JPH05114863A - High-efficiency encoding device and decoding device - Google Patents

Abstract

PURPOSE:To obtain excellent characteristics even for an isolated spectrum by varying a noise shaping factor according to a signal relating to the level of an input signal or the smoothness of the spectrum of the input signal. CONSTITUTION:An MDCT(Modified Discrete Cosine Transform) coefficient is supplied to an input terminal 301 and a band-by-band energy calculating circuit 302 calculates signal energy regarding respective bands obtained by fractionalizing a critical band in the critical band or a high range. A reducing circuit 303 for high-range signal level reduces the high-range signal level more and more as the energy of a band of >=4KHz is smaller and smaller. A spectrum smoothness calculating circuit 305 makes the noise shaping factor small judging that the spectrum is smooth as the ratio of the entire-band addition value of the square of the difference value of energy between adjacent blocks to the entire-band addition value of energy is smaller and an energy dependency bit distributing circuit 304 distributes bits.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to high-efficiency coding of digital data such that input digital data is coded by so-called high-efficiency coding, transmitted or recorded / reproduced, and decoded to obtain a reproduced signal. The present invention relates to a device and a decoding device.

[0002]

2. Description of the Related Art There are various techniques for high-efficiency coding of audio or voice signals. For example, an audio signal on the time axis is not divided into blocks but is divided into a plurality of frequency bands by a filter and coded. As a non-blocking frequency band division method, the band division coding (sub band
Coding: SBC), etc.
A block frequency band division method that transforms a signal on the time axis into a signal on the frequency axis (orthogonal transform), divides it into multiple frequency bands, and encodes each band, so-called transform coding, etc. You can

A high-efficiency coding method combining the above-mentioned band-division coding and transform coding has also been considered. In this case, for example, after performing band-division by the above-mentioned band-division coding. , The signal in each band is orthogonally transformed into a signal on the frequency axis, and each orthogonally transformed band is encoded.

Here, as a filter for the above band division, for example, there is a QMF filter, which is 1976 RE.
Crochiere, Digital coding of speech in subbands, B
ellSyst.Tech. J. Vol.55, No.8 1976. Also ICASSP 83, BOSTON Polyphase Quadrature fil
ters-A new subband coding technique, Joseph H. Roth
Weiler describes an equal bandwidth filter partitioning technique. Next, as the above-mentioned orthogonal transform, for example, an input audio signal is divided into blocks in a predetermined unit time (frame), and fast Fourier transform (FFT), discrete cosine transform (DCT), modified DCT transform (MDCT) For example, the time axis is converted to the frequency axis by performing the above. ICASSP for MDCT
1987Subband / Transform Coding Using Filter Bank Des
igns Based on Time DomainAliasing Cancellation,
JPPrincen, ABBradley, Univ. Of Surrey RoyalMel
Bourne Inst. of Tech.

Further, as a frequency division width for quantizing each frequency component divided into frequency bands, for example, band division is performed in consideration of human auditory characteristics. That is, the audio signal may be divided into a plurality of bands (for example, 25 bands) with a bandwidth that is generally called a critical band and becomes wider in a higher band. Further, at the time of encoding the data for each band at this time, encoding is performed by predetermined bit allocation for each band or adaptive bit allocation (bit allocation) for each band. For example, when the coefficient data obtained by the MDCT processing is coded by the bit allocation, it is adaptive to the MDCT coefficient data for each band obtained by the MDCT processing for each block. Encoding will be performed with the allocated number of bits.

The following two methods are known as bit allocation methods. First, IEEE Transactions of Accoustics,
Speech, and Signal Processing, vol.ASSP-25, No.4,
In August 1977, based on the signal size of each band,
Bit allocation is done. In this method, the quantization noise spectrum becomes flat and the noise energy becomes minimum, but
Since the masking effect is not used auditorily, the actual noise feeling is not optimal.

Next, ICASSP 1980 The critical band c
oder-digital encoding of the perceptual requirem
In ents of the auditory system, MAKransner, MIT, a method of performing fixed bit allocation by using auditory masking to obtain a necessary signal-to-noise ratio for each band is described. However, in this method, even when the characteristic is measured with a sine wave input, the characteristic value is not so good because the bit allocation is fixed.

Here, in the former method of the above two methods,
In order to optimize the actual noise feeling, a fixed noise shaping factor is used when determining the bit allocation so that the spectrum of the quantization noise is auditorily matched. At this time, bit allocation is performed according to the following expression (1).

[0009]

[Equation 1]

In this equation (1), b (k) is the word length of the MDCT coefficient of the kth block, δ is the proper bias,
σ ² (k) is the signal power of the k-th block, D is the average quantization error power in all blocks, and γ is the noise shaping factor. The calculation is performed by changing δ so that the sum of b (k) of all blocks falls within all usable bits, and the optimum value is obtained.

However, even in this method, when the characteristic is measured with a sine wave input, the bit allocation does not sufficiently concentrate on the frequency where the sine signal exists, and the characteristic value is not so good.

[0012]

As described above, when bit allocation is performed according to the signal size of each band and the quantization noise energy is minimized, the auditory noise level is not minimized,
If a fixed noise shaping factor is introduced or fixed bits are allocated to each band in consideration of the masking effect, it is difficult to obtain good signal-to-noise characteristics when a sine wave is input.

The present invention has been made in view of the above circumstances, and is a bit allocation method that is desirable from the viewpoint of hearing and also has good characteristics for an isolated spectrum input such as a 1 kHz sine wave input. It is an object of the present invention to provide such a high-efficiency encoding device and decoding device.

[0014]

A high efficiency coding apparatus for digital data according to the present invention divides an input digital signal into a plurality of frequency bands and adaptively changes a block size for each frequency band. After that, the spectrum data is obtained by performing orthogonal transformation, and this spectrum data is coded by adaptively assigning bits so that the spectrum of the quantization noise is auditorily adapted to each other using the noise shaping factor for each critical band. A high-efficiency encoding apparatus for obtaining data, wherein the noise shaping factor is changed according to at least one of the signals related to the magnitude of the input signal or the smoothness of the spectrum of the input signal, To solve.

In this case, it is preferable to use a signal related to the difference information between adjacent spectra as an index of the smoothness of the spectrum. Further, the amount of calculation can be reduced by using the signal related to the difference information of the signal size between blocks for the block floating as an index of the smoothness of the spectrum. Also, the smaller the signal size, the smaller the high-frequency signal size is estimated to be the basis for bit allocation. Since the minimum audible level in the high frequency band is increased, effective bit allocation is performed. Useful above. Particularly, the smaller the signal size above the minimum frequency of the minimum audible level of approximately 4 kHz or higher, the more effective it is to perform bit allocation based on the spectrum in which the high frequency spectrum is reduced.

Further, the above problem can also be solved by a high efficiency decoding device which decodes a signal which is coded by the high efficiency coding device and transmitted or recorded / reproduced.

[0017]

According to the present invention, even when the spectrum is dispersed like a music signal, the noise level seen from the auditory sense can be lowered by the masking effect, and even when the sine wave is input, bits are set in a large band of the signal. As they are collected, the signal-to-noise ratio can be increased.

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An input digital signal such as an audio PCM signal is subjected to band division coding (SBC) and adaptive transform coding (A).
An embodiment in which high efficiency coding is performed using the techniques of TC) and adaptive bit allocation (APC-AB) will be described with reference to FIG.

In the concrete high-efficiency encoder shown in FIG. 1, the input digital signal is divided into a plurality of frequency bands by a filter and the like, and orthogonal transformation is performed for each frequency band to obtain the obtained frequency axis. The spectrum data is coded by adaptively allocating bits for each so-called critical band width in consideration of human auditory characteristics described later. Of course, the non-blocking frequency division width by a filter or the like may be an equal division width. Furthermore, in the embodiment of the present invention, the block size (block length) is adaptively changed according to the input signal before the orthogonal transformation, and the critical band width is further subdivided in the critical band unit or in the high band. Floating processing is performed on the converted blocks. The critical band is a frequency band divided in consideration of human auditory characteristics, and the band of a pure tone when the pure tone is masked by narrow band noise of the same strength near the frequency of the pure tone. That is. This critical band has a wider bandwidth as it goes to higher frequencies, and the above 0-2
The entire frequency band of 0 kHz is divided into 25 critical bands, for example.

That is, in FIG. 1, an audio PCM signal of 0 to 20 kHz, for example, is supplied to the input terminal 10. The input signal is divided into a band of 0 to 10 kHz and a band of 10 to 20 kHz by a band dividing filter 11 such as a so-called QMF filter, and a signal of a band of 0 to 10 kHz is 0 to 0 by a band dividing filter 12 such as a so-called QMF filter. It is divided into a 5 kHz band and a 5 kHz to 10 kHz band. 10k from band splitting filter 11
The signal in the band of up to 20 kHz is Mo, which is an example of an orthogonal transformation circuit.
dified Discrete Cosine Transform (MDCT) circuit 1
3 to 5k to 10k from the band division filter 12
The Hz band signal is sent to the MDCT circuit 14, and the 0 to 5 kHz band signal from the band division filter 12 is sent to the MDCT circuit 15 for MDCT processing.

Here, each MDCT circuit 13, 14, 15
FIG. 2 shows a concrete example of the block size. In the specific example of FIG. 2, the frequency band is widened and the time resolution is increased (the block length is shortened) on the higher frequency side. That is, a signal in the 0-5 kHz band on the low frequency side and 5 k in the middle frequency band
For a signal in the 10 kHz band, the number of samples in one block b _L , b _{M is} , for example, 256 samples.
Each length of 1/2 with respect to the k~20kHz band signal, b _H of the low and middle band side block b _L, b _M
Blocked with lengths of BL _L / 2 and BL _M / 2. In this way, the number of orthogonal transform block samples in each band is the same. Further, each band can be further adaptively divided into ½, ¼, etc. blocks assuming a case where a signal temporal change is large.

Referring again to FIG. 1, each MDCT circuit 13,
The spectrum data on the frequency axis or MDCT coefficient data obtained by MDCT processing at 14 and 15 is a block obtained by further subdividing the critical band width in each so-called critical band (critical band) or in a high band. It is collected for each and sent to the adaptive bit allocation encoding circuits 16, 17, and 18.

Adaptive bit allocation coding circuits 16, 17, 1
In accordance with the number of bits allocated to each critical band (critical band) in the high band or to each block obtained by further subdividing the critical band width in the high frequency band, each spectrum data (or MDCT coefficient data) is re-reproduced. I try to quantize. The data encoded in this way is taken out via the output terminals 22, 24 and 26. At this time, floating information indicating what kind of signal magnitude is normalized and bit length information indicating what bit length the quantization is performed are sent at the same time.

FIG. 3 shows an adaptive bit allocation encoding circuit 16,
It is a functional block diagram which shows the specific example of the internal function of 17, 18, and the output of each MDCT circuit 13, 14, 15 in FIG. 1 is for every band via the input terminal 301 of the adaptive bit allocation function part 300 of FIG. Energy of each critical band (critical band), or energy of each band obtained by further subdividing the critical band in the high frequency band, is sent to the energy calculation circuit 302 of FIG. It can be obtained by calculating the square root of the mean. Instead of the energy for each band, a peak value, an average value, etc. of the amplitude value may be used. FIG. 4 shows, as an output from the energy calculation circuit 302, an example of a spectrum of a sum value for each band obtained by further subdividing the critical band width in the critical band (critical band) or in a high band. However, in order to simplify the illustration in FIG. 4, the number of bands in the number of critical bands (critical bands) or the band in which the critical band width is further subdivided in the high range is 12 bands (B1). ~ B12).

The operation of the adaptive bit allocation circuits 16, 17, 18 will be further described with reference to FIG. Now, the magnitude of the MDCT coefficient for each block is the MDCT circuits 13, 14,
The MDCT coefficient obtained at 15 is supplied to the input terminal 300. The supplied MDCT coefficient is given to the energy calculation circuit 302 and the high frequency signal level reduction circuit 303 for each band. The energy calculation circuit 302 for each band calculates the signal energy for the critical band or each band obtained by further subdividing the critical band in the high band. The energy regarding each band calculated by the energy calculation circuit 302 for each band includes a high-frequency signal level reduction circuit 303 and a spectrum smoothness calculation circuit 3.
05 and the noise shaping factor determination circuit 306. First, in the high frequency signal level reduction circuit 303, the higher the high frequency signal level is reduced, the smaller the energy in the band of 4 kHz or higher. This makes it possible to take into consideration the influence of the minimum audible limit characteristic in the high frequency range. Next, in the spectrum smoothness calculation circuit 305, the smaller the ratio between the sum of the squared difference values of the energy of adjacent blocks over the entire band and the total value of the energy over the entire band, the smoother the spectrum, And reduce the noise shaping factor. Energy-dependent bit allocation circuit 304
Uses the noise shaping factor from the noise shaping factor determination circuit 306 and the high-frequency-reduced signal spectrum from the high-frequency signal level reduction circuit 303 to perform bit allocation according to the above equation (1). The value of this bit allocation is taken out via the output terminal 307 and used in quantization.

Next, bit allocation in the case where the spectrum of the input signal is smooth and in the case where the spectrum is not smooth and the resulting quantization noise spectrum will be described with reference to FIGS. 5 to 8. That is, FIG. 5 shows a state of bit allocation when the signal spectrum is flat, and FIG. 6 shows a state of quantization noise (noise spectrum) corresponding to the bit allocation. Further, FIG. 7 shows the state of bit allocation when the spectrum is not smooth, for example, when the tonality of the signal spectrum is high and the pitch of the signal is audibly perceived, and the frequency is biased. The state of noise (noise spectrum) is shown in FIG. 6 and 8, the curve a shows the signal level and the curve b shows the quantization noise level.

That is, FIGS. 5 and 6 show the case where the spectrum of the signal is relatively flat, and show that the noise shaping factor approaches approximately -1.
On the other hand, as shown in FIGS. 7 and 8, when the signal spectrum is not smooth, the quantization noise spectrum can be brought close to white by bringing the noise shaping factor close to 0. This achieves improved performance with isolated spectrum input signals.

FIG. 9 shows a decoding circuit for decoding the signal thus highly efficient coded, after transmitting or recording / reproducing it. The quantized MDCT coefficients in each band are decoded circuit input terminals 122 and 12
The block size information given to and used in No. 4, 126 is given to the input terminals 123, 125, 127. The decoding circuits 116, 117 and 118 cancel the bit allocation using the adaptive bit allocation information. Next, the IMDCT (inverse MDCT) circuits 113, 114 and 115 convert the signals on the frequency axis into signals on the time axis. The signals on the time axis of these partial bands are IQMF (inverse QMF) circuit 11
2, 111 decoded into full band signal and output terminal 1
It is taken out from 10.

[0029]

As is apparent from the above description, according to the digital data high-efficiency coding apparatus of the present invention, the input digital signal is divided into a plurality of frequency bands, and each frequency band is adapted. After changing the block size, the spectrum data is obtained by performing orthogonal transformation, and this spectrum data is adaptively applied so that the spectrum of the quantization noise is auditorily adapted by using the noise shaping factor for each critical band. A high-efficiency coder for performing bit allocation to obtain coded data according to at least one of a signal related to a magnitude of an input signal or a smoothness of a spectrum of the input signal. This bit allocation method is also desirable audibly because it is changed, and like the 1 kHz sine wave input, Good characteristics even for standing spectrum input, without bit amount adjustment again and again, only 1
It is possible to realize the bit allocation obtained by performing the calculation once.

The same effect can also be obtained by a high-efficiency decoding device that decodes a signal that has been encoded by the high-efficiency encoding device for digital data and transmitted or recorded / reproduced.

[Brief description of drawings]

FIG. 1 is a block circuit diagram showing a configuration example of an encoding device as an embodiment of the present invention.

FIG. 2 is a diagram showing a specific example of frequency division and time division of a signal of the apparatus of the embodiment.

FIG. 3 is a functional block diagram for explaining an example of a bit allocation algorithm of an adaptive bit allocation encoding circuit used in the device of the embodiment.

FIG. 4 is a diagram showing a Bark spectrum.

FIG. 5 is a diagram showing an example of bit allocation at the time of inputting a signal having a substantially flat spectrum in the above embodiment.

FIG. 6 is a diagram showing an example of a quantization noise spectrum at the time of inputting a signal having a substantially flat spectrum in the above embodiment.

FIG. 7 is a diagram showing an example of bit allocation at the time of inputting a signal having high tonality according to the above-described embodiment.

FIG. 8 is a diagram showing an example of a quantization noise spectrum at the time of inputting a signal having high tonality according to the above-mentioned embodiment.

FIG. 9 is a block circuit diagram showing a configuration example of a decoding device for the encoding device of the above embodiment.

[Explanation of symbols]

10 ... High efficiency coding circuit input terminal 11, 12 ... QMF circuit 13, 14, 15 ... MDCT circuit 16, 17, 18 ... Adaptive bit allocation coding circuit 19, 20, 21 ... Block size determination circuit 22, 24, 26 ... Encoding output terminal 23, 25, 27 ... Block size information output terminal 122, 124, 126 ... Encoding input terminal 123, 125, 127 ... Block size information input terminals 116, 117, 118 ... Adaptive bit allocation decoding circuit 113, 114, 115 ... IMDCT circuit 112, 111 ... IQMF circuit 110 ... High efficiency decoding circuit output terminal 300. ..Adaptive bit allocation function unit 301 ... Adaptive bit allocation coding input terminal 302 ... Energy calculation circuit for each band 303 ... High band signal level Bell reduction circuit 304 ... Energy-dependent bit allocation circuit 305 ... Spectral smoothness calculation circuit 306 ... Noise shaping factor calculation circuit 307 ... Bit allocation amount output terminal

Claims (7)

[Claims] 1. An input digital signal is divided into a plurality of frequency bands, and a block size is adaptively changed for each frequency band, and then orthogonal transformation is performed to obtain spectrum data. A high-efficiency encoding device that obtains encoded data by adaptively allocating bits so that the spectrum of quantization noise is auditorily matched using a noise shaping factor for each band, wherein the noise shaping factor is: A high-efficiency coding device, characterized in that it is changed according to at least one of the signals related to the magnitude of the input signal or the smoothness of the spectrum of the input signal. 2. A high-efficiency decoding apparatus for decoding a signal transmitted or recorded / reproduced by the high-efficiency encoding apparatus for digital data according to claim 1 to obtain a reproduced signal. 3. The noise shaping factor is changed so that the quantization noise level becomes flatter as the input signal becomes larger or the spectrum of the input signal becomes smoother. High efficiency encoder. 4. The high efficiency coding apparatus according to claim 3, wherein a signal related to difference information between adjacent spectra is used as an index of spectrum smoothness. 5. The high-efficiency code according to claim 4, wherein a signal related to signal size difference information between blocks for block floating on the frequency axis is used as an index of spectrum smoothness. Device. 6. The high efficiency coding apparatus according to claim 4, wherein the smaller the size of the input signal, the more the bit allocation is performed based on the spectrum in which the high frequency spectrum is reduced. 7. The high-order according to claim 6, wherein the smaller the signal size above the minimum frequency of the approximate minimum audible level is, the more the bit allocation is performed based on the spectrum in which the high frequency spectrum is reduced. Efficiency encoder.