JP2000078017A - Decoding method and decoding device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To define the upper limit value which can be added together in a desired section when the reproduction level is controlled or the filter effect is obtained by detecting the maximum value in a prescribed range in regard to the normalized information obtained by a number using several bits of a two-dimensional block concerning the time and frequency. SOLUTION: The MDCT coefficient of every band and the used block size information are given to the input terminals 908 and 910 respectively. An adaptive bit allocation decoding circuit 906 cancels the bit allocation based on the adaptive bit allocation information. The inverse orthogonal transform units 903, 904 and 905 transform the signals existing on a frequency axis into those existing on a time base. The signals existing on the time base of a partial band are decoded into the full band signals by the band synthesizing filters 902 and 901. An adder 909 adds the numeric value received from a normalized information control circuit 911 to the scale factor value of a unit block. The same numeric value is added to all unit blocks to perform the level control and the respective different numeric values are added to every block to obtain the filter effect.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a decoding method and a decoding device.

[0002]

2. Description of the Related Art There are various conventional methods and devices for efficient encoding of audio signals.
Three examples will be described. The audio signal in the time domain is divided into blocks for each unit time, and the signal on the time axis for each block is converted into a signal on the frequency axis (orthogonal transform), divided into a plurality of frequency bands, and encoded for each band. There is a transform coding method which is one of the blocking frequency band division methods. Sub-band coding (SBC), which is one of the non-blocking frequency band division methods for dividing and encoding a time-domain audio signal into a plurality of frequency bands without dividing the signal into unit frequency units. : Subband Codin
g) There is a method. There is also a high-efficiency coding method that combines the above-described band division coding and transform coding. In this case, for example, after performing band division by the above-described band division coding system, the signal of each band is orthogonally transformed into a signal in the frequency domain by the above-described transformation encoding system, and this orthogonal transformation is performed. Encoding is performed for each band.

Here, as a band division filter used in the above-mentioned band division coding system, for example, QMF (Q
There is a filter such as uadrature Mirror filter. This filter is a 1976 RECrochiere Di
gital coding of speech insubbands Bell Syst.Tech.
J. Vol.55, No.8, 1976. Also, ICA
SSP 83, BOSTON Polyphase Quadrature filters-A new
The subband codingtechnique Joseph H. Rothweiler has a polyphase quadrature filter (Polyphase Quadrat
ure Filter: a polyphase quadrature filter) and the like, and an equal bandwidth filter dividing method and apparatus are described.

As the above-mentioned orthogonal transform, for example, an input audio signal is divided into blocks in a predetermined unit time (frame), and a fast Fourier transform (F
FT), a discrete cosine transform (DCT), a modified DCT transform (MDCT), and the like, to transform the time axis to the frequency axis. Regarding the above MDCT, ICASSP 1987 Subband / Transfor
m Coding Using FilterBank Designs Based on Time Do
main Aliasing Cancellation JPPrincen ABBradley
Univ. Of Surrey Royal Melbourne Inst. Of Tech.

[0005] Further, as a frequency division width when quantizing each frequency component divided into frequency bands, there is band division in consideration of human auditory characteristics. That is, an audio signal may be divided into a plurality of bands (for example, 25 bands) with a bandwidth generally called a critical band (critical band) such that the bandwidth becomes wider as the band becomes higher.
Also, when encoding data for each band at this time,
Encoding by predetermined bit allocation for each band or adaptive bit allocation for each band is performed. For example, when MDCT coefficient data (signal components in a two-dimensional block relating to time and frequency) obtained by the above-described MDCT processing is encoded by the above-described bit allocation, the MDCT coefficient data obtained by the above-described MDCT processing for each block is obtained. MDC for each band
Encoding is performed on the T coefficient data with an adaptive number of allocated bits.

[0006] Further, in encoding for each band, a so-called block floating process for realizing more efficient encoding is performed by normalizing and quantizing each band. For example, when encoding the MDCT coefficient data obtained by the above-mentioned MDCT processing, quantization is performed by performing normalization corresponding to the maximum value of the absolute value of the MDCT coefficient described above for each band. Thus, more efficient encoding is performed. For normalization, numbering corresponding to information of a plurality of sizes is defined in advance, and this number is used as normalization information. The information on the magnitude of the normalization prepared in advance is numbered at a constant proportion.

Conventionally, the following two methods are known as the above-mentioned bit allocation method and the device for the same.

[0008] IEEE Transactions of Accoustics, Speec
h, and Signal Processing, vol.ASSP-25, No.4, August 19
In 77, bit allocation is performed based on the magnitude of the signal for each band. Also ICASSP 1980 The critical band
coder--digital encoding of the perceptual requireme
At the nts of the auditory system MA Kransner MIT,
A method is described in which a required signal-to-noise ratio is obtained for each band and fixed bit allocation is performed by using auditory masking.

A signal which has been subjected to high-efficiency encoding by the above-described method is decoded by the following method. First, a signal subjected to high-efficiency encoding is calculated as MDCT coefficient data (signal components in a two-dimensional block relating to time and frequency) using bit allocation information, normalization information, and the like for each band. Become. This MDCT coefficient data is subjected to a so-called inverse orthogonal transform, and is converted into data in a time domain. If band division has been performed by a band division filter at the time of encoding, synthesis is further performed using a band synthesis filter. By these operations, the original data in the time domain is decoded.

With respect to the number of the normalization information in the above-described high efficiency coding, a signal subjected to the high efficiency coding is decoded by adding or subtracting a desired value to the value. For a signal in the time domain, there is known a method of adjusting the magnitude of the amplitude, that is, the reproduction level, and obtaining a so-called filter effect. According to this method, since only the addition process is performed for the number of the normalized information, the configuration of the system can be easily realized, and there is no need to perform unnecessary decoding or encoding.
It is possible to prevent quality deterioration.

A signal obtained by decoding a signal which has been subjected to high-efficiency encoding by the above-described method is a so-called general digital audio signal. For example, a digital audio signal is converted into an analog signal, and the obtained analog audio signal is converted into an analog signal. In the case of recording on an analog recording medium, as an index for adjusting the input level of the analog recording device,
There is a method of detecting the maximum value of a digital audio signal. This detection method is realized by comparing the magnitude of the data of the digital audio signal in the amplitude direction for each point. However, when importance is attached to the speed of detection, the magnitude of all data is not compared. Then, certain data is selected, and the data is compared. In this case, if the amount of the selected data is large, the maximum value can be detected with higher accuracy. However, the speed of the detection is slow. If the amount of the selected data is small, the speed of the detection is increased, but the accuracy is reduced.

[0012]

By the way, by adding a desired value to the value of the number of the normalization information in the above-described high-efficiency coding, the signal which has been subjected to the high-efficiency coding is decoded. When performing a method of adjusting the magnitude of the amplitude, that is, the reproduction level, or the so-called filtering effect of the signal in the time domain, since the number of the normalization information in the high-efficiency coding actually has an upper limit, the addition is performed. There is also an upper limit on the possible values. If the addition process is performed without recognizing the upper limit value that can be added, a normalized information number exceeding the upper limit may be obtained depending on the added value. Processing to suppress the number to the upper limit is required. When such processing is performed, there is a block in which the effect of the desired added value cannot be obtained, and the effect varies from block to block, and the desired effect cannot be obtained. There is a possibility that an imbalanced operation between blocks may cause a change in sound quality in a reproduced signal after decoding.

In a case where a signal which has been subjected to high-efficiency encoding is decoded, the obtained digital audio signal is converted into an analog signal, and the obtained analog audio signal is recorded on an analog recording medium, an analog recording apparatus is used. When trying to detect the maximum value of the digital audio signal as an index for adjusting the input level of the digital audio signal, decoding must be performed to detect the maximum value because the existing method uses the digital audio signal in the time domain. Occurs,
The processing process becomes longer and the detection speed becomes slower.

Therefore, an object of the present invention is to adjust the reproduction level,
It is an object of the present invention to propose a decoding method and a decoding device capable of clarifying an upper limit value that can be added in a desired section when a filter effect is realized.

Another object of the present invention is to realize the detection of the maximum level of a digital signal obtained by decoding a signal which has been subjected to high-efficiency encoding, with a smaller processing step and at a high speed. It is intended to propose a possible decoding method and decoding device.

[0016]

According to a first decoding method of the present invention, an input digital signal is decomposed into a plurality of frequency band components to obtain signal components in a plurality of two-dimensional blocks relating to time and the number of local waves. For each two-dimensional block relating to time and frequency, one of several positive values that are numbered using several bits in advance is selected as normalization information based on signal components in the two-dimensional block. In this way, a number using the corresponding bits is used as normalization information, and a quantization coefficient representing the characteristic of the signal component in the two-dimensional block is obtained for each two-dimensional block in terms of time and frequency. A bit allocation amount is determined based on the quantization coefficient, a signal component in the block is quantized by the normalization information and the bit allocation amount for each two-dimensional block relating to time and frequency, and information is compressed. The information compression parameter for each two-dimensional block related to the frequency is obtained, and the number information of the two-dimensional blocks to be effective is selected from a predetermined value of several bits, and the signal component for each two-dimensional block related to the time and frequency of the information compression is obtained. Is decoded using information compression parameters for each two-dimensional block regarding time and frequency,
In a decoding method in which a level is adjusted for each two-dimensional block by adding a desired value to normalization information based on a number using several bits for each two-dimensional block during decoding, With regard to the normalized information by a number using several bits, the maximum value is detected within a desired range, and the limit value at which the level adjustment is effective can be recognized based on the detected maximum value. It was done.

According to the first aspect of the present invention, the maximum value in the desired range is detected for the normalized information by the number using several bits of the two-dimensional block, and the value of the detected maximum value is detected. To recognize the limit value at which the level adjustment is effective.

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A first aspect of the present invention is to decompose an input digital signal into a plurality of frequency band components, obtain signal components in a plurality of two-dimensional blocks relating to time and the number of local waves, and obtain signals relating to time and frequency. For each two-dimensional block, based on the signal components in the two-dimensional block, normalization is performed by selecting, as normalization information, one of several positive values numbered using several bits in advance. Then, a number using the corresponding bits is used as normalization information, a quantization coefficient representing the characteristic of the signal component in the two-dimensional block is obtained for each two-dimensional block relating to time and frequency, and the quantization coefficient is determined based on the quantization coefficient. The bit allocation amount is determined, the signal components in the block are quantized by the normalization information and the bit allocation amount for each two-dimensional block relating to time and frequency, and the information is compressed. Obtain information compression parameter for each click,
The number information of the two-dimensional blocks to be validated is selected from a predetermined number of bits, and the signal components of the two-dimensional blocks related to the time and frequency of the information compression are converted to the information compression parameters for the two-dimensional blocks related to the time and frequency. And a level adjustment for each two-dimensional block by adding a desired value to the normalized information based on a number using several bits for each two-dimensional block at the time of decoding. In the method, for the normalized information by number using several bits of a two-dimensional block,
This is a decoding method in which the maximum value is detected in a desired range and the limit value at which the level adjustment is effective can be recognized based on the detected maximum value.

According to a second aspect of the present invention, there is provided a band dividing means for dividing an input digital signal into a plurality of frequency band components, encoding in a plurality of two-dimensional blocks relating to time and frequency by orthogonally transforming the signal, and / or Orthogonal transform means for obtaining signal components for analysis, and several positive values previously numbered using several bits based on the signal components in the two-dimensional block for each two-dimensional block relating to time and frequency A normalization data calculating means for performing normalization by selecting one of them as normalization information and obtaining a number using the corresponding few bits as the normalization information; and a two-dimensional block for time and frequency. Quantization coefficient calculation means for obtaining a quantization coefficient representing a characteristic of a signal component in a two-dimensional block; bit allocation calculation means for determining a bit allocation amount based on the quantization coefficient; Compression encoding means for quantizing the signal components in the block by the normalized data and the bit allocation amount for each two-dimensional block relating to the number and compressing the information, and the information compression parameter for obtaining the information compression parameter for each two-dimensional block relating to time and frequency Determining means, valid two-dimensional block number information determining means for selecting the number information of the two-dimensional blocks to be valid from a predetermined number of bits, and signal information in the two-dimensional block relating to time and frequency of information compression. Decoding means for decoding using the information compression parameter for each two-dimensional block relating to time and frequency, and normalization information adjustment for adding a desired value to the normalization information to enable level adjustment for each two-dimensional block Means for normalizing information by a number using several bits of a two-dimensional block. A decoding device comprising: detecting means for detecting a maximum value in a desired range; and display means for displaying a parameter capable of recognizing a limit value at which level adjustment is effective based on the detected maximum value. It is.

According to a third aspect of the present invention, an input digital signal is decomposed into a plurality of frequency band components to obtain signal components in a plurality of two-dimensional blocks relating to time and frequency. Based on the signal components in the two-dimensional block, normalization is performed by selecting, as normalization information, one of several positive values that have been numbered using several bits in advance. Numbers using several bits are used as normalization information, quantized coefficients representing the characteristics of signal components in the two-dimensional block are obtained for each two-dimensional block relating to time and frequency, and the bit allocation amount is determined based on the quantized coefficients. Then, for each two-dimensional block related to time and frequency, the signal components in the block are quantized by the normalization information and the bit allocation amount to compress the information, and the two-dimensional blocks related to time and frequency are used.
The information compression parameter for each dimensional block is obtained, and the number information of the two-dimensional blocks to be effective is selected from a predetermined value of several bits. And decoding using the information compression parameter for each two-dimensional block related to frequency and frequency, and adding a desired value to normalized information based on a number using several bits of the two-dimensional block at the time of decoding. In the decoding method in which the level is adjusted for each dimensional block, the position of the maximum value in a desired range is detected for the normalized information by the number using several bits of the two-dimensional block, and the detected maximum value This is a decoding method in which a position where level adjustment is effective is recognized based on the position.

According to a fourth aspect of the present invention, there is provided a band dividing means for dividing an input digital signal into a plurality of frequency band components, encoding in a plurality of two-dimensional blocks relating to time and frequency by orthogonally transforming the signal, and / or Orthogonal transform means for obtaining signal components for analysis, and several positive values previously numbered using several bits based on the signal components in the two-dimensional block for each two-dimensional block relating to time and frequency A normalization data calculating means for performing normalization by selecting one of them as normalization information and obtaining a number using the corresponding few bits as the normalization information; and a two-dimensional block for time and frequency. Quantization coefficient calculation means for obtaining a quantization coefficient representing a characteristic of a signal component in a two-dimensional block; bit allocation calculation means for determining a bit allocation amount based on the quantization coefficient; Compression encoding means for quantizing the signal components in the block by the normalized data and the bit allocation amount for each two-dimensional block relating to the number and compressing the information, and the information compression parameter for obtaining the information compression parameter for each two-dimensional block relating to time and frequency Determining means, valid two-dimensional block number information determining means for selecting the number information of the two-dimensional blocks to be valid from a predetermined number of bits, and signal information in the two-dimensional block relating to time and frequency of information compression. Decoding means for decoding using the information compression parameter for each two-dimensional block relating to time and frequency, and normalization information adjustment for adding a desired value to the normalization information to enable level adjustment for each two-dimensional block Information using numbers using several bits of a two-dimensional block in a decoding device having means A decoding device comprising: a detecting unit for detecting a position of a maximum value in a desired range; and a display unit for displaying a parameter capable of recognizing a position where level adjustment is effective based on the detected position of the maximum value. .

[Specific Example of Embodiment of the Invention] Hereinafter, a specific example of an embodiment of the present invention will be described with reference to the drawings.

In a specific example of this embodiment, an input digital signal such as an audio PCM signal is subjected to band division coding (SBC), adaptive transform coding (ATC: Adaptive Transform).
ormCoding) and adaptive bit allocation techniques. This technique will be described with reference to FIG.

In the specific high-efficiency coding apparatus shown in FIG. 1, the input digital signal is divided into a plurality of frequency bands, and at the same time, orthogonal transform is performed for each frequency band, and the obtained spectrum data on the frequency axis is obtained. In the low frequency range, adaptively, for each so-called critical bandwidth (critical band) in consideration of human hearing characteristics described later, and in the middle and high frequency ranges, each critical bandwidth is subdivided in consideration of the block flow efficiency. Bits are assigned and encoded. Usually, this block is a quantization noise generating block. Further, in a specific example of the embodiment of the present invention, the block size (block length) is adaptively changed according to the input signal before the orthogonal transform.

That is, in FIG. 1, when the sampling frequency is 44.1 kHz, for example,
An audio PCM signal of 22 kHz is supplied.
This input signal is divided into a band of 0 to 11 kHz by a band division filter 101 such as a so-called QMF filter.
It is divided into the band of kHz to 22kHz, and it is 0 to 11kHz.
The signals in the band are similarly divided into 0 to 5.5 kHz band and 5.5 by a band division filter 102 such as a so-called QMF filter.
It is divided into a kHz to 11 kHz band.

The 11 from the above-described band division filter 101
A signal in the band of 22 kHz to 22 kHz is an MDCT (Modified Discrete Cosine Transform) which is an example of an orthogonal transform circuit.
The signal in the 5.5 kHz to 11 kHz band from the above-described band division filter 102 is sent to the MDCT circuit 104, and the 0 to 5.5 kHz band signal from the above-described band division filter 102 is sent to the MDCT circuit 105. By being sent, each is subjected to MDCT processing. Each of the MDCT circuits 103, 104, and 105 includes a block determination circuit 109, 110, 111 provided for each band.
MDCT processing is performed based on the block size (block length) (information compression parameter) determined by the above.

Here, each of the MDCT circuits 103, 104,
FIG. 2 shows a specific example of a standard input signal for a block for each band to be supplied to 105. In the specific example of FIG. 2, each of the three filter output signals has a plurality of orthogonal transform block sizes (information compression parameters) independently for each band, and switches the time resolution according to the time characteristics, frequency distribution, and the like of the signal. I am trying to be.
If the signal is quasi-stationary in time, the orthogonal transform block size is increased to 11.6 ms, ie, as in the long mode of FIG. 2A, and if the signal is non-stationary, the orthogonal transform block size is increased. (Information compression parameter)
It is divided into four parts. Like the short mode of FIG. 2B,
When the whole is divided into four and 2.9 mS or the middle mode A in FIG. 2C and the middle mode B in FIG.
Division into 5.8 mS, division of one part into four parts, and a time resolution of 2.9 mS allow adaptation to an actual complicated input signal. It is obvious that the division of the orthogonal transform block size is more effective if a more complicated division is performed if the scale of the processing device permits. This block size is determined by the block size determination circuit 10 in FIG.
9, 110, 111, and each MDCT circuit 10
3, 104, 105 and the bit allocation calculation circuit 118, and output from the output terminals 113, 115, 117 as block size information of the corresponding block.

Referring again to FIG. 1, each MDCT circuit 10
Spectrum data or MDCT coefficient data on the frequency axis obtained by performing MDCT processing in 3, 104 and 105 (signal components in a two-dimensional block related to time and frequency)
In the low band, the critical band is divided for each so-called critical band (critical band), and in the middle and high bands, the critical bandwidth is subdivided in consideration of the effectiveness of block floating, and the adaptive bit allocation coding circuits 106, 107, 108, And transmitted to the bit assignment calculation circuit 118. The critical band is a frequency band divided in consideration of human auditory characteristics, and a band of a pure tone when the pure tone is masked by a narrow band noise near the frequency of the pure tone. That is. In this critical band (critical band), the higher the band, the wider the bandwidth,
The entire frequency band of 0 to 22 kHz described above is divided into, for example, 25 critical bands.

The bit assignment calculation circuit 11 in FIG.
Reference numeral 8 denotes masking for each divided band in consideration of the above-described critical band and block floating, based on the above-described block size (information compression parameter) information and spectrum data or MDCT coefficient data, in consideration of a so-called masking effect or the like. The amount and the energy or peak value for each of the divided bands are calculated, and the number of allocated bits (bit allocation amount) is calculated for each band based on the calculation result. The adaptive bit allocation coding circuits 106 and 107 in FIG. , 10
8. In these adaptive bit allocation coding circuits 106, 107, and 108, according to the above-mentioned block size (information compression parameter) information and the number of bits allocated to each divided band in consideration of the critical band and the block floating, Each spectrum data or MD
The CT coefficient data is requantized (normalized and quantized). The data encoded in this way is
It is taken out via the output terminals 112, 114, 116 in FIG. For the sake of convenience in the following description, each of the above-mentioned critical bands and each divided band that takes into account block floating, which is a unit of bit allocation, is referred to as a unit block.

Next, a specific method of the bit allocation performed by the bit allocation calculating circuit 118 in FIG. 1 will be described. FIG. 3 is a block circuit diagram showing a schematic configuration of one specific example of the bit allocation calculating circuit 118 in FIG. 1 described above. 3, an input terminal 301 is connected to the MDCT circuits 103 and 104 in FIG.
Spectrum data or MDC on the frequency axis from 105
T coefficient and the above-described block determination circuit 1 in FIG.
Block size information from 09, 110 and 111 is supplied. Hereinafter, in the system of the bit allocation calculating circuit 118 shown in FIG.
Processing is performed using constants, weighting functions, and the like that are adapted to the above-described block size information.

In FIG. 3, spectrum data or MDCT coefficients on the frequency axis inputted from an input terminal 301 are as follows:
The energy is sent to the band-by-band energy calculation circuit 302, and the energy for each unit block is calculated by, for example, calculating the sum of the amplitude values in the unit block. Instead of the energy for each band, a peak value or an average value of the amplitude value may be used. As an output from the energy calculation circuit 302, for example, the spectrum of the sum of each band is shown as SB in FIG. However, in FIG. 4, the number of divisions by unit blocks is represented by 12 blocks (B1 to B12) to simplify the illustration. In the energy calculation circuit 302,
Indicates the block floating state of the unit block.
A scale factor value that is normalized data (information compression parameter) is also determined. Specifically, for example, several positive values are prepared in advance as candidates for the scale factor value, and the minimum value is selected from among the positive values that are equal to or more than the maximum value of the absolute value of the spectrum data or MDCT coefficient in the unit block. Is adopted as the scale factor value of the unit block. The scale factor value is numbered using several bits in a form corresponding to the actual value, and the number is stored in a ROM or the like (not shown).
May be stored in the memory. The scale factor value corresponding to the number is defined to have a value in the order of the number, for example, at an interval of 2 dB. Here, as for the scale factor value determined by the above method in a certain unit block, the above-described number corresponding to the determined value is used as sub-information indicating the scale factor of the unit block.

Next, in order to consider the influence on the so-called masking of the above-described spectrum SB obtained by the above-described energy calculating circuit 302, a convolution such as multiplying the spectrum SB by a predetermined weighting function and adding the result is performed. (Convolution) processing is performed. Therefore, the output of the energy calculation circuit 302 for each band, that is, each value of the spectrum SB, is sent to the convolution filter circuit 303. The convolution filter circuit 303 includes, for example, a plurality of delay elements for sequentially delaying input data, a plurality of multipliers for multiplying outputs from these delay elements by a filter coefficient (weighting function), and a sum of outputs of the respective multipliers. And a sum adder which takes By this convolution process, the sum of the parts indicated by the dotted lines in FIG. 4 is obtained.

Next, the output of the convolution filter circuit 303 is sent to the subtractor 304. The subtractor 304 is
A level α corresponding to an allowable noise level (quantized coefficient) described later in the convolved region is obtained. The level α corresponding to the permissible noise level (permissible noise level) is, as described later, a level at which the permissible noise level of each critical band is obtained by performing inverse convolution processing. It is. Here, an allowance function (a function expressing a masking level) for obtaining the level α is supplied to the subtractor 304. The level α is controlled by increasing or decreasing the allowable function. The permissible function is supplied from the (n-ai) function generation circuit 305 described below.

That is, the level α corresponding to the allowable noise level can be obtained by the following equation (1), where i is a number sequentially given from the low band of the critical band.

[0035]

Α = S− (n−ai)

In the equation (1), n and a are constants and a
> 0, S is the intensity of the convolved spectrum,
(N-ai) in the equation (1) is an allowable function. N as an example
= 38, a = 1 can be used.

In this way, the above-mentioned level α is obtained, and this data is transmitted to the division circuit 306. The division circuit 306 is for performing inverse convolution of the level α in the convolved region. Therefore, by performing the inverse convolution processing, a masking spectrum can be obtained from the level α. That is, this masking spectrum becomes an allowable noise spectrum. Although the above-described inverse convolution processing requires a complicated operation, the inverse convolution is performed using the simplified divider 306 in this specific example.

Next, the above-mentioned masking spectrum is transmitted to the subtraction circuit 309 via the synthesis circuit 308. Here, the output from the above-described energy calculation circuit 302 for each band, that is, the above-described spectrum SB is supplied to the subtraction circuit 309 via the delay circuit 310. Therefore, by performing the subtraction operation of the above-described masking spectrum and the spectrum SB in the subtraction circuit 309, as shown in FIG.
The level below the level indicated by the level of the masking spectrum MS is masked.

By the way, at the time of synthesizing by the synthesizing circuit 308 described above, data indicating a so-called minimum audible curve RC which is a human auditory characteristic as shown in FIG. The above-mentioned masking spectrum MS can be synthesized. At this minimum audible curve, if the absolute noise level is below this minimum audible curve, the noise will not be heard. This minimum audible curve will differ depending on the playback volume during playback, for example, even if the coding is the same, but in a realistic digital system, for example, music enters the 16-bit dynamic range very little. Since there is no difference, for example, if quantization noise in the most audible frequency band around 4 kHz is not heard, it is considered that quantization noise below the level of the minimum audible curve is not heard in other frequency bands. Therefore, it is assumed that the system is used in such a manner that the noise of the word length of the system, for example, around 4 kHz cannot be heard. Is obtained, the permissible noise level in this case can be up to the shaded portion in FIG. In this specific example, the 4 kHz level of the above-described minimum audible curve is adjusted to the lowest level corresponding to, for example, 20 bits. FIG. 6 also shows the signal spectrum SS.

Thereafter, in the allowable noise correction circuit 311, the allowable noise level in the output from the subtraction circuit 309 is corrected based on, for example, information on the equal loudness curve. Here, the equal loudness curve is a characteristic curve relating to human auditory characteristics. For example, the loudness curve is obtained by calculating the sound pressure of sound at each frequency that sounds as loud as the pure tone of 1 kHz, and connecting the curves with each other. Also called a sensitivity curve. Further, this equal loudness curve draws substantially the same curve as the minimum audible curve RC shown in FIG. In this equal loudness curve, for example, at around 4 kHz, even if the sound pressure falls by 8 to 10 dB below 1 kHz, it sounds as large as 1 kHz, and conversely, at around 50 Hz, the sound pressure must be about 15 dB higher than the sound pressure at 1 kHz. It doesn't sound the same size. For this reason, it can be seen that noise (allowable noise level) (quantization coefficient) exceeding the level of the minimum audible curve described above preferably has a frequency characteristic given by a curve corresponding to the equal loudness curve. . From this, it can be understood that correcting the above-mentioned allowable noise level in consideration of the above-mentioned equal loudness curve is suitable for human auditory characteristics. Through the series of processing up to this point, the allowable noise correction circuit 310 calculates bits to be assigned to each unit block based on various parameters such as the above-described masking and auditory characteristics.

The data output from the allowable noise correction circuit 311 is output from the output terminal 312 as the output of the bit assignment calculation circuit 118 in FIG.

That is, in the bit allocation calculating circuit 118 in FIG. 1, the data obtained by processing the orthogonal transform output spectrum by the sub-information as the main information and the block as the sub-information by the system shown in FIG. A scale factor indicating the floating state and a word length indicating the word length are obtained, and based on the maximum values, the adaptive bit allocation encoding circuit 106 in FIG.
At 107 and 108, requantization is actually performed, and encoding is performed in a form conforming to the encoding format.

Here, the normalization information adjustment circuit 119 of FIG. 1 will be described. As described above, for the scale factor value that is the normalized data (information compression parameter), several positive values are prepared in advance as scale factor value candidates, and the spectrum data or the MDCT coefficient in the unit block are prepared from among them. Among the values that are greater than or equal to the maximum value of the absolute value of, the smallest one is adopted as the scale factor value of the unit block, and the scale factor value is numbered using several bits in a form corresponding to the actual value. And the number is used as sub-information indicating the scale factor of the unit block. The scale factor value corresponding to the number is defined to have a value in the order of the number, for example, at an interval of 2 dB. Therefore, for example, by manipulating the number of the sub-information indicating the scale factor, the level of the unit block can be adjusted by 2 dB. Normalization information adjustment circuit 11
Reference numeral 9 denotes a circuit for outputting a numerical value for performing the level adjustment.

The adders 120, 121, and 122 add the numerical value from the normalization information adjusting circuit 119 for each band to the scale factor value of the unit block. When the numerical value output from the normalization information adjustment circuit 119 is a negative number, the adders 120, 121,
122 acts as a subtractor. In other words, the level adjustment of 2 dB can be performed only by adding the output of the same numerical value from the normalization information adjustment circuit 119 to all the unit blocks. At this time, the addition result is restricted so that it falls within the range of the scale factor value defined in the format.
In the above-described example, the output of all the same numerical values from the normalization information adjustment circuit 119 is added to all the unit blocks. However, by adding the numerical values independently for each unit block, Can be adjusted, and as a result, a so-called filter effect can be realized. In this case, the normalization information adjustment circuit 119 outputs a unit block number and an added value for the unit block. This normalization information adjustment circuit 11
The effect using No. 9 can also be realized at the time of decoding described later.

Next, an encoding format of data to be actually encoded will be described with reference to FIG. The numerical value shown on the left side of FIG. 7 represents the number of bytes, and in this embodiment, 212 bytes are used as a unit of one frame.

In the position of 0 byte located at the forefront in FIG. 7, the block determination circuits 109, 1 in FIG.
The block size information of each band determined in 10 and 111 is recorded.

Information on the number of unit blocks to be recorded is recorded at the next byte position. This is because, for example, the bit allocation becomes 0 and the recording is unnecessary in many cases as the frequency becomes higher on the higher frequency side by a series of bit allocation calculation circuits. A lot of bits are allocated to the middle and low frequencies that have a large effect. In addition, the number of unit blocks in which bit allocation information is double-written and the number of unit blocks in which scale factor information is double-written are recorded at the position of this 1 byte. Double writing is a method for recording the same data as data recorded at a certain byte position in another location for error correction. The more this double-write information is, the higher the error resistance is, but the less this information is, the more bits that can be used for the spectrum data. For each of the above-mentioned bit allocation information and scale factor information, the number of unit blocks for which double writing is performed is set independently, and the strength against errors and the number of usable bits for spectrum data are adjusted. ing. Note that for each piece of information, the correspondence between the code in the prescribed bits and the number of unit blocks is determined in advance as a format. Specifically, as shown in FIG. 8, three bits out of the eight bits at the position of one byte are used as information of the number of unit blocks to be actually recorded, and two bits out of the remaining five bits are used as bit allocation information. The number of unit blocks on which double writing is performed, and the number of unit blocks on which double writing of scale factor information is performed for the remaining three bits are recorded.

At the position from the second byte in FIG. 7, bit allocation information of a unit block is recorded. For recording bit allocation information, for example, 4
The use of bits is defined as a format.
As a result, bit allocation information for the number of unit blocks actually recorded in FIG. 7 described above is recorded in order from the 0th unit block.

After the data of the bit allocation information recorded by the above method, the scale factor information of the unit block is recorded. For the recording of scale factor information, for example, use of 6 bits for one unit block is defined as a format. Thus, the scale factor information is recorded in the same order as the recording of the bit allocation information, starting from the 0th unit block and by the number of the unit blocks to be actually recorded.

After the scale factor information thus recorded, the spectrum data of the unit block is recorded. The spectrum data is also recorded in order from the 0th unit block by the number of unit blocks to be actually recorded. The number of pieces of spectrum data that exist in each unit block is determined in advance by the format, so that it is possible to correspond to the data by the above-described bit allocation information. It should be noted that recording is not performed for a unit block whose bit allocation is 0.

After this spectrum information, the above-mentioned double writing of the scale factor information and the two writing of the bit allocation information are performed.
Write overwriting. With regard to this recording method, only the correspondence of the number corresponds to the double-written information shown in FIG.
Others are the same as the recording of the scale factor information and the bit allocation information described above.

In FIG. 7, the scale factor 2
The overwriting and / or the bit allocation double writing may be stopped, and the portion may be allocated to the spectrum data area.

For the last two bytes, as shown in FIG.
Overwritten. The double writing for two bytes is defined as a format, and the variable writing amount of the double writing cannot be set like the double writing of the scale factor information or the double writing of the bit allocation information.

That is, in the bit allocation calculating circuit 118 in FIG. 1, the data obtained by processing the orthogonal transform output spectrum by the sub-information as the main information and the block as the sub-information by the system shown in FIG. A scale factor indicating the floating state and a word length indicating the word length are obtained, and based on the maximum values, the adaptive bit allocation encoding circuit 106 in FIG.
At 107 and 108, requantization is actually performed, and if necessary, the normalization information is adjusted by the output from the normalization information adjustment circuit 119, and is encoded in a form conforming to the encoding format.

FIG. 9 shows a decoding circuit (decoding device) for decoding again a signal which has been highly efficiently coded by the system shown in FIG. 1 described above. The quantized MDCT coefficient of each band, that is, output terminal 1 in FIG.
Data equivalent to the output signals of 12, 114, and 116 are supplied to a decoding circuit input terminal 908 in FIG. 9 and used block size (information compression parameter) information, that is,
Data equivalent to the output signals of the output terminals 113, 115, and 117 in FIG. 1 are given to the input terminal 910 in FIG. Adaptive bit allocation decoding circuit 90 in FIG.
In step 6, bit allocation is canceled using the adaptive bit allocation information. Next, an inverse orthogonal transform (IMDCT) circuit 9 in FIG.
In 03, 904, and 905, a signal on the frequency axis is converted to a signal on the time axis. The signals on the time axis of these partial bands are output to a band synthesis filter (reverse band splitting filter) (IQMF: reverse right-angle mirror filter) circuit 902 in FIG.
By 901, the signal is decoded into a full band signal.

Here, the normalization information adjustment circuit 911 will be described. Here, a case where the movable contact m of the switch 912 is switched to the fixed contact a will be described.
The operation of the switch 912 when the movable contact m is switched to the fixed contact b will be described later. The normalization information adjustment circuit 911 basically performs the same operation as the normalization information adjustment circuit 119 in FIG. That is, the circuit outputs a numerical value for adjusting the level of the unit block by adding the normalized information.

Reference numeral 909 denotes a normalization information adjustment circuit 91
This is an adder that adds a numerical value from 1 to the scale factor value of the unit block. When the numerical value output from the normalization information adjusting circuit 911 is a negative number, the adder 909 acts as a subtractor. That is, as in the case of the encoding, the level adjustment of every 2 dB is made possible only by adding the output of the same numerical value from the normalization information adjustment circuit 911 to all the unit blocks. A so-called filter effect can be obtained only by adding different numerical values for each unit block. At this time, the addition result is restricted so that it falls within the range of the scale factor value defined in the format. The scale factor value for which the level of the unit block has been adjusted by the adder 909 is used in the decoding process from the adaptive bit allocation decoding circuit 906, and is used only for adjusting the level of the decoded signal. For example, a scale factor value is read from a recording medium on which encoded information is recorded, and the adjusted scale factor value is output to an output terminal 907 to adjust the scale factor value recorded on the recording medium. It is also possible to change it to the value given. The information on the recording medium can be changed as necessary. This is a very simple system,
The level information of the recording medium can be changed.

In the above description, an example in which both the encoding circuit and the decoding circuit have the normalization information adjusting circuit has been described.
Even with the decoding circuit alone, this effect can be fully utilized.

The operation of the normalization information adjustment circuit 911 will be described in more detail with reference to FIGS. FIG. 10 shows a state in which normalization is performed for each unit block. That is, a normalization candidate corresponding to the spectrum data or the MDCT coefficient on the largest frequency axis in the unit block is selected, and the number of the normalization candidate becomes the normalization information of the corresponding unit block. In this example, the normalized information of the block with the block number 0 is 5 and the normalized information of the normalized information with the block number 1 is 7. Similarly, the numbers of the normalization information correspond to the other blocks.
Because the encoded data has this normalization information,
At the time of decoding, decoding is generally performed based on these pieces of normalization information.

On the other hand, FIG. 11 shows that the normalized information determined based on the spectrum data or MDCT coefficient on the maximum frequency axis in the unit block is used as the normalized information adjusting circuit 911.
Indicates the state of operation. This is the encoding
It is possible to perform either of them at the time of decoding. However, here, an example in which, when decoding the encoded data recorded on the recording medium, the operation is performed by the normalization information adjustment circuit 911 in FIG. 9 will be described. Actually, the recording medium shown in FIG.
The normalization information determined in the manner as described above, that is, scale factor information is recorded in the format shown in FIG. By decoding this recording information as it is, it is possible to reproduce it in the form shown in FIG. 10, but here, a value of -1 is calculated for all unit blocks by the normalization information adjustment circuit 911 in FIG. When this value is added to the normalized information by the adder 909 in FIG.
The shape is as shown in That is, the normalized information has a value one less than the original state. This is equivalent to setting the normalization information to a value lower by 2 dB. Since the value of the spectrum data or the MDCT coefficient in each unit block is determined based on the normalization information, the spectrum data or the MDCT coefficient is recorded by setting the normalization information to a value lower by 2 dB. 2 dB from the information recorded on the medium
It will be calculated as a small value. Since the same applies to other unit blocks, a value of -1 is calculated for all the unit blocks by the normalization information adjustment circuit 911 in FIG. 9 and the value is added by the adder 909, as shown in FIG. As a result, decoding at a level lower by 2 dB can be realized. Although an example of the level adjustment has been described here, for example, the normalization information adjustment circuit 911 in FIG. 9 assigns −6 to the block with the block number 3 in FIG.
By outputting and adding -4 to the block of, the normalization information of the blocks of block numbers 3 and 4 can be set to 0 to realize a filter effect at the time of decoding. In this example, for convenience of explanation, the number of unit blocks is five (0 to 4), and the number of normalization candidate numbers is one (0 to 9).
Although the number is 0, for example, in the format used for the mini-disc, the number of unit blocks is 52 from 0 to 51, and the number of normalization candidate numbers is 64 from 0 to 63. By specifying the normalization candidate number, it is possible to perform more precise level adjustment operation or filter processing operation.

Here, as an application example of the reproduction level adjustment operation by the above-described method, an example of realizing a so-called fade-in effect will be described with reference to FIG. A1 to E in FIG.
Reference numeral 1 denotes a state in which five unit blocks are normalized as in the example of FIG. 10 described above.
It is assumed that the time is updated from left to right, that is, from A1 to E1. That is, FIG.
12 are decoded first, and then FIG. 12B1 is decoded. Similarly, FIG.
Also, FIG. 12E shows a state in which decoding is performed from FIG. 12E1. For example, this state can be assumed to be sequentially recorded on a recording medium as encoded information. Here, the coded information at a certain time is referred to as a frame for convenience of description. In this example, five unit blocks exist in one frame, and five frames exist.

On the other hand, FIGS.
The above-described normalization information adjustment circuit 911 outputs an added value for each frame to the coded information indicated by A1 to E1 to perform level adjustment. That is, in FIGS. 12A2 to E2, in the first frame (FIG. 12A2) on the time axis, -8 is added to the normalization information, and -16 is added to the initial encoding state.
The reproduction level of dB is adjusted. Similarly, in the frames of FIGS.
By adding 6, -4, -2, and 0, the level gradually becomes the same as that of the original coded information, so that a so-called fade-in effect can be obtained. Here, for the sake of convenience of explanation, a simple fade-in method has been described, but a value output from the normalization information adjustment circuit is set to a fine value, for example, based on a sine curve, so that it is more audible. A natural fade-in effect can be realized. At this time, it is obvious that the normalization information adjusting circuit has a function of calculating an added value corresponding to a frame, so that a wider variety of effects can be realized, but only a limited number of desired effects are achieved. In this case, the system can be configured by storing a pattern corresponding to the effect in a ROM or the like.

As described above, the normalized information, that is, the scale factor value is read from the recording medium on which the encoded information is recorded, and the adjusted scale factor value is output to 907 in FIG. 9 and recorded on the recording medium. It is also possible to change the scale factor value to an adjusted value. This corresponds to, for example, re-recording the encoded information originally recorded on the recording medium in the form shown in FIGS. 12A1 to E1 into the form shown in FIGS. 12A2 to E2. Thus, even when decoding is performed by a decoding device having no normalization information adjustment circuit, decoding can be performed in a form as shown in FIGS. 12A2 to E2.

Here, the coded information changed as shown in FIGS. 12A2 to E2 is returned to the initial coded information shown in FIGS. 12A1 to E1, that is, so-called Undo (undo).
A method of realizing the operation (cancel the immediately preceding operation and return to the original state) will be described. As shown in FIGS. 12A3 to E3, the normalization information adjustment circuit adds 8 to the normalization information in the frame corresponding to the first on the time axis (FIG. 12A3), and similarly, in FIG. By adding 6, 4, 2, 0 to the frame of E3,
It is possible to restore the same state as in FIG. 12A1 to E1. The value added at this time is a value obtained by inverting the sign of the added value in order to obtain the effects of FIGS. 12A2 to E2.
That is, for example, in FIG. 9, as described above, the normalization information adjustment circuit 911 stores patterns for realizing desired level adjustment, filter processing, and the like in a ROM or the like.
Is realized, at the time of the so-called Undo operation, the movable contact m of the switch 912 is switched to the fixed contact b side, and-
It is only necessary to invert the sign by the one multiplier 913 and to add a value obtained by inverting the sign of the same pattern. That is, the Undo operation as shown in FIG. 12 can be realized only by adding the two of the switch 912 and the -1 multiplier 913 shown in FIG. Of course, some time position information about the edited part is necessary, but in the case of general fade-in or fade-out, for example, the beginning or end of the song often becomes the edit position, Positioning in such a case is easy. Even in the case of editing the entire music, basically the position information only needs to be able to grasp the beginning and end of the music.

By writing the result of re-editing the normalization information, that is, the scale factor value in this manner, to the disk again, it is possible to completely return the disk to the initial state disk before editing.

Here, problems that may occur in the adjustment operation by the above-described normalization information adjustment circuit will be described. By outputting a positive value from the normalization information adjusting circuit 911 in FIG. 9 and adding a positive value to the normalization information by the above-described method, a so-called boost effect of increasing the reproduction level can be obtained. However, at this time, the value that can be added has an upper limit value indicated by the format. This example will be described with reference to FIG. For example, when the number of normalization candidate numbers is 64, that is, 0 to 63, the upper limit value of the normalization information is 63. Therefore, for example, when an attempt is made to increase the reproduction level by adding the same value to the normalization information of all the blocks shown in FIG. 10, the value that can be added is the normalization of the block number 2 which is the maximum normalization information. It is defined by 9 which is the value of the conversion information, and the value is 54 which is a difference value from the upper limit value of 63. If a value larger than this is added, it is necessary to suppress the normalized information exceeding the upper limit indicated by the format to the upper limit indicated by the format, and it is not possible to obtain a desired effect. May cause sound quality change in the reproduced signal after decoding. In addition, for blocks that have exceeded the upper limit value due to the addition operation, it is not possible to perform the so-called undo operation as described with reference to FIG.

In order to avoid or recognize these problems, the maximum value of the normalized information at the partition where the adjustment operation is performed may be detected. This example is shown in FIGS.
5 will be described.

The scale factor information shown in FIG. 7 (represented in each of A in FIGS. 13 to 15) is shown in B in FIGS. 13 to 15 when the frame numbers are n-1, n, and n + 1. The maximum scale factor information max is obtained according to the method of the flowchart. Note that it is assumed that max is determined one for one frame, and for example, it is assumed that the max of the frame number n is defined as an array that becomes max [n]. As other variables shown in the flowchart, i indicates a block number,
sf indicates scale factor information and is represented in the form of an array called sf [i]. end_block indicates the upper limit of the number of records indicated by the format, but it is sufficient if the number of records of the block in the frame indicated by the first byte is sufficient. In this case, the end_block is always equal to or less than the upper limit of the number of records indicated by the format. Is set, so that the processing steps can be reduced.

A description will be given with reference to a flowchart when the frame number (NO.) Shown in FIG. 14B is n. In step ST-1, after setting the block number i to i = 0,
Move to step ST-2. In step ST-2, the maximum schedule factor information max [n] of the frame number n
Is set to max [n] = 0, and the process proceeds to step ST-3. In step ST-3, the block number i is i <end
_ Blck or not, and if NO, end
If YES, the process moves to step ST-4. In step ST-4, it is determined whether or not the maximum scale factor information max [n] of the frame number n is max [n] <sf [i].
If so, the process proceeds to step ST-6, and if YES, the process proceeds to step ST-5. In step ST-5, the maximum scale factor information max [n] of the frame number n is set to sf
After setting [n], the process proceeds to step ST-6. In step ST-6, after increasing the block number i by 1,
It returns to step ST-3.

The process shown in the flow chart is a general process of detecting the maximum value.
The comparison process will be performed as many times as -block,
Normally, the numerical value of end_block is much smaller than the number of points of the digital signal corresponding to the case where the signal is decoded. For example, when the number of points of the digital signal corresponding to the case of decoding is 512 points and the upper limit of the number of recordings indicated by the format is 52, the level of the reproduced signal is simply obtained by a repetition operation of about 1/10. It is possible to predict the maximum value. It is clear that the maximum value prediction can be refined by performing more detailed calculations as much as the processing step allows.

According to this method, as shown in FIGS. 13 to 15, the maximum value of the normalized information is obtained for each frame.
max can be detected. First, when level adjustment is performed by adding the added value output from the normalization information adjustment circuit 911 to the normalization information described with reference to FIG. 9, a maximum value is further detected for these max. This makes it possible to obtain the above-described threshold value for addition in a desired section in which level adjustment is to be performed. By finding this threshold before actually performing the addition process,
It is possible to prevent a change in sound quality due to a level imbalance between blocks having an upper limit of the added value, or to clearly recognize a change in sound quality that occurs when adding an added value exceeding the upper limit. Further, the position where the desired addition value exceeds the upper limit can be obtained from the frame number by verifying max shown in FIGS.

In each B of FIGS. 13 to 15, each frame is used to determine the threshold value of the added value when the level adjustment is performed by performing the adding process on the normalized information of all the blocks. In the above example, the maximum value of the normalized information is calculated.For example, when the addition processing is performed only on the normalized information of a specific block, and when so-called filtering is performed, only the corresponding specific block is deleted. It is clear that the detection has to be performed between frames.

The value of max obtained by this method, or
The detection result of the position where max exists, or the threshold value obtained by subtracting max and the upper limit of the normalized information is FL
(Fluorescent display) By displaying on a tube or liquid crystal display, etc.
It is possible to prevent the occurrence of a problem caused by editing or recording work in advance.

[0074]

According to the first aspect of the present invention, an input digital signal is decomposed into a plurality of frequency band components to obtain signal components in a plurality of two-dimensional blocks relating to time and the number of local waves, and to obtain time and frequency components. For each two-dimensional block, based on the signal components in the two-dimensional block, normalization is performed by selecting, as normalization information, one of several positive values numbered using several bits in advance. Then, a number using the corresponding bits is used as normalization information, a quantization coefficient representing the characteristic of the signal component in the two-dimensional block is obtained for each two-dimensional block relating to time and frequency, and based on the quantization coefficient The bit allocation amount is determined, the signal components in the block are quantized by the normalization information and the bit allocation amount for each two-dimensional block relating to time and frequency, and the information is compressed. Obtain information compression parameter for each click,
The number information of the two-dimensional blocks to be validated is selected from a predetermined number of bits, and the signal components of the two-dimensional blocks related to the time and frequency of the information compression are converted to the information compression parameters for the two-dimensional blocks related to the time and frequency. And a level adjustment for each two-dimensional block by adding a desired value to the normalized information based on a number using several bits for each two-dimensional block at the time of decoding. In the method, for the normalized information by number using several bits of a two-dimensional block,
Since the maximum value is detected in a desired range and the limit value at which the level adjustment is effective can be recognized based on the detected maximum value, the addition process is performed on the normalized information. By doing so, it is possible to clarify the upper limit value that can be added in a desired section when adjusting the reproduction level and realizing the filter effect, thereby reducing sound quality deterioration and data correction caused by inappropriate addition processing. Thus, it is possible to obtain a decoding method capable of preventing in advance the phenomenon in which decoding becomes impossible.

According to the second aspect of the present invention, band dividing means for dividing an input digital signal into a plurality of frequency band components,
Orthogonal transform means for orthogonally transforming a signal to obtain signal components for encoding and / or analysis in a plurality of two-dimensional blocks related to time and frequency, and a signal in the two-dimensional block for each two-dimensional block related to time and frequency Based on the components, normalization is performed by selecting one of several positive values that have been numbered using several bits in advance as normalization information, and the number using the corresponding several bits A normalized data calculating unit that obtains as a normalized information, a quantized coefficient calculating unit that obtains, for each two-dimensional block relating to time and frequency, a quantized coefficient representing a characteristic of a signal component in the two-dimensional block,
Bit allocation calculating means for determining a bit allocation amount based on the quantized coefficient, and a compression code for quantizing signal components in the block by using normalized data and the bit allocation amount for each two-dimensional block relating to time and frequency and compressing information. Information compression parameter determining means for obtaining an information compression parameter for each two-dimensional block relating to time and frequency, and the number of valid two-dimensional blocks for selecting the number information of valid two-dimensional blocks from a predetermined value of several bits Information determining means, decoding means for decoding signal components in a two-dimensional block related to time and frequency using information compression parameters for each two-dimensional block related to time and frequency, Including normalization information adjusting means for enabling level adjustment for each two-dimensional block by adding the values of Detecting means for detecting the maximum value within a desired range for the normalized information by the number using several bits of the two-dimensional block, and determining whether the level adjustment is effective based on the detected maximum value. Display means for displaying a parameter capable of recognizing a certain limit value,
By performing the addition processing on the normalized information, it is possible to clarify the upper limit value that can be added in a desired section when adjusting the reproduction level or realizing the filter effect, thereby making it possible to perform improper addition. It is possible to obtain a decoding device capable of preventing in advance the sound quality deterioration caused by the processing and the phenomenon that the data cannot be corrected.

According to the third aspect of the present invention, an input digital signal is decomposed into a plurality of frequency band components to obtain signal components in a plurality of two-dimensional blocks relating to time and frequency, and two-dimensional blocks relating to time and frequency are obtained. For each, based on the signal components in the two-dimensional block, normalization is performed by selecting, as normalization information, one of several positive values that have been numbered using several bits in advance. A number using the corresponding several bits is used as normalization information, and a quantization coefficient representing the characteristic of the signal component in the two-dimensional block is obtained for each two-dimensional block relating to time and frequency, and the bit allocation amount is determined based on the quantization coefficient. Is determined, the signal components in the block are quantized using the normalization information and the bit allocation amount for each two-dimensional block related to time and frequency, and the information is compressed. To obtain the information compression parameters of each,
The number information of the two-dimensional blocks to be validated is selected from a predetermined value of several bits, and the signal components of the two-dimensional blocks related to the time and frequency of the information compression are converted to the information compression parameters of the two-dimensional blocks for the time and frequency. And a level adjustment for each two-dimensional block by adding a desired value to normalized information based on a number using several bits of the two-dimensional block at the time of decoding. In the method, the position of the maximum value is detected in a desired range with respect to the normalized information by the number using several bits of the two-dimensional block, and the position where the level adjustment becomes effective based on the detected position of the maximum value is performed. The maximum level of a digital signal obtained by decoding a signal with high efficiency encoding is detected by a smaller processing unit. In, it is possible to obtain a decoding method which can realize a high speed.

According to the fourth aspect of the present invention, band dividing means for dividing an input digital signal into a plurality of frequency band components,
Orthogonal transform means for orthogonally transforming a signal to obtain signal components for encoding and / or analysis in a plurality of two-dimensional blocks related to time and frequency, and a signal in the two-dimensional block for each two-dimensional block related to time and frequency Based on the components, normalization is performed by selecting one of several positive values that have been numbered using several bits in advance as normalization information, and the number using the corresponding several bits A normalized data calculating unit that obtains as a normalized information, a quantized coefficient calculating unit that obtains, for each two-dimensional block relating to time and frequency, a quantized coefficient representing a characteristic of a signal component in the two-dimensional block,
Bit allocation calculating means for determining a bit allocation amount based on the quantized coefficient, and a compression code for quantizing signal components in the block by using normalized data and the bit allocation amount for each two-dimensional block relating to time and frequency and compressing information. Information compression parameter determining means for obtaining an information compression parameter for each two-dimensional block relating to time and frequency, and the number of valid two-dimensional blocks for selecting the number information of valid two-dimensional blocks from a predetermined value of several bits Information determining means, decoding means for decoding signal components in a two-dimensional block related to time and frequency using information compression parameters for each two-dimensional block related to time and frequency, Decoding means having a normalization information adjusting means for enabling level adjustment for each two-dimensional block by adding Detecting means for detecting the position of the maximum value in a desired range with respect to the normalized information using a number using several bits of a two-dimensional block, and level adjustment based on the detected position of the maximum value is effective. Display means for displaying a parameter capable of recognizing the position,
A decoding device capable of detecting the maximum level of a digital signal obtained by decoding a signal that has been subjected to high-efficiency encoding, with a smaller processing step and at a high speed can be obtained.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a circuit of a high-efficiency compression encoding encoder that can be used for bitrate compression encoding according to a specific example of an embodiment of the present invention.

FIG. 2 is a diagram illustrating a structure of an orthogonal transform block at the time of bit compression.

FIG. 3 is a block diagram showing a specific configuration of a bit allocation calculating circuit (bit allocation calculating means) of the high efficiency compression encoding encoder of FIG. 1;

FIG. 4 is a diagram illustrating spectra of each critical band and a band divided in consideration of block floating.

FIG. 5 is a diagram showing a masking spectrum.

FIG. 6 is a diagram in which a minimum audible curve and a masking spectrum are combined.

FIG. 7 is a diagram showing how data is encoded.

FIG. 8 is a diagram showing details of data of a first byte in FIG. 7;

FIG. 9 is a block diagram showing a circuit of a high-efficiency compression encoding decoder that can be used for bitrate compression encoding in a specific example of the embodiment of the present invention.

FIG. 10 is a diagram showing how normalization information is determined.

FIG. 11 is a diagram illustrating a state in which normalization information is adjusted.

FIG. 12 is an explanatory diagram of a fade-in.

FIG. 13 is a diagram showing a processing step (n-1) of maximum value detection.

FIG. 14 is a diagram showing a processing step (n) for detecting a maximum value.

FIG. 15 is a diagram showing a processing step (n + 1) of maximum value detection.

[Explanation of symbols]

101, 102 band division filter, 103, 104,
105 orthogonal transform circuit (MDCT), 109, 110,
111 block determination circuit, 118 bit allocation calculation circuit, 106, 107, 108 adaptive bit allocation coding circuit, 119 normalization information adjustment circuit, 120, 121,
122 adder, 302 energy calculator per band, 3
03 convolution filter, 304 adder, 305 function generator, 306 divider, 307 synthesizer, 308
Subtractor, 309 delay circuit, 310 allowable noise corrector,
701, 702 Band synthesis filter (IQMF), 70
3, 704, 705 inverse orthogonal transform circuit (IMDCT),
706 Adaptive bit allocation decoding circuit, 901, 902
Band synthesis filter, 903 to 905 inverse orthogonal transform circuit,
906 adaptive bit allocation decoding circuit, 909 adder,
911 normalization information adjustment circuit, 912 switch, 9
13 -1 multiplier.

Claims (4)

[Claims] An input digital signal is decomposed into a plurality of frequency band components to obtain signal components in a plurality of two-dimensional blocks relating to time and the number of local waves.
For each dimension block, based on the signal components in the two-dimensional block,
Normalization is performed by selecting one of several positive values that have been numbered using several bits in advance as normalization information, and the number using the corresponding several bits is used as normalization information. A quantization coefficient representing a characteristic of a signal component in the two-dimensional block is obtained for each of the two-dimensional blocks relating to time and frequency, and a bit allocation amount is determined based on the quantization coefficient; For each time, the information component in the block is quantized based on the normalization information and the bit allocation amount to compress the information, and the information compression parameter for each two-dimensional block relating to the time and frequency is obtained, and the number information of the effective two-dimensional block is obtained. Is selected from a predetermined value of several bits, and the signal component of each two-dimensional block related to the time and frequency of the information compression is converted to the value related to the time and frequency. The decoding is performed by using the information compression parameter of each two-dimensional block, and a desired value is added to the normalized information by the number using several bits of each two-dimensional block at the time of decoding. In the decoding method in which the level is adjusted every time, a value of a maximum value in a desired range is detected for normalized information by a number using several bits of the two-dimensional block, and a value of the detected maximum value is detected. A decoding method characterized in that a limit value at which a level adjustment is effective can be recognized based on the above. 2. Band splitting means for splitting an input digital signal into a plurality of frequency band components, encoding in a plurality of two-dimensional blocks relating to time and frequency by orthogonally transforming the signal,
And or orthogonal transformation means for obtaining a signal component for analysis,
Based on the signal components in the two-dimensional block for each of the two-dimensional blocks relating to time and frequency, one of several positive values that have been numbered using several bits in advance is selected as normalization information. A normalized data calculating means for performing normalization to obtain a number using the corresponding several bits as the normalized information, and representing a characteristic of a signal component in the two-dimensional block for each two-dimensional block relating to the time and frequency. Quantization coefficient calculation means for obtaining a quantization coefficient, bit allocation calculation means for determining a bit allocation amount based on the quantization coefficient, and the above-described normalized data and bit allocation amount for each of the two-dimensional blocks relating to time and frequency. Compression encoding means for quantizing a signal component in the block to compress the information, and an information pressure for obtaining an information compression parameter for each two-dimensional block relating to the time and frequency. Parameter determining means, valid two-dimensional block number information determining means for selecting the number information of valid two-dimensional blocks from a predetermined number of bits, and a signal in the two-dimensional block relating to the time and frequency of the compressed information. Decoding means for decoding the component using the information compression parameter for each of the two-dimensional blocks relating to the time and frequency; and a normalization means for adding a desired value to the normalized information to enable level adjustment for each two-dimensional block. A decoding device having a normalized information adjusting means, a detecting means for detecting a maximum value in a desired range with respect to the normalized information by a number using several bits of the two-dimensional block, Display means for displaying a parameter capable of recognizing a limit value at which level adjustment is effective based on the value. . 3. An input digital signal is decomposed into a plurality of frequency band components to obtain signal components in a plurality of two-dimensional blocks relating to time and frequency.
For each dimension block, based on the signal components in the two-dimensional block,
Normalization is performed by selecting one of several positive values that have been numbered using several bits in advance as normalization information, and the number using the corresponding several bits is used as normalization information. A quantization coefficient representing a characteristic of a signal component in the two-dimensional block is obtained for each of the two-dimensional blocks relating to time and frequency, and a bit allocation amount is determined based on the quantization coefficient; In each case, the signal components in the block are quantized using the normalization information and the bit allocation amount to compress the information, and the information compression parameters for the two-dimensional blocks relating to the time and frequency are obtained.
The number information of the two-dimensional blocks to be validated is selected from a predetermined value of several bits, and the signal components of the two-dimensional blocks related to the time and frequency of the compressed information are converted into the information of the two-dimensional blocks related to the time and frequency. The decoding is performed using the compression parameter, and at the time of the decoding, the level is adjusted for each two-dimensional block by adding a desired value to the normalization information based on the number using several bits of the two-dimensional block. In the decoding method described above, the position of the maximum value is detected in a desired range with respect to the normalized information using the number of bits of the two-dimensional block, and the level adjustment is performed based on the detected position of the maximum value. A decoding method characterized by recognizing a valid position. 4. A band dividing means for dividing an input digital signal into a plurality of frequency band components, encoding in a plurality of two-dimensional blocks relating to time and frequency by orthogonally transforming the signal,
And or orthogonal transformation means for obtaining a signal component for analysis,
Based on the signal components in the two-dimensional block for each of the two-dimensional blocks relating to time and frequency, one of several positive values that have been numbered using several bits in advance is selected as normalization information. A normalized data calculating means for performing normalization to obtain a number using the corresponding several bits as the normalized information, and representing a characteristic of a signal component in the two-dimensional block for each two-dimensional block relating to the time and frequency. Quantization coefficient calculation means for obtaining a quantization coefficient, bit allocation calculation means for determining a bit allocation amount based on the quantization coefficient, and the above-described normalized data and bit allocation amount for each of the two-dimensional blocks relating to time and frequency. Compression encoding means for quantizing a signal component in the block to compress the information, and an information pressure for obtaining an information compression parameter for each two-dimensional block relating to the time and frequency. Parameter determining means, valid two-dimensional block number information determining means for selecting the number information of valid two-dimensional blocks from a predetermined number of bits, and a signal in the two-dimensional block relating to the time and frequency of the compressed information. Decoding means for decoding the component using the information compression parameter for each of the two-dimensional blocks relating to the time and frequency; and a normalization means for adding a desired value to the normalized information to enable level adjustment for each two-dimensional block. A decoding device having digitized information adjusting means, a detecting means for detecting a position of a maximum value within a desired range with respect to the normalized information by a number using several bits of the two-dimensional block, and a position of the detected maximum value Display means for displaying a parameter capable of recognizing a position at which a level adjustment is effective based on the parameter. .