High-efficiency coding device

High-efficiency coding device

The present invention relates to a high-efficiency encoding device that encodes input digital data by so-called high-efficiency encoding.

Although there have been various methods for encoding audio or sound signals in the past, for example, there are: the audio signals on the time axis are not divided into blocks, but are divided into complex frequency bands and then encoded The non-blocking frequency band division method is the so-called "Sub-Band-Coding (SBC)", or the signal on the time axis is transformed (orthogonal transformation) into the signal on the frequency axis and then divided into plural The frequency band is then coded for each band. The so-called "transform coding" and other various methods are used to block the frequency band division method. In addition, a method of combining the aforementioned band division coding and transform coding into high-efficiency coding can also be considered. In this case, for example, the aforementioned band division coding is used to perform band division, and then the signals of each band are orthogonally transformed into The signal on the frequency axis is then coded for each band after the orthogonal transformation.

Here, the filter used to perform the aforementioned band division is, for example, a QMF filter. The QMF filter is disclosed in: 1976 RECrochiere Digital Coding of speech in S-ubband Bell Syst.Tech.J. Vol.55, No.8 in the 1976 publication. Another example is: ICASSP 83, BOSTON Polyphase Q-uadrature filters-A new Subband Coding technique Joseph H. Rothweiler's publication reveals a filter division method with equal band width.

In addition, as for the aforementioned orthogonal transform method, there are, for example: block the input digital signal in a predetermined unit time (frame), and then apply fast Fourier transform (FFT) or cosine to each block. Orthogonal transformation (MDCT) that transforms the time axis into the frequency axis through transformation (DCT), or modified DCT transformation (MDCT), etc., and transforms the time axis into the frequency axis orthogonal transformation method. Also, the MDCT system is described in: ICASSP 1987. Subband/Transform Coding Using Filter Bank Designs Based on Time Domain Al-iasing Cancellation JPPrincen ABBradley Univ. Of Surrey Royal Melbourne Inst. Of Tech.

Furthermore, as for the frequency division width for quantizing each frequency component that has been divided by the frequency band, for example, the frequency division width of the band division that takes into account the characteristics of human hearing is used. That is, the band width of the type that is generally called critical band (critical band) is divided into a plurality of bands (for example, 25 bands) by using a type of band width in which the higher the band width increases. At this time, when the data of each band is encoded, it is encoded with a predetermined bit allocation for each band; or is encoded with a compliant bit allocation (bit allocation) for each band. For example: when the coefficient data obtained by the above MDCT processing is coded according to the aforementioned compliance bit allocation, the MDCT coefficient data of each band obtained by the aforementioned MDCT processing of each block is used. Encoding is implemented according to the number of compliance bits.

As for the aforementioned bit allocation method, there are the following two known ones. For example: According to IEEE Transactions of Accoustics, Speech, and Signal Processing, Vol.·ASSP-25, No. 4, August 1977, bit allocation is performed according to the signal size of each band. However, although this method has flattened the quantized noise spectrum and minimized the noise energy, it does not make use of the so-called "masking effect" in the auditory sense, so the actual sense of noise is not the best. Appropriately, this is its shortcoming.

In addition, for example, according to the record of ICASSP 1980 The critical band coder digital encoding of the perceptual requirem-ents of the auditory system MAKransner MIT, the necessary signal-to-noise ratio for each band is obtained by using auditory masking. Fixed bit allocation method. However, the disadvantage of this method is that even when the characteristics are measured with a sine wave input, the characteristic value cannot be a very good value because of the fixed bit allocation.

<p>In order to solve the aforementioned problems of the aforementioned two bit allocation methods, someone once proposed a highly efficient encoding device, such as a system that first divides all the bits that can be used for bit division into: The part of the fixed bit allocation pattern set in advance for a small block and the part used for bit allocation that depends on the signal size of each block are used; and the division ratio depends on the input signal For related signals, the smoother the frequency spectrum of the aforementioned signal, the greater the division ratio of the portion of the aforementioned fixed bit allocation pattern.</p><p>According to the bit allocation method in this high-efficiency encoding device, for example, when the energy of the input of a sine wave is concentrated in a specific frequency spectrum, it is remarkable by allocating more bits in the block containing the frequency spectrum. Improve the overall signal noise characteristics (S/N ratio). Therefore, in general, human hearing is extremely sensitive to signals with steep spectral components. By using this method to improve the signal noise characteristics, not only the measurement value is improved, but also the sound quality of the hearing. The improvement is also helpful.</p><p>However, if the bit allocation that depends on the input signal is purely based on the consideration of improving the signal-to-noise ratio, for example, a triangle sound like a signal that contains many steep spectral components, think When a lower bit rate is used for compression, enough bits are not allocated to the blocks corresponding to each spectrum, and therefore, sufficient sound quality cannot be obtained auditorily.</p><p>Therefore, the present invention was developed in view of the foregoing facts, and its purpose is to provide a high-efficiency encoding device, even if it is to compress a signal containing a large number of steep spectral components at a lower bit rate. Under the circumstances, it is also possible to obtain a very good sound quality in terms of hearing.</p>

<p>The high-efficiency coding device of the present invention is designed to achieve the aforementioned purpose, and belongs to a high-efficiency coding device that performs high-efficiency coding by assigning input digital data to compliant bit positions. Its characteristics are: The device has a compliant bit allocation method, which divides all bits available for bit allocation into: for each small block that has been subdivided in time and frequency, a predetermined fixed bit allocation is performed. This part, as well as the part that executes the bit allocation that depends on the signal size in each small block, is used; and when the aforementioned bit allocation that depends on the signal size in each small block is performed, it corresponds to each small block The corresponding frequency band will be processed with a heavier weight.</p><p>That is, the high-efficiency encoding device of the present invention is made by dividing all the bits available for bit allocation into: no matter how short the time is, a predetermined fixed bit allocation pattern is performed. , And the implementation of the bit allocation that depends on the signal size of each block is used, and in addition to the fixed bit allocation pattern, the bit allocation that depends on the signal size of each block is based on To solve the aforementioned problems by adding weight to the corresponding band of the block.</p><p>Here, as for the above-mentioned method of "emphasizing weight", it is based on the so-called: "High-range noise is more difficult to detect by human ears than low-range noise because of the sensitivity of the human ear. The fact that signals in the lower range are easier to be masked". Therefore, more bits are allocated in the block on the lower range side, which can effectively solve the aforementioned problems.</p><p>Therefore, the aforementioned compliant bit allocation coding method can be made as follows: the bit allocation corresponding to the frequency band corresponding to the aforementioned small block is processed to increase the weight according to the same signal energy. It lies in the principle of assigning more bits to the lower range to increase the weight.</p><p>In addition, although the division ratio between the fixedly allocated bits and the bits allocated depending on the input signal, whether it is to make the division ratio depend on the signal related to the input signal, or to make the division The ratio does not depend on the signal related to the input signal and can be applied to the present invention. However, it is made so that the aforementioned division ratio depends on the input signal to obtain a better sound quality.</p><p>Therefore, the aforementioned compliant bit allocation coding method can also be made: divide all the bits available for bit allocation into: predetermined fixed bits for each small block that has been subdivided in time and frequency The pattern part of the allocation and the part that performs bit allocation that depends on the signal size in each small block is used. The division ratio of the two parts is treated as the division ratio of the input signal for bit allocation. By.</p><p>In addition, a plurality of the aforementioned pa-tterns can be prepared in advance, and these pa-tterns can be switched according to the input signal. By slowly changing the variable of the weighting weight according to the input signal, It can be changed into a bit allocation that is more in line with the sense of hearing, which can further improve the sound quality.</p><p>Therefore, the aforementioned compliant bit allocation coding method can be made: the bit allocation that depends on the signal size in the small block, which is performed according to the weight of the frequency band corresponding to the aforementioned small block, is based on the reason. The input signal is different and the coefficient of the aggravating weight is changed to execute it.</p>

<p>According to the high-efficiency encoding device of the present invention, it is made by dividing all the bits available for bit allocation into: a predetermined fixed bit allocation pattern part no matter how short the time is , And execute the bit allocation that depends on the signal size of each block to use. In addition to the fixed bit allocation pattern, the bit allocation that depends on the signal size of each block also corresponds to Because the band corresponding to this block is weighted, even for a signal with multiple sharp and steep spectral components, an effective bit allocation suitable for auditory perception can be easily achieved, and a lower bit is used. Yuan rate can encode high-quality music signals.</p>

Hereinafter, the embodiments of the present invention will be described with accompanying drawings.

The high-efficiency encoding device of this embodiment is shown in Figure 1. It is a high-efficiency encoding device capable of compliantly assigning input digital data (digital audio data) supplied to the input terminal 10 to perform high-efficiency encoding. The encoding device has: divide all the bits available for bit allocation into: for each small block that has been subdivided in time and frequency, a predetermined fixed bit allocation pattern is performed, and execution The part of the bit allocation that depends on the signal size in each small block is used, and when the aforementioned bit allocation that depends on the signal size in each small block is performed, it depends on the frequency band corresponding to each small block. It is constituted by the compliance bit coding means 16, 17, and 18 which carry out heavy weight processing.

Here, the aforementioned "heavy weight" is based on the so-called "due to the sensitivity of the human ear, the noise on the high-range side is more difficult to perceive by the ear than the noise on the low-range side, and the high-range side is easily affected by the low-range side. The fact that "the signal is masked by the "can be quite effective if more bits are allocated in the block on the low range side.

Therefore, the aforementioned compliant bit allocation coding means 16, 17, 18 can be made: the processing of the weighted weight of the bit allocation implemented for the frequency band corresponding to the aforementioned small block can be made if it has the same If the signal energy is higher, the more bit distributors will be emphasized on the low range side.

In addition, although the division ratio between the fixedly allocated bits and the bits allocated depending on the input signal, whether it is to make the division ratio depend on the signal related to the input signal, or to make the division This does not depend on the signal related to the input signal and can be applied to the present invention. However, if the aforementioned division ratio depends on the input signal, a better sound quality can be obtained.

Therefore, the aforementioned compliant bit allocation coding methods 16, 17, 18 can also be made: all the bits available for bit allocation are divided into: for each small block that has been subdivided in time and frequency The pattern part of the predetermined fixed bit allocation and the part of the bit allocation that depends on the signal size in each small block are used. The division ratio of the two parts is regarded as the division ratio dependent on the input signal. And those who perform bit allocation.

In addition, it is also possible to prepare a plurality of the aforementioned patterns of the weighted weight in advance, and switch these patterns according to the input signal, which can be changed by slowly changing the variable of the weighted weight according to the input signal. The bit allocation more in line with the sense of hearing can further improve the sound quality.

Therefore, the aforementioned compliant bit allocation coding means 16, 17, 18 can be made: the bits that depend on the signal size in the small block will be executed according to the weight of the frequency band corresponding to the small block. The method of distribution is performed based on the coefficient of the weighted weight that changes due to the difference in the input signal.

The following is accompanied by Figure 1 to illustrate the embodiment of the present invention that uses band division coding (SBC), compliant transform coding (ATC), and compliant bit allocation (APC-AB) for input digital signals such as audio PCM signals. The various technologies to be composed of high-efficiency coding.

The high-efficiency encoding device of this embodiment shown in Figure 1 is obtained by dividing the input digital signal (input audio PCM signal) into a plurality of frequency bands using filters and the like, and performing orthogonal conversion for each frequency band. The frequency spectrum data on the frequency axis is assigned a compliant status element to perform coding for each so-called critical band that takes into account the characteristics of human hearing. Of course, the non-blocking frequency division width performed by filters etc. can also be made into equal division width.

Furthermore, in the embodiment of the present invention, before the orthogonal transformation, the block size (block length) is compliantly changed in response to the input signal. At the same time, the block size (block length) is changed in a critical band width unit or in the high range. The critical band width is further subdivided into blocks for floating processing. In addition, the aforementioned "critical band" means: a frequency band divided after taking into account the characteristics of human hearing. It means: using a narrow band noise of the same intensity next to the frequency of a certain pure tone to mask the pure tone (mask) in terms of the band of the noise. This critical band is the higher the frequency band, the wider the band width, and the 0-20KHz full frequency band is divided into, for example, 25 critical bands.

That is, in Figure 1, an audio PCM signal of 0-20KHz is input into the input terminal 10, for example. This input signal is divided into a band division filter 11 composed of a so-called QMF filter, etc.: 0~10KHz band and 10KHz~20KHz band, and 0~10KHz band signal is filtered by the same QMF The band division filter 12 composed of a device and the like is further divided into a band of 0~5KHz and a band of 5KHz~10KHz.

The signal in the 10KHz~20KHz band sent by the aforementioned band division filter 11 is then sent to the MDCT (Modified Discrete Cosine Transform) circuit 13 and the block size determination circuit 19 as an example of the orthogonal conversion circuit; The signal in the 5KHz~10KHz band sent by the split filter 12 is sent to the MDCT circuit 14 and the block size determination circuit 20; and the signal in the 0~5KHz band sent by the band split filter 12 is sent To the MDCT circuit 15 and the block size determination circuit 21. The block size determination circuits 19, 20, and 21 described above are used to determine the block size during MDCT processing. The block size determines the block size of the circuits 19, 20, and 21 to perform MDCT processing, respectively.

Here, a specific example of the block size of each MDCT circuit 13, 14, 15 is shown in Fig. 2. In the specific example shown in Fig. 2, the higher the frequency band, the wider the frequency band, and the higher the time resolution (that is, the shorter the block length). That is, for the signal in the 0~5KHz band on the low-range side and the signal in the 5KHz~10KHz band on the mid-range side with b_<i>L</i>, B_<i>M</i>During the period of MDCT processing performed once for the block of, the signal of the 10KHz~20KHz band on the high-range side is b_<i>H1</i>, B_<i>H2</i>The blocks of are implemented twice MDCT processing.

In this embodiment, in this way, the number of samples of the orthogonal transform block of each band is made the same, in order to simplify the device on the one hand, and on the other hand to improve the frequency resolution ability on the low-frequency side with a narrower critical band width. The high-range side that contains more advantageous components for transitional signals improves the time resolution ability. In addition, in this embodiment, since it is also assumed that there is a large temporal change in the signal, each band is allowed to be divided into 1/2 and 1/4 compliance blocks.

Next, returning to Figure 1, the spectrum data or MDCT coefficient data on the frequency axis obtained by MDCT processing of each MDCT circuit 13, 14, 15 is a so-called c-ritical band each time Or if it is in the high range, the critical band is divided into a band and sent to the aforementioned compliant bit allocation coding circuits 16, 17, and 18.

In the aforementioned compliant bit allocation coding circuits 16, 17, 18, the number of bits allocated to the aforementioned critical band or the bit rate of each band in which the critical band is subdivided in the high range is allocated to the bit rate. , Requantize each spectrum data (or MDCT coefficient data). As for the specific configuration of the compliant bit allocation coding circuits 16, 17, 18, it will be described later.

The data encoded in this way is taken out via the output terminals 22, 24, and 26. At this time, the aforementioned encoded data and the floating information used to indicate the size of the signal that has been normalized have been executed, and the bit length used to indicate which bit length is used for quantization. The element length information is also sent to the circuit configuration of the subsequent stage.

Here, Fig. 3 is used to explain the specific configuration of the aforementioned compliant bit allocation coding circuit.

The output of the MDCT circuits 13, 14, 15 shown in Figure 1 is sent to the input terminal 400 of the compliant bit allocation coding circuit shown in Figure 3 to calculate the energy of each band. Circuit 303.

In this energy calculation circuit 303, which is used to calculate the energy of each band, the energy of each critical band is calculated by calculating the square root of the square root of the average value of each amplitude value in each band. , Or find the energy of each band after the critical band is subdivided in the high range. In addition, the peak value or average value of the amplitude value can also be used to replace the energy of each band.

The output sent from the aforementioned energy calculation circuit 303, for example, the spectrum of the sum of each critical band, or the spectrum of the sum of each band after the critical band is subdivided in the high range, becomes, for example, The frequency spectrum (Bark Spectrum) SB shown in Figure 4. However, in Figure 4, in order to simplify the illustration, the 12-band (B_<i>1</i>~B_<i>12</i>) To express the number of bands in the aforementioned critical band or in the high range, the critical band is subdivided.

Here, it is assumed that in this embodiment, if the number of bits that can be used for transmission or recording for representing MDCT coefficients is 100 Kbps, a fixed bit allocation pattern of 100 Kbps is used. In addition, the 100Kbps of the usable bit number is set in the usable total bit number output circuit 302, for example. The total number of bits that can be used can also be input from outside.

Moreover, in this embodiment, a plurality of bit allocation patterns for the aforementioned fixed bit allocation are prepared. Various choices can be made according to the nature of the signal. In the configuration of FIG. 3, the fixed bit allocation circuit 305 has various patterns that can distribute the bit amount of the short-term block corresponding to the aforementioned 100 Kbps to each frequency.

That is, in the aforementioned fixed bit allocation circuit 305, in particular, a plurality of different patterns having different bit allocation rates in the mid-range and high-range ranges are prepared. And made: choose the so-called "the smaller the size of the signal, the smaller the amount of bit allocation in the higher the range". By doing this, it can be made to exert the so-called "loundness effect in which the sensitivity of the high range decreases when the signal is smaller." In addition, as for the signal size at this time, although the signal size of the entire band can be used, for example, the output of a non-blocking frequency division circuit such as a filter or the output of MDCT can also be used.

In addition, in this embodiment, regarding the bit allocation that depends on energy, the dB value of the energy for each small block is multiplied by the coefficient preset for each small block to increase the weight of the row, and the aforementioned bit The distribution is proportional to the value of the weighted portion. Here, by setting the aforementioned weighting coefficient to a larger value in the low range, more bits can be allocated to the low range. In addition, this energy-dependent bit allocation is performed by the energy-dependent bit allocation circuit 304 supplied with the output of the energy calculation circuit 303.

That is, in this energy-dependent bit allocation circuit 304, similar to the fixed bit allocation described above, a plurality of patterns of the aforementioned weighting coefficients are prepared, and the input signal is used to switch these plural patterns. , Or use, for example, the input signal to interpolate the shape of the two weighted weights to the shape of the weighted weight to calculate the energy-dependent bit allocation. Therefore, in this embodiment, by using the input signal to change the coefficient of the weighting weight, it can become a bit allocation that is more suitable for the sense of hearing (hearing sense), so as to improve the sound quality.

In Figure 3, the division ratio between the allocation value of the aforementioned fixed bit allocation pattern and the bit allocation value dependent on the Bark Spectrum SB is represented by the smoothing of the signal spectrum It is determined by the index of the degree. That is, in this embodiment, the output of the aforementioned energy calculation circuit 303 is sent to the spectrum smoothness calculation circuit 308, and the spectrum smoothness calculation circuit 308 calculates the difference between adjacent values of the signal spectrum The value obtained by dividing the sum of absolute values by the sum of the signal spectrum is calculated as an index, and this index is sent to the bit division rate determination circuit 309 for calculating the division rate of the aforementioned bit allocation.

The division rate data sent from the bit division rate determination circuit 309 are respectively input to: the multiplier 402 supplied with the output of the fixed bit distribution circuit 305 and the multiplication of the output supplied with the energy-dependent bit distribution circuit 304Device401. The outputs of these multipliers 401 and 402 are sent to the sum calculation circuit 306. That is, the fixed bit allocation and the value of the bit allocation depending on the critical band of each band, or depending on the high range is the bit allocation of the spectrum of each band after the critical band is divided The sum of the values is calculated by the aforementioned sum calculation circuit 306, and the result of this calculation is sent from the output terminal (the bit allocation amount output terminal of each band) 307 to the subsequent circuit configuration for use in quantization.

In addition, other methods for obtaining the aforementioned division ratio can also be used. For example, first calculate the exponent R that multiplies the dB value of the aforementioned weighted energy by the value between 0 and 1 obtained from the aforementioned index. The sum S1 of the obtained values; and calculate the sum S2 of the value obtained by multiplying the value of (1-R) by the fixed bit allocation pattern for each small block, and take this one S1 and S2 The ratio of the total value of S1 and S2 is used as the method of dividing the ratio between the energy-dependent bit allocation and the fixed-pattern bit allocation.

Here, the state of bit allocation when a signal having the frequency spectrum shown in FIG. 5 is input for the configuration of this embodiment is shown in FIG. 6. However, in Figures 5 and 6, the number of bands for which the frequency spectrum is quantized is set to 12. In Figure 6, the slashed rectangles indicate the number of bits allocated according to the size of the signal components of each block, and the rectangles outside the slashes in the figure indicate that they are allocated according to the fixed pattern. The number of bits in each block is allocated the number of bits equivalent to the sum value in the block. In addition, although the numerical values in the aforementioned Figures 5 and 6 are not expressed as integer values but as real values, this is only a representation of the calculation process on the way. In the end, by These values are rounded up to find the number of allocated bits for each block.

The input signal shown in Figure 5 contains signal components with isolated frequency spectrums of roughly the same size as those of A, B, and C in the figure. At this time, in the conventional way, in order to allocate the bits allocated according to the size of the signal component of each block to make the signal noise characteristic close to the best, for example: in the second block in the figure In addition to the bits allocated according to the fixed pattern, the 6th and 10th blocks are also added with roughly the same number of bits. In contrast, in the bit allocation method of the embodiment of the present invention, the more bits are added in the order of the second, sixth, and tenth blocks in the figure. Therefore, according to the bit allocation method according to the embodiment of the present invention, compared with the conventional method, although the signal noise characteristics on the high-range side will be slightly deteriorated, the signal noise characteristics on the low-range side will be better. Effectively improve.

In addition, the noise on the high-range side generated when the bit allocation method of the embodiment of the present invention is used is compared with the noise on the low-range side in terms of the dependence of the sensitivity of the human ear on the frequency. It is more unpleasant, and it is also easy to be masked by the signal on the low range side, because the sense of hearing does not pose a big problem.

Therefore, if the bit allocation method of this embodiment is used, since the noise on the low-range side that is relatively easy to be felt by the ear can be suppressed, the sound quality in the sense of hearing can be effectively improved.

In addition, it should be noted here that in the bit allocation method according to this embodiment, for example, the number of bits in block No. 6 in Fig. 5 is less than that of block No. 2. , But it is allocated more bits than the block in Figure 4. This kind of bit allocation method is not simply obtained by applying the so-called "fixed bit allocation pattern" that allocates more bits to the low range. The bit allocation of the signal size in each small block can be achieved by adding weight according to the corresponding band of the small block.

In addition, in this embodiment, if the spectrum obtained as a result of MDCT is in the low frequency range below 100 Hz, it is assumed that only a few can be obtained. In this case, there are many signals equivalent to frequencies higher than those in the lower range mixed in each frequency spectrum on the low range side obtained by calculation, and it is necessary to allocate a sufficient number of bits on the low range side. . Therefore, the approximation can be made as long as the bits allocated according to the size of the signal component of each block are allocated so that the more they are on the lower range, the more they are allocated.

Furthermore, the interval used to obtain the frequency spectrum is longer. For example, if it is a high-efficiency encoding device that can obtain a sufficient density of the frequency spectrum below 100 Hz, it will depend on the low sensitivity to human ears. For example, the allocation of bits for the signal size of the spectrum equivalent to less than 50Hz can also be made: less than the number of bits allocated for bands above 50Hz, that is, the allocation of bands below 50Hz With fewer bits, more bits are allocated for bands above 50 Hz.

Fig. 7 shows the structure of a decoding device for re-decoding a signal that has been highly-encoded by the aforementioned high-efficiency encoding device of this embodiment.

That is, in this figure 7, the quantized MDCT coefficients of each band are given to the decoding circuit input terminals 122, 124, 126, and information on the block size used is given to the input terminals 123, 125, 127. The signals after passing through these input terminals are sent to decoding circuits (compliant bit allocation coding circuits) 116, 117, and 118. The aforementioned decoding circuits 116, 117, and 118 use the aforementioned compliance bit allocation information to release bit allocation.

Next, the outputs of the aforementioned decoding circuits 116, 117, and 118 are sent to the IMDCT circuits 113, 114, and 115. In addition, the aforementioned IMDCT circuits 113, 114, and 115 are fed with block size information sent via the aforementioned terminals 123, 125, and 127. In these IMDCT circuits 113, 114, 115, the signal on the frequency axis is transformed into the signal on the time axis. The partial-band signals on the time axis from the IMDCT circuits 113, 114, and 115 are decoded into full-band signals by the IQMF circuits 112, 111. These decoded signals are output from the output terminal 110 to the subsequent circuit configuration.

According to the decoding device of Fig. 7, the encoded data sent from the aforementioned high-efficiency encoding device can be decoded by a small-scale configuration, and at the same time, a decoded signal with very excellent sound quality can be obtained.

[Effects of the present invention]

It can be seen from the above description that according to the high-efficiency encoding device of the present invention, all the bits that can be used for bit allocation are divided and used for: a fixed bit allocation pattern that is preset no matter how short the time is. This part, and the part that can be allocated based on the signal size of each block, and in addition to the fixed bit allocation pattern, even for the bit allocation that depends on the signal size of each block It is also due to the weight of the band corresponding to the block, regardless of whether the signal spectrum is flat, or not only a case where there is only an isolated spectrum, even if there are many steep spectrums in the input In the case of a signal in the form of this, it is not necessary to go through complex masking calculations to achieve a bit allocation that meets the sense of hearing, and only a small-scale and low bit rate can achieve high-efficiency coding with good sound quality.

<p>10Input terminal of high efficiency coding device</p><p>11,12Band segmentation filter</p><p>13,14,15MDCT circuit</p><p>16, 17, 18Compliant bit allocation coding circuit</p><p>19, 20, 21Block size determination circuit</p><p>22, 24, 26Code output terminal</p><p>23, 25, 27Block size information output terminal</p><p>122, 124, 126Code input terminal</p><p>123, 125, 127Block size information input terminal</p><p>116, 117, 118Compliant bit allocation decoding circuit</p><p>113, 114, 115IMDCT circuit</p><p>112, 111IQMF circuit</p><p>110High efficiency decoding device output terminal</p><p>302The total number of bits that can be used output circuit</p><p>303Calculating circuit of energy of each band</p><p>304Energy-dependent bit distribution circuit</p><p>305Fixed bit distribution circuit</p><p>306Bit sum calculation circuit</p><p>307Bit allocation output terminal for each band</p><p>308Spectrum smoothness calculation circuit</p><p>309Bit division rate determination circuit</p>

Fig. 1 is a block circuit diagram of a configuration example of a high-efficiency encoding device according to an embodiment of the present invention.

Fig. 2 is a diagram for explaining the signal frequency and time division of the high-efficiency encoding device according to the embodiment of the present invention.

Figure 3 is a diagram showing the configuration (an example of the number of bit allocations) of the compliant bit allocation coding circuit of the high-efficiency encoding device according to the embodiment of the present invention.

Figure 4 shows the three-masted spectrum.

Fig. 5 is a diagram showing an example of the frequency spectrum of the input signal in the high-efficiency encoding device according to the embodiment of the present invention.

Fig. 6 is a diagram for explaining the allocation of bits of the input signal by the high-efficiency encoding device of this embodiment.

Fig. 7 is a block circuit diagram showing an example of the configuration of the high-efficiency decoding apparatus of this embodiment.