direction, that is, to improve the resolution in the time axis direction.

Further, in the first mode, since the data are grouped so that the length of each data group in the time axis direction becomes short, the length in the time axis direction, that is, the so-called code length is set in the attack portion. It gets shorter. Therefore, as shown in FIG. 8, the quantization noise generated when the data grouped in the first mode is quantized appears in a narrow range in the time axis direction.

The second mode is adopted when the attack portion is not detected, and as shown in FIG. 5, for example, S0 (0) to S0 (9) form the first data group. As in the configuration, the sub-bands are in a mode that directly forms a data group, and
As shown in FIG. 7, the whole data is divided into fine pieces in the frequency axis direction, that is, the resolution in the frequency axis direction is increased. Further, as shown in FIG. 9, the quantization noise generated when the data grouped in the second mode is quantized spreads over the entire time domain as shown in FIG. Since there is no attack part, the quantization noise can be suppressed even if each subband is used as it is.

Further, comparing each data group defined in each of the above modes, as shown in FIGS. 6 and 7,
The first mode is short in the time axis direction, and
The second mode is long in the time axis direction. Therefore, hereinafter, the first mode is referred to as a short mode, and the second mode is referred to as a long mode.

Further, the scale factor calculation section 3 is
A scale factor is calculated for each data group grouped in the short mode or the long mode. In addition, the scale factor calculation unit 3 has
As shown in, the data group to which the scale factor is assigned is sent to the bit allocation unit 4.

The bit allocating unit 4 allocates a quantized bit for performing a quantization process to this data group according to the scale factor assigned to each data group. The data of the data group unit to which the quantization bit is assigned is sent to the quantization unit 5. The quantizer 5 also performs a quantization process on the data according to the quantization bit. This quantized data is sent to the bit stream forming unit 7 together with the corresponding scale factor.

The bit stream forming unit 7 forms a bit stream from the quantized data, the scale factor, and the mode data. The format of this bit stream is, for example, as shown in FIG. 10, composed of a plurality of frames F0 to F9, and each of the frames F0 to F9 is composed of six blocks B0 to B5.
Composed of. Further, each of these blocks B0 to B5
Is an auxiliary data area 56, a bit allocation data area 55 composed of 32 bit allocation data, a scale factor data area 54 composed of 32 scale factors, and 32 data group areas 53 corresponding to each data group. Composed of and. Further, ten pieces of data D0 to D9 are put in each of the data groups G0 to G31.

Further, in the auxiliary data area 56, the mode data, that is, the short mode or the long mode which is a mode in which the band-divided data by the frequency band division filter bank 2 when performing the quantization processing is grouped. Contains mode information. Further, the data group area 53 is arranged in order from the data group G0 including the lowest frequency data to the data group G31 including the highest frequency data. Since each of the blocks B0 to B5 has 320 pieces of band-divided data, each of the frames F0 to F9 has 1920 pieces of data.

Further, the bit stream forming section 7 is
As shown in FIG. 1, the bitstream is output from the bitstream output terminal.

A second example of the subband coding apparatus of the present invention is shown in FIG. This subband coding device is
A frequency band division filter bank 12 for dividing the sampled data of the input digital audio signal into data of each band divided into frequency bands, that is, subbands.
And the maximum absolute value in the frequency band unit is detected, the amount of change in the maximum absolute value between time-continuous frequency bands is detected, and the amplitude that outputs the mode data set based on the amount of change is detected. A mode data output unit 16 as a detection means, and a scale for grouping the data divided into the subbands based on the mode data and calculating the maximum value of the amplitude of the data of each data group, a so-called scale factor. Factor calculation unit 3,
A bit allocation unit 4 for allocating a quantization bit for quantizing each data to each data group based on the scale factor, and a quantization for quantizing each data for each data group according to the quantization bit. Part 5
And.

In the sub-band encoding device, the input data from the digital audio signal input terminal 9 is supplied to the frequency band division filter bank 12. Further, the output data from the frequency band division filter bank 12 is supplied to the mode data output unit 16.

The output data from the mode data output unit 16 is supplied to the scale factor calculation unit 3, and the mode data is supplied from the mode data output side to the scale factor 3 and bit stream formation unit 7. Further, the mode data output unit 16 includes the control unit 1
The threshold value for determining the mode data is supplied from 8.

The output data from the scale factor calculation unit 3 is supplied to the quantization unit 5 via the bit allocation unit 4. The output from the quantizing unit 5 is supplied to the bit stream forming unit 7.

Here, the frequency band division filter bank 1
Reference numeral 2 denotes the sample data of the digital audio signal sent from the audio processor or the like through the digital audio signal input terminal 9 from DC to 24 as described above.
The data is divided into data of each of 32 frequency bands, that is, subbands (Sub0 to Sub31) in 0.75 kHz steps up to the band of kHz. Further, the frequency band division filter bank 12 sends the data divided into the subbands to the mode data output unit 16.

The mode data output unit 16 detects the maximum absolute value of the data for each of the subbands, and also detects the amount of change by taking the ratio of the maximum absolute value between the frequency bands continuous in time. The mode data output unit 16 compares the threshold value sent from the control unit 18 with the ratio of the maximum absolute value,
That is, attack detection is performed. Here, when the ratio of the maximum absolute value is larger than the threshold value, that is, when the attack portion is detected, the mode data of the short mode is set, and when the ratio of the maximum absolute value is smaller than the threshold value, that is, the attack. If no part is detected, the mode data of the long mode is sent to the scale factor calculation part 3 and the bitstream formation part 7.

Further, according to each of the above sub-band encoding devices, the attack section is added to the digital audio signal input from the digital audio signal input terminal 9 by the mode data output section 6 or the mode data output section 16. When it is detected, the short mode is adopted for grouping the subband-divided data, and the scale factor calculation unit 3 selects the subband-divided data from the group G0 to the group G3 as shown in FIG.
It is divided into data groups up to 1. In the short mode, only data having similar amplitudes are grouped in the same data group. By doing so, it becomes possible to separately handle the data having a small amplitude and the data having a large amplitude of the digital audio signal having the attack portion.

When the attack section is not detected by the mode data output section 6 or the mode data output section 16 in the digital audio signal, the long mode is adopted for grouping the subband-divided data. In the scale factor calculation unit 3, as shown in FIG.
As shown in, the subband-divided data is subgrouped as it is into a data group, that is, a group G0 to a group G31.

Further, as described above, each sub-band coding apparatus calculates a scale factor for each data group, allocates a quantization bit to each data group based on this scale factor, and The data in each data group is quantized according to the quantization bit. Further, each of the subband coding devices described above forms a bitstream based on the quantized data and the data necessary for the quantization processing, and outputs the bitstream.

To decode the above bitstream,
A subband decoding device is used. As shown in FIG. 13, this sub-band decoding device detects the bit stream as shown in FIG. 10 for each block, extracts the mode data from the data of the auxiliary data area 56 of each block B0 to B5, and also allocates the bit allocation data. Based on the bit stream decoding unit 21 for extracting the data in the area 55, the bit allocation decoding unit 22 for extracting the bit allocation data from the data in the bit allocation data area 55, and the block 55 based on the bit allocation data and the mode data. Inverse quantization unit 23 for performing inverse quantization processing on the data D0 to D9 in the data group area 53 of
And a synthesis filter bank 24 for synthesizing the plurality of dequantized data and reproducing the digital audio signal before the subband coding processing.

In the subband decoding device, the input data from the bitstream input terminal 25 is supplied to the bitstream decoding unit 21. The output data from the bitstream decoding unit 21 is supplied to the bit allocation decoding unit 22 and the inverse quantization unit 23. Further, the output data from the bit allocation decoding unit 22 is supplied to the inverse quantization unit 23. The output data from the inverse quantization unit 23 is supplied to the synthesis filter bank 24 via the synthesis filter bank 24. The output data from the synthesis filter bank 24 is supplied to the digital audio signal output terminal 26.

According to the above sub-band decoding device, the bit stream obtained by subjecting the digital audio signal to the sub-band decoding process in each of the above-mentioned sub-band decoding devices is subjected to the decoding process to obtain the digital audio signal. Reproduce.

In this embodiment, an example in which a signal sampled at a sampling frequency of 48 kHz is used as the input digital audio signal has been described, but the present invention is not limited to this, and other sampling signals are used. Even if a digital audio signal sampled using frequency is used, the same effect as that of the present invention can be obtained by changing the number of subband divisions and the subband division interval to be optimal for the sampling frequency. .

In the present embodiment, the data obtained by band-dividing the sample data of the digital audio signal is grouped into a short mode for increasing the resolution in the time axis direction and a long mode for increasing the resolution in the frequency axis. I set the mode and, but not limited to these,
In accordance with the magnitude of the maximum absolute value in the attack section, for example, data obtained by dividing the band only between high frequency subbands may be grouped.

Further, as the short mode, the entire band is divided into eight, and a grouping mode in which the code length of each data group becomes 80 code length, and as the long mode, the entire band is divided into 32, An example of using the grouping mode in which the code length of each data group is 320 code lengths has been given, but the present invention is not limited to this. For example, the entire band is divided into 4 and the code length of each data group is divided. A grouping mode may be used in which is 160 code lengths.

[0065]

As described above, according to the present invention, a sub-band coding method having a band dividing step, a grouping step, and a quantizing step is used. The quantization step is performed by grouping the data obtained by dividing the audio signal into frequency bands in the band division step according to the amplitude of the digital audio signal in consideration of the spread in the time axis direction. In the above, the data is quantized corresponding to a portion of the digital audio signal in which the energy change is large, that is, the quantization processing of the signal having the small amplitude and the quantization processing of the signal having the large amplitude are separately performed. Can be performed, and the efficiency of subband encoding of the above data is improved. Furthermore, pre-echo can be suppressed.

In the subband coding method, when the amount of change in the maximum absolute value between temporally consecutive blocks is larger than a predetermined threshold in the grouping step, the length in the time axis direction is increased. By grouping the data obtained in the band dividing step with a short period, the grouping of the data can be performed based on a portion of the digital audio signal, which has a large energy change detected in block units. Therefore, in the quantizing step, the sample data is quantized corresponding to a portion of the digital audio signal in which the energy change is large, that is, the quantizing process of the signal having the small amplitude and the quantizing process of the signal having the large amplitude are performed. It is possible to perform the quantization processing separately, and the efficiency of subband encoding of the above data is improved. Furthermore, pre-echo can be suppressed.

In the subband coding method, when the amount of change in the maximum absolute value between temporally consecutive frequency bands is larger than a predetermined threshold in the grouping step, the length in the time axis direction is increased. By grouping the data obtained in the band dividing step by shortening the length, the grouping of the data can be performed based on the portion of the detected digital audio signal in which the energy change is large in each frequency band. it can. Therefore, in the quantization step, the data is quantized corresponding to a portion of the digital audio signal in which the energy change is large, that is, the quantization of the signal having a small amplitude and the quantization of the signal having a large amplitude are performed. It is possible to perform the encoding process separately, and the efficiency of subband encoding of the above data is improved. Furthermore, pre-echo can be suppressed.

Further, according to the present invention, band dividing means,
By using the sub-band coding device having the amplitude detecting means, the grouping means, and the quantizing means, the grouping means can perform the band division according to the amplitude of the input digital audio signal. By performing the grouping of the data obtained by dividing the digital audio signal in the frequency band in consideration of the spread in the time axis direction, in the quantizing means, the portion where the energy change of the digital audio signal is large is obtained. Correspondingly, it becomes possible to separately perform the quantization processing of the data, that is, the quantization processing of the signal having the small amplitude and the quantization processing of the signal having the large amplitude. Efficiency increases. Furthermore, pre-echo can be suppressed.

Further, in the sub-band coding apparatus, the amplitude detecting means detects the maximum absolute value of the sample data of the digital audio signal in block units of a fixed number of samples, and the time between consecutive blocks. The amount of change in the maximum absolute value is detected, and when the amount of change is larger than a predetermined threshold value, the grouping unit shortens the length in the time axis direction to obtain the data obtained by the band dividing unit. When the data is grouped, the grouping of the data can be performed based on a portion of the digital audio signal, which has a large energy change detected in block units. Therefore, in the quantizing means, the data is quantized corresponding to the portion of the digital audio signal in which the energy change is large, that is, the quantization of the signal having the small amplitude and the quantization of the signal having the large amplitude are performed. It is possible to perform the encoding process separately, and the efficiency of subband encoding of the above data is improved. Furthermore, pre-echo can be suppressed.

In the sub-band encoding device, the amplitude detecting means detects the maximum absolute value of the sample data of the digital audio signal in each of the frequency bands, and detects the maximum absolute value between frequency bands continuous in time. Detecting the amount of change in absolute value, the grouping means,
When the amount of change is larger than a predetermined threshold value, when the data obtained by the band dividing means is grouped by shortening the length in the time axis direction, the grouping of the data is performed by frequency. This can be performed based on the portion of the digital audio signal in which the energy change is large detected in band units. Therefore, in the quantizing means, the data is quantized corresponding to the portion of the digital audio signal in which the energy change is large, that is, the quantization of the signal having the small amplitude and the quantization of the signal having the large amplitude are performed. It is possible to perform the encoding process separately, and the efficiency of subband encoding of the above data is improved. Furthermore, pre-echo can be suppressed.

[Brief description of drawings]

FIG. 1 is a block diagram showing a main part of a first example of a subband coding apparatus to which a subband coding method of the present invention is applied.

FIG. 2 is a block diagram showing a main part of a second example of a subband coding apparatus to which the subband coding method of the present invention is applied.

FIG. 3 is a diagram illustrating the operation of a frequency band division filter bank of the subband encoding device.

FIG. 4 is a diagram explaining a short mode of a grouping mode of the subband coding device.

[Fig. 5] Fig. 5 is a diagram illustrating a long mode of a grouping mode of the subband coding device.

FIG. 6 is a diagram for explaining the resolution on the frequency axis and the time axis of the data group obtained in the short mode.

FIG. 7 is a diagram for explaining the resolution on the frequency axis and the time axis of the data group obtained in the long mode.

FIG. 8 is a diagram showing quantization noise generated when the sample data is quantized in the data group obtained in the short mode.

FIG. 9 is a diagram showing quantization noise generated when the sample data is quantized in the data group obtained in the long mode.

FIG. 10 is a diagram illustrating a format of a bitstream obtained by the subband encoding device.

FIG. 11 is a diagram illustrating a grouping process in the short mode.

FIG. 12 is a diagram illustrating a grouping process in the short mode.

FIG. 13 is a block diagram showing a main part of a subband decoding device that decodes a signal encoded by the subband encoding device.

FIG. 14 is a block diagram showing a main part of a conventional subband encoding device.

FIG. 15 is a block diagram showing a main part of a conventional subband decoding device.

FIG. 16 is a block diagram illustrating a conventional subband coding method.

[Fig. 17] Fig. 17 is a diagram for explaining the operation of the conventional subband encoding device.

FIG. 18 is a diagram for explaining the operation of the frequency band filter of the conventional subband encoding device.

[Fig. 19] Fig. 19 is a diagram for explaining the operation of the scale factor calculation unit of the conventional subband encoding device.

[Fig. 20] Fig. 20 is a diagram for describing the operation of the bit allocation unit of the conventional subband encoding device.

[Fig. 21] Fig. 21 is a diagram for explaining the operation of the bitstream forming unit of the conventional subband encoding device.

[Fig. 22] Fig. 22 is a diagram for describing a problem of the conventional subband encoding device.

[Fig. 23] Fig. 23 is a diagram for describing a problem of the conventional subband encoding device.

[Fig. 24] Fig. 24 is a diagram for describing a problem of the conventional subband encoding device.

[Explanation of symbols]

 1 block delay unit 2 and 12 frequency band division filter bank 3 scale factor calculation unit 4 bit allocation unit 5 quantization unit 6 and 16 mode data output unit 7 bit stream formation unit 8 and 18 control unit

Claims (6)

[Claims] 1. A sub-band encoding method for dividing sample data of a digital audio signal into data of a plurality of frequency bands and encoding the data, wherein the sample data of the input digital audio signal is a data of a plurality of frequency bands. A band division step of dividing the data into the frequency bands, a grouping step of grouping the data divided into the frequency bands according to the amplitude of the input digital audio signal, and the data group obtained in the grouping step. And a quantization step of quantizing the data for each subband encoding method. 2. The maximum absolute value of the sample data of the digital audio signal is detected in block units of a fixed number of samples, and in the grouping step, a change amount of the maximum absolute value between the blocks that are temporally continuous. 2. The subband coding method according to claim 1, wherein the data obtained in the band division step is grouped by shortening the length in the time axis direction when is larger than a predetermined threshold value. 3. A maximum absolute value is detected in each of the frequency bands of the sample data of the digital audio signal, and in the grouping step, a change amount of the maximum absolute value between frequency bands continuous in time is predetermined. 2. The subband coding method according to claim 1, wherein the data obtained in the band division step is grouped by shortening the length in the time axis direction when the value is larger than the threshold value. 4. A sub-band coding apparatus for dividing data of a digital audio signal into sample data of a plurality of frequency bands and coding the sample data, wherein the input digital audio signal data is converted into a plurality of frequency band data. Band dividing means for dividing, amplitude detecting means for detecting the amplitude of the input digital audio signal, and data divided by the band dividing means are defined based on the amplitude detected by the amplitude detecting means. A subband coding apparatus comprising grouping means for grouping the data groups according to each other, and quantizing means for quantizing the data for each of the data groups obtained by the grouping means. 5. The amplitude detecting means detects the maximum absolute value of the sample data of the digital audio signal on a block-by-block basis with a fixed number of samples, and determines the amount of change in the maximum absolute value between temporally consecutive blocks. 5. The sub-band encoding device according to claim 4, wherein the grouping means detects the change amount and reduces the length in the time axis direction to group the data when the change amount is larger than a predetermined threshold value. 6. The amplitude detecting means detects the maximum absolute value in each of the frequency bands of the sample data of the digital audio signal and also detects the amount of change in the maximum absolute value between frequency bands continuous in time. 5. The subband coding apparatus according to claim 4, wherein the grouping means groups the data by shortening the length in the time axis direction when the change amount is larger than a predetermined threshold value.