JP3304739B2 - Lossless encoder, lossless recording medium, lossless decoder, and lossless code decoder - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a high-definition encoder having no degradation by operating a run length compressing operation due to division into plural sub-bands, reduction in the number of bits on the noise floor of linear PCM and the decision of a scale factor. SOLUTION: Mode information is inputted to a mode information input terminal 44, and a full operating mode is set by a system control part 14. An input signal is inputted to a wide band audio data input terminal 41, and noise floor information and input signal band information of sub-band for each signal frequency band 24KHz are respectively impressed to a noise floor information input terminal 42 and an input signal band information input terminal 43. Information for commanding the encoding request of 1st mode and wide band audio mode is inputted to a mode information input terminal 44 and based on this command, a code system control part 14 sets a sub-band dividing filter 10. Besides, redundant run length compression parts 31-34 detect and compress the peaks of signal level in sub-bands at every prescribed time. Thus, the high- definition encoder can be provided.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a lossless encoder, a lossless recording medium, a lossless decoder, and a lossless code decoder for transmitting digital signals with high quality and without deterioration.

[0002]

2. Description of the Related Art Music recording and reproduction using digital signals such as compact discs and DATs are widely performed. For example, a compact disc has a sampling frequency of 44.1 kHz.
z, 16 bits are recorded by a linear coding method. In this system, it is theoretically impossible to reproduce a sound exceeding 22.05 kHz and obtain a dynamic range exceeding 98 dB. 22.0 for acoustic signals generated by live musical instruments
It was stated that there was no need to reproduce this component because it was out of the audible band despite containing components above 5 kHz. However, in recent years, studies have been conducted on the possibility that ultra-treble activates α-wave, which is a human brain wave, and it has begun to be considered that ultra-treble has some effect on brain wave. Whether or not it can be heard by humans can not be said unconditionally due to individual differences, but it has some physical and physiological influences and effects, and in order to leave higher sound quality as a cultural heritage in the future, it is extremely high in the reproduced signal It has been pointed out that regional components are important. In addition, while there is an actual sound whose dynamic range exceeds 100 dB and reaches 130 dB, clip distortion is likely to occur when this is expressed by a linear code of 16 bits. It has been pointed out that the dynamic range is insufficient because distortions due to quantization errors make the sound muddy.

[0003] Therefore, as shown in the radio and experiment magazine, February 1995, pages 100 to 101, published by Seibundo Shinkosha, LSB of 16-bit data is used, and 2 bits are used for this bit.
A method of recording music signal information of 2.05 kHz or more using ADPCM (method 1), and a radio technology magazine, April 1991, pp. 147 to 150, published by IAI Publishing Co., Ltd.
As shown on the page, a method of driving the quantization noise to 15 kHz to 22.05 kHz by using noise shaping and improving the dynamic range of the audibility (method 2)
Has been proposed.

As a next-generation format, the development of a high-density recording disk has been developed in recent years, and an SD (super-density disk) which has been studied as a base of a DVD (digital video disk or digital versatile disk) has been studied. ) There is a format. For the audio part of the SD format, AES
Issued by ES), 99th Convention 1995 October 6-9
New York Abstract No. 4121 (D-9) `` The Application of
a New High-Density Optical Disc for Audio ”, with a sampling frequency of 48 kHz
16bit linear PCM to sampling frequency 96k
A specification of a Hz / 24-bit linear PCM and an SD format proposal that greatly exceeds the specifications of a conventional CD has been disclosed. It is expected that a reproduction frequency band of about 45 kHz and a dynamic range of 140 dB can be achieved with this format proposal.

[0005]

[Problems to be Solved by the Invention] However, recent research "Riki Ohashi et al .: Difference in sound quality between LP and CD-Physiological / Kansei scientific study-" IEICE Technical Report HC9
4-06 (1994-06) Institute of Electronics, Information and Communication Engineers, "
It was confirmed that a signal reaching Hz was recorded in a reproducible state, and in the fortesimo part of the ethnic musical instrument Gamelan, a spectrum was obtained that wrapped around 50 kHz and reached 100 kHz. From this research result, it must be said that the band is still insufficient for faithfully reproducing such music even in the band of about 45 kHz in the SD format proposal. If the sampling frequency is further increased in the SD format proposal, for example, if it is expanded to 240 kHz, the reproduction band is extended to 100 kHz or more. The amount of information required for that is 5.76 (Mbps) per channel, 2c.
In the case of h stereo, it becomes huge at 11.52 (Mbps). Then, even if the DVD capacity is 4.7 GB, it is about 5
Only 0 minutes can be recorded, and the bit rate exceeds the temporary upper limit of 6.75 (Mbps), which is not practical.

An object of the present invention is to solve the above problems and to provide a lossless encoder, a lossless recording medium, a lossless decoder, and a lossless code decoder having a wide dynamic range and a high dynamic range.

[0007]

In order to achieve this object, a lossless coding apparatus according to the present invention inputs noise floor information and information indicating a signal band, and converts wideband audio data to a base band and a high band according to band information. Each sub-band is divided by a sub-band division filter that divides the sub-band into sub-bands. When the noise floor is high for each sub-band, a predetermined noise margin is secured to reduce the number of bits. Further, a plurality of redundant runs for detecting signal peak levels for a predetermined time in all or some of the sub-bands, determining a scale factor, and outputting signal data excluding a higher-order redundant run length. The encoded data is extracted from a length compression unit and a multiplexer for multiplexing the output data of the plurality of scale factors and the output data of the plurality of redundant run-length compression units.

In order to achieve the above object, a lossless decoding apparatus according to the present invention comprises a demultiplexer for reproducing a plurality of subband data and additional information such as a scale factor and a time code from input coded data, A redundant run length restoring unit for restoring a redundant run length based on a scale factor of a predetermined time section for each band; an oversampling unit for determining an oversampling frequency according to signal band information and oversampling; A sub-band synthesis filter for synthesizing data and outputting wide-band signal data to output wide-band audio data.

[0009]

BEST MODE FOR CARRYING OUT THE INVENTION A lossless encoder (hereinafter sometimes simply referred to as an encoder) of the present invention converts a sub-band into a signal band when high-output data is input in a wide band covering a maximum band. It has the effect of extending to 120 kHz and allocating up to 24 bits (144 dB or more), and has the ability to encode as an ultra-high frequency Hi-Fi signal. Normally, when the noise floor is high for each sub-band, a predetermined noise margin is secured and the number of bits is reduced, so that useless information is not allocated to data including only noise components. The minimum necessary bit rate according to the noise floor determined by the system is allocated. Further, in all or some of the sub-bands, a signal peak level at a predetermined time is detected, a scale factor is determined, high-order redundant run lengths are eliminated, and unnecessary headroom is eliminated. A code can be obtained in which the wideband audio data thus compressed is compressed to the minimum necessary bit rate without any loss.

Further, the lossless recording medium of the present invention can record the above-mentioned code as a signal. Further, the lossless decoding device of the present invention (hereinafter, sometimes simply referred to as a decoding device) receives the above-described code as a signal, and converts a plurality of subband data, a scale factor, a time code, and the like from input coded data. Regenerate additional information, restore the redundant run length based on the scale factor of the predetermined time section for each subband, determine the oversampling frequency according to the signal band information, oversample, and combine multiple subband data Therefore, if high-bandwidth and high-output data covering a maximum band is input according to the quality of the transmission signal, the sub-band is expanded to a signal band of 120 kHz and the 24-bit (144 bits) signal is output.
(dB or more) and has the ability to be encoded as an ultra-high-frequency hi-fi signal. Normally, when the noise floor is high for each sub-band, a predetermined noise margin is secured to reduce the number of bits, and only the noise component is reduced. No unnecessary information is allocated to the data of the subband, and the scale factor of the subband is determined, and high-efficiency wideband audio data excluding the higher-order redundant run length can be output.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram illustrating a lossless encoding device according to an embodiment of the present invention.
In the figure, reference numeral 10 denotes n (n
Is an integer of 2 or more), a subband division filter for dividing into subbands, 20 to 24 are subsampling units,
1 is a linear prediction encoder, 30 is a baseband code processing unit, 3
1 to 34 are redundant run-length compression units, 12 is a multiplexer, 13 is a code dither generator, and 14 is a code system control unit. As information representing the characteristics of the audio data to be input, noise floor information is input from a noise floor information input terminal 42, input signal band information is input from an input signal band information input terminal 43, and code mode information is input from a mode information input terminal 44. input. Broadband audio data is input from a wideband audio data input terminal 41, and encoded data is extracted from an encoded data output terminal 45.

The encoded data is transmitted to a transmission device (not shown,
Here, it is output to a high-density optical disk writing device), a recording format is formed, and it is recorded on a medium of the high-density optical disk (lossless recording medium 15 in FIG. 1). The lossless recording medium 15 is reproduced by a reproducing means (not shown), and the encoded data is decoded and taken out.

FIG. 2 is a block diagram showing a lossless decoding device according to the embodiment of the present invention. In the figure, reference numeral 50 denotes a demultiplexer for reproducing a plurality of subbands and additional information from input encoded data and extracting separated data;
Is a base band decoding processing section, 61 to 64 are redundant run length restoring sections, 51 is a linear predictive decoder, 70 to 74
Denotes an oversampling unit, 52 denotes a subband synthesis filter, and 54 denotes a decoding system control unit. Mode information for instructing the operation of the entire lossless decoding device is input from the mode information input terminal 94 to the decoding system control unit 54. The noise floor information and the input signal band information multiplexed on the data input from the coded data input terminal 91 are input to the demultiplexer 5.
At 0, it is separated and identified and input to the decoding system control unit 54. The decoding system control unit 54 controls the entire operation based on these pieces of information.

In the embodiment configured as above, various modes will be described in detail with reference to the drawings. First, a coding operation of the first mode lossless coding apparatus according to the present embodiment will be described. In FIG. 1, mode information is input to a mode information input terminal 44, and a mode of the entire operation is set by the coding system control unit 14. An input signal (here, the sampling frequency is 240 kHz and the word length is 24 bits) is input from the wideband audio data input terminal 41. Aside from this,
The noise floor information and the input signal band information of the sub-band every 24 kHz of the signal frequency band are input from the noise floor information input terminal 42 and the input signal band information input terminal 43, respectively. The mode information input terminal 44 receives information for instructing an encoding request for the first mode, that is, the wideband audio mode. The coding system control unit 14 sets the sub-band division filter 10 based on these commands. The subband division filter 10 divides the wideband audio data into five bands, that is, a base band B0 and subbands B1 to B4.

FIG. 3 is a spectrum diagram showing a sub-band division and sub-sampling method in the first mode. The sub-band division filter 10 divides the base band B0 into a signal frequency band of 24 kHz, the sub-band B1 into a band with a signal frequency of 24 kHz to 48 kHz, and thereafter sequentially divides the sub-bands B2 to B4 into a band up to 120 kHz. Preferably, it is composed of a polyphase quadrature mirror filter (hereinafter referred to as QMF) by an oversampling process of 8 times or more. The reason why the QMF is used is to cancel aliasing distortion that occurs near the band crossover.

Hereinafter, a description will be given with reference to FIGS. Output 100 of base band B0 of subband division filter 10
Is B0 in FIG. 3B. Fo in FIG.
vs is the QMF oversampling frequency of 384kHz
And Base band B0 of subband division filter 10
Is subsampled by a subsampling unit 20 at a subsampling frequency (fsub) of 48 kHz. The sub-sampled signal spectrum is converted into low-frequency data SB0 in FIG. Similarly, subbands B1, B
2, B3 and B4 signal spectrum B in FIG.
1, B2 of the same (f), B3 of the same (h) and B4 of the same (j) are sub-sampled to the low band, and
(E) SB1, (g) SB2, (i) SB3
And SB4 of the same (k). The bit rate of each sub-band is 1.152 (Mbps) because the sub-sampling frequency is 48 kHz and the number of bits is 24 bits, and the maximum bit rate from the base band to the sub-band B4 is 5.76 (Mbps). .

FIG. 5 is a diagram showing a sub-band division method in the first mode. In FIG. 5, the horizontal axis represents the signal frequency, the left vertical axis represents the signal level, and the right vertical axis represents the number of code bits. As described above, when high-output data is input in a wide band covering the maximum band, the sub-band has an effect of expanding the signal band to 120 kHz and allocating up to 24 bits (144 dB or more), and extends to the ultra-high sound range. Has the ability to encode Hi-Fi signals. While maintaining this maximum capability, lossless coding is performed by the following two method procedures, focusing on the signal components of each subband band.

One method is to know in advance the noise floor of the input signal so as not to uselessly allocate information. Due to background noise and microphone amplifier thermal noise even in recent excellent studio environments, the total dynamic range of sound pickup is at most about 120 dB without auditory sensation correction. Further, in the frequency band of 20 kHz to 100 kHz, the characteristics of the AD converter become severe, the dynamic range is further narrowed, and about 80 dB is required at most. Although future breakthroughs are expected, improvements are expected to be made at a moderate pace for the time being.

FIG. 6 shows the sound pickup noise floor of the broadband audio data in the above-described sound pickup environment, and shows an example in which code quantization noise is set based on information read from the noise floor information input terminal 42. FIG. In FIG. 6, the axis is the same as in FIG. The coding system control unit 14 controls the subsampling units 20 to 24 by determining a required quantization step size for each subband from the information input from the noise floor information input terminal 42. The base band B0 has a quantization step size of 24 bits according to the format selected from the SD format proposal. In addition, 20 bits and 16 bits can be selected. The sub-band B1 has a pickup noise floor of a band corresponding to the sub-band B1 from -105 to -115d.
Since the quantization step size is B, the quantization step size is set to 22 bits at which encoding quantization noise of -132 dB or less is obtained. In this way, the number of bits of the quantization step size is set with a margin of about 20 dB for the sound pickup noise floor. This is because even the signal buried under the pickup noise floor is faithfully encoded. Similarly, the quantization step sizes of the sub-bands B2 to B4 are determined, and the number of quantization bits is set to 20, 18, and 16 bits, respectively.

Since the quantization step size is specific to the sound collection system, it is preferable to fix the quantization step size in a sound collection unit, that is, in a program. The identification ID indicating the quantization step size information or the bit number information is multiplexed together with the signal data. Subsampling units 20 to 24
Is a subsampling frequency (fsub) of 48 kHz
And performs low-frequency conversion by sub-sampling, and outputs the data at the predetermined quantization step size. In this way, by assigning a useless amount of data to data consisting only of noise components while securing a margin, the minimum necessary bit rate according to the noise floor unique to the sound pickup system is assigned. In the example in FIG. 6, the maximum bit rate of each sub-band data is 1.152 (Mbp) for the base band B0.
s), subband B1 is 1.056 (Mbps), subband B
2 is 0.960 (Mbps), subband B3 is 0.864 (M
bps) and the sub-band B4 is 0.768 (Mbps). The total bit rate is 4.800 (Mbps), which can be reduced to about 83% of the original bit rate of 5.76 (Mbps) shown in FIG.

In the case of linear quantization, there is a possibility that the sound spectrum has an eigen spectrum at a minute level. Therefore, by adding dither from the code dither generator 13, the eigen noise spectrum can be scattered. In order to enhance the effect, it may be more preferable. For this purpose, the output of the code dither generator 13 may be added to the data of the wideband audio data input terminal 41. The outputs of the sub-sampling units 20 to 24 may be passed through the linear predictive encoder 11 as preprocessing for enhancing the compression effect of the redundant run-length compression units 31 to 34. Although the linear prediction encoder 11 does not have to obtain the basic function, if the linear prediction coder 11 is inserted and the prediction residual is coded, the linear prediction coder 11 has the function of suppressing the peak of the low band spectrum, and the signal spectrum distribution is low. The spectrum can be flattened in the case of the band concentration and high band attenuation type.

The output of the subsampling units 20 to 24 or the linear predictive encoder 11 is output to the baseband code processing unit 30.
And to the redundant run-length compression units 31 to 34. The base band code processing unit 30 has the same fixed scale factor as the input data. Next, the second method will be described. The redundant run length compression units 31 to 34 detect the peak of the signal level of the sub-band in units of a program or every predetermined time, and set a scale factor.
The redundant run length part of the upper bits is deleted, and the scale factor is transmitted instead of the upper bits. In this way, the redundant run length is compressed.

Next, an example of encoding actual music will be described. FIG. 7 shows an example in which a spectrum reaching 50 kHz is observed at 50 kHz in the fortesimo part of the folk instrument Gamelan. The thick broken line in FIG. 7 is a plot of the peak level in the program. Thus, in the natural world, components up to 100 kHz are observed, but the energy of sound tends to decrease as the frequency becomes higher. The redundant run-length compression units 31 to 34 determine the scale factor based on the peak level in each subband and delete and compress the upper redundant run-length.

As shown in FIG. 7, the sub-band B1 is set to have a scale factor of 4, and 8, 12, and 13 in that order. As a result, the sub-band B1 can be expressed without leaving all information by 18 bits by deleting 4 bits out of 22 bits. Similarly, the sub-band B2 has 12 bits, the sub-band B3 has 7 bits, and the sub-band B4 has 4 bits, so that all information can be expressed without any deterioration. Coding without deterioration is called lossless coding. As described above, all information can be restored from the data which has been lossless-encoded with the scale factor without any deterioration by decoding after transmission.

Here, the transmission bit rate of each sub-band data in FIG. 7 is 1.152 (Mbp) for the base band B0.
s), sub-band B1 is 0.864 (Mbps), sub-band B
2 is 0.576 (Mbps), subband B3 is 0.336 (M
bps) and sub-band B4 are 0.192 (Mbps).
The total bit rate is 3.120 (Mbps). This can be reduced to about 54% of the bit rate of 5.76 (Mbps) without the noise floor and run length compression shown in FIG. As a result, 6.24 (Mbps) is obtained even for two channels, which is the provisional upper limit of the SD format provisional allocation.
Does not exceed 75 (Mbps).

FIG. 8 is a diagram showing the relationship between the spectrum of the part of Fortissimo and the bit rate when encoding classical music. Since the sound collecting system environment is the same as that of FIG. 7, the noise floor and the code quantization step size can be made the same, but the compression effect of the redundant run length is even greater because the peak spectrum in the high frequency range is lower than that of the gamelan music. In FIG. 8, the required number of bits and the bit rate are respectively 15 bits / 0.72 (Mbps) for subband B1, and 8 bits / bit for subband B2.
0.384 (Mbps), sub-band B3 is 5 bits / 0.2
4 (Mbps), sub-band B4 is 4 bits / 0.192 (Mbp
s), and the total bit rate becomes 2.688 (Mbps). This is about 4 of the bit rate of 5.76 (Mbps) without the noise floor and run length compression shown in FIG.
It can be reduced to 7%. As a result, even with two channels, 5.3
76 (Mbps), which does not exceed about 6.75 (Mbps), which is the upper limit of DVD allocation, and when a DVD is inserted, the recording time is about 2 hours, which is practically excellent.

In the above description, the reason why the base band B0 is not compressed and remains at 1.152 (Mbps) is to make it common with the audio signal format of SD. S
In the D format, combinations of the sampling frequency, the number of bits, and the number of channels are as follows. Sampling frequency: 48 kHz, 96 kHz Number of bits: 16 bits, 20 bits, 24 bits Number of channels: 2 to 8 channels From the viewpoint of ensuring compatibility, it is preferable to select the base band B0 from these. This allows SD
In the basic format and the base band. In other words, there is a possibility that even a player that supports only the basic format of SD can realize compatibility in which data in the common part can be extracted from the data of the ultra-high fidelity disc and reproduced. Further, in the ultra-high-fidelity disc player, there is a possibility that the playback section of the disc on which the sound of the SD basic format is recorded and the base band processing section can be shared, which is advantageous in terms of cost reduction.

FIG. 4A illustrates a method of dividing a sub-band in the second mode. Basal band B0
To the signal frequency 48 kHz, the sub-band B1 to the band of the signal frequency 48 kHz to 72 kHz, and the sub-band B2
To subband B3 are sequentially divided into bands up to 120 kHz. QMF oversampling frequency (fovs)
Is set to 384 kHz. The signal spectrum of the output 100 of the base band B0 of the sub-band division filter 10 is shown in FIG.
(B). This is the sub-sampling frequency (fsub0)
The signal spectrum subsampled at 96 kHz is shown in FIG.
(c). Thus, the data is converted to low-frequency data. Similarly, the signal spectra of the sub-bands B1, B2 and B3 are as shown in (d), (f) and (h) of FIG. 4, and the spectra sub-sampled to low frequencies at the sub-sampling frequencies fsub1 to fsub3 are respectively shown in FIG. (E),
(G) and (i).

FIG. 9 is a diagram showing an example of encoding characteristics of the lossless encoding device of the second mode when a signal of a gamelan music of gamelan music having an extremely large treble range is input. When wideband audio data indicated by a broken line in FIG. 9 is input, the total bit rate is 3.36.
(Mbps). This value also does not exceed 6.75 (Mbps) for the two channels, which is the provisional allocation upper limit of the SD format.

FIG. 10 is a diagram showing an example of encoding characteristics when the number of bits of wideband audio data is 20 bits. This information is input from the noise floor information input terminal 42, and the sub-band is set by the coding system control unit 14. In this case, the maximum is 2 including the base band B0.
It is 0 bit. The total bit rate in this case is 2.83
2 (Mbps).

FIG. 11 is a diagram showing an example of encoding characteristics when the number of bits of wideband audio data is 16 bits. In this case, the maximum is 16 including the base band B0.
Bit. The total bit rate in this case is 2.20
8 (Mbps). Next, FIG. 12 is a diagram illustrating an example of encoding characteristics when the original sampling frequency of the wideband audio data is 96 kHz and the signal frequency band is 48 kHz. In this case, this information is inputted from the input signal band information input terminal 43, the required subband is obtained by the coding system control unit 14, and the operations of the subband division filter 10 and the subsampling units 20 to 24 are controlled. Bits are allocated only to the base band B0 and the sub-band B1, and no bits are allocated to the sub-bands B2 to B4. The total bit rate in this case is 2.0
16 (Mbps).

FIG. 13 is a diagram for explaining a frequency division method in the lossless encoder of the third mode. This is an embodiment in which the signal frequency band of the base band B0 is 48 kHz, and the band of the sub-band B1 is two, from 48 kHz to 120 kHz. FIG. 14 is a diagram illustrating a frequency division method in the lossless encoding device of the fourth mode. This is an embodiment in which the signal frequency band of the base band B0 is 24 kHz and the band of the sub-band B1 is two, from 24 kHz to 120 kHz. Detailed description is omitted.

The wide-band audio data output in the various modes described above is output to a transmission device, here, a writing device (not shown) for a high-density optical disk, and a recording format is formed to form a medium for the high-density optical disk (lossless recording medium 15). ). The medium is reproduced by reproducing means (not shown) to extract a decoded signal. Next, the operation of the lossless decoding device will be described in detail.

In FIG. 2, mode information for instructing the operation of the entire lossless decoding device is input from the mode information input terminal 94 to the decoding system control unit 54. Further, the demultiplexer 50 separates and identifies noise floor information and input signal band information multiplexed on data input from the encoded data input terminal 91 and inputs the information to the decoding system control unit 54. The decoding system control unit 54 controls the entire operation based on these pieces of information. The demultiplexer 50 reproduces a plurality of subbands and various IDs such as noise floor information 92 and input signal band information 93 from the input encoded data, extracts separated data, and supplies the separated data to the decoding system control unit 54. The base band decoding processing unit 60 performs a process dedicated to the base band, and the redundant run length restoring units 61 to 64 restore the redundant run length. The linear prediction decoder 51 turns on / off the operation based on the ID indicating the presence or absence of the linear prediction process, and the oversampling units 70 to 74 oversample the subband data and synthesize the wideband audio data with the subband synthesis filter 52. Broadband audio data is output from a wideband audio data output terminal 95. Thereafter, a wideband high-bit DA converter (not shown) reproduces a wideband audio analog signal.

A case where the lossless recording medium on which the ethnic instrument gamelan music shown in FIG. 7 is recorded is reproduced to perform lossless decoding will be described. As shown in FIG. 7, the subband B1 has a scale factor of 4, and 8, 12, and 13 in that order. Since the subband B1 expresses all the information by 18 bits by removing 4 bits out of 22 bits, 2 bits of the noise floor are added to the LSB side, and 4 bits of the redundant headroom are added to the MSB side to obtain 24 bits. Restore to bit data. Similarly, since all information is represented by 12 bits for subband B2, 7 bits for subband B3, and 4 bits for subband B4, bits are added to the LSB side and the MSB side, respectively, to restore 24-bit data. . As described above, since the lossless-encoded data with the scale factor is used to restore all information by the transmission and decoding device, no deterioration occurs. In this way, the transmission bit rate of each sub-band data is 1.1 for the base band B0.
52 (Mbps), subband B1 is 0.864 (Mbps), subband B2 is 0.576 (Mbps), and subband B3 is 0.
336 (Mbps) and sub-band B4 are 0.192 (Mbps) and the total bit rate is 3.120 (Mbps). Restore to bit data. This wideband audio data is converted into a wideband audio analog signal by a high-performance DA converter (not shown) having a sampling frequency of 240 kHz and 24 bits.

There are many portable and vehicle-mounted applications among audio reproducing devices. For such applications, power saving as well as miniaturization and weight reduction are important. Because the noise floor of the reproduction environment is high, it is meaningless to reproduce the high dynamic range at the time of sound pickup. Decoding signal processing and DA conversion are performed so that the minimum dynamic range and frequency characteristics are required according to the application. The power consumption can be reduced by substituting the required performance.

A case where the lossless recording medium on which the ethnic instrument gamelan music shown in FIG. 7 is recorded is reproduced to perform lossless decoding will be described. Subband B3 and subband B4
Omit processing. In other words, the sub-band up to 120 kHz is reproduced, for example, up to 72 kHz. As a result, the sub-band B3 and the sub-band B4 in the redundant run-length restoring units 61 to 64, the oversampling units 70 to 74, and the sub-band synthesis filter 52 can be reduced. As for the number of bits, the maximum band number is set to, for example, 18 bits, so that the baseband decoding processing unit 6
0, the operation word length of the redundant run-length restoration unit, the oversampling unit, and the sub-band synthesis filter 52 can be reduced to 18 bits. The power consumption of the lossless decoding device can be suppressed by these processing rates and circuit reduction.

Next, a case will be described in which the lossless recording medium of the present invention is reproduced by a reproducing apparatus that supports only the SD format specifications, that is, an apparatus in which the combination of the sampling frequency and the number of bits is limited as follows. Sampling frequency: 48 kHz, 96 kHz Number of bits: 16 bits, 20 bits, 24 bits Number of channels: 2 to 8 channels The base band B0 in the first mode of the embodiment is the same as 48 kHz, 24 bits, and 2 channels in the SD format. , The sub-band B1 to the sub-band B4 are framed to form a lossless recording medium which is recorded in the area up to the remaining six channels.
By reproducing only the channel, the reproduction of the base band B0 can be performed by a reproducing apparatus that supports only the SD format specification. Even if the player supports only the basic format of SD, there is a possibility of realizing compatibility in which data of the common part can be extracted from the data of the ultra-high fidelity disc and reproduced.

Further, in the case of reproducing a disc on which sound in the basic format of SD is recorded in a reproducing apparatus equipped with the decoding device according to the embodiment of the present invention, the base band B0 in the first mode or the second mode is used. Can be shared. The baseband decoding unit 60 and the oversampling unit 70 can be shared to output audio data.

The embodiment of the present invention is configured with the format and apparatus of the framework so as to realize the highest possible performance. There is a great distance between this maximum frame and the attainment of the environment that can be realized for the time being and the attained specifications of device and circuit technology. Therefore, for the time being, the lossless decoding device, lossless recording medium, and lossless decoding device of the present invention are produced and sold by setting parameters based on realistic specifications, and a device or circuit that can collect sound in a better environment in the future and further improve performance is provided. When is developed, future specifications can be improved simply by changing the parameters without changing the basic format. Therefore, it is possible to independently improve the specification characteristics of the lossless encoding device, the lossless recording medium, and the lossless decoding device, and it is not necessary to change the characteristics at the same time. This is a scalable format that incorporates future specification expandability.

The redundant run-length compressing sections 31 to 34 and the redundant run-length restoring sections 61 to 64 perform transmission by scaling from a peak level for a predetermined time, but it goes without saying that other compression methods may be used. An advanced one, for example, a table reference type vector quantization may be used. Oversampling frequency 38
The frequency of 4 kHz is not limited to this. If the operation and characteristics of the sub-band splitting filter 10 and the sub-band synthesizing filter 52 can be obtained, the same effect can be obtained at other frequencies. Similarly, the sub-sampling frequency,
The number of input and output bits of the encoding device is not limited. Further, the bandwidth of the sub-band may not be uniform, but is preferably uniform for simplicity of the circuit configuration.

In the embodiment of the present invention, the lossless encoding device and the lossless decoding device are separated from each other. However, at the time of encoding, the digital signal processor and the memory use the subband division filter 10, the subsampling units 20 to 24,
Base band code processing unit 30, redundant run-length compression unit 3
1 through 34 and the multiplexer 12 to obtain a code output, and at the time of decoding, the operation mode is changed by switching the system control unit, so that the demultiplexer 50, the base band decoding processing unit and the digital signal processor and memory are used. The code decoding device can be configured by extracting and outputting a decoded output by configuring and operating the 60, the redundant run-length restoring units 61 to 64, the oversampling units 70 to 74, and the subband synthesis filter 52. In this case, for example, when applied to a recording / reproducing apparatus, most circuits can be shared for recording and reproduction, and the circuit can be simplified.

[0043]

As described above, according to the present invention, based on the input of noise floor information and signal band information, the number of bits is reduced by the sub-band division of the multiple and the noise floor in the linear PCM, and the scale factor is determined. In the event that high-output data is input in a wide band up to the maximum band, the band is automatically expanded to 120 kHz and allocated to 24 bits, and ultra-high band hi-fi While retaining the ability to encode as a signal, the effect of efficiently lossless encoding natural sounds can be obtained. This code can be recorded on a medium. In addition, since the decoding apparatus reproduces the medium on which this code is recorded and obtains the original wideband audio data based on the noise floor and the redundant run-length information, the encoding parameters are losslessly recorded according to the technological progress. The effect that the decoding device can be linked by writing to the medium, or by providing limited change of the reproduction specification according to the required bandwidth and the number of bits of the reproducing device, a more power-saving and more economical decoding device can be provided. To play. In addition to the above effects, there are the following specific effects.

(A) At a low bit rate of 3.12 (Mbps) per channel, it is possible to simultaneously realize both a high dynamic range of 140 dB or more and a wide band of Nyquist frequency of 120 kHz. 44. The turbidity of the sound due to the deterioration of the distortion rate at a minute level is eliminated, and the original sound of the ultrahigh frequency signal from 20 kHz to 120 kHz can be reproduced.
It is possible to realize a natural recording / reproducing operation that escapes from the limited space of 1 kHz 16 bits and is infinitely transparent and reproduces up to a high frequency range, a medium for recording this signal and a reproducing sound field.

(B) Since the energy in the ultrasonic band in the music signal is small, lossless encoding can be performed with 24-bit accuracy when the signal is only in the ultra-high frequency range or when the signal is mainly in the ultra-high frequency range. As a result, an increase in the average bit rate can be suppressed. (C) Since the required bit rate is determined by the product of the reproduction band and the dynamic range, the highest performance device / circuit that can be realized for the time being can be coded at a low bit rate, and if the performance is improved in the future, it can be freely adjusted according to progress Can be adjusted by adjusting the bit rate.

(D) With a common medium, it is possible to reproduce even a portable type simplified decoding device aiming at power saving. (E) Since the base band can be made similar to the conventional format, compatibility with the conventional format can be obtained in the frequency band of the base band.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a lossless encoding device according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a lossless decoding device according to the embodiment;

FIG. 3 is a spectrum diagram showing a subband division and subsampling method in a first mode.

FIG. 4 is a spectrum diagram showing a sub-band division and sub-sampling method in a second mode.

FIG. 5 is a diagram showing a sub-band division method and a maximum number of allocated bits in a first mode.

FIG. 6 is a diagram showing a relationship between a pickup noise floor and code quantization noise in the first mode.

FIG. 7 is a diagram showing an example in which a spectrum reaching 100 kHz is observed in the part of the folk instrument Gamelan Fortessimo in the first mode, and the relationship between the bit rate when encoding the spectrum.

FIG. 8 is a diagram showing a bit rate when encoding classical music in the first mode.

FIG. 9 shows that the base band B0 is set to 96 k in the second mode.
Figure showing the sub-band division method and the maximum number of allocated bits when using Hz / 24 bits

FIG. 10 is a diagram illustrating an example of encoding characteristics when the number of bits of wideband audio data is 20 bits.

FIG. 11 is the same as FIG.
FIG. 7 is a diagram showing an example of encoding characteristics in the case of 6 bits

FIG. 12 is a diagram showing an example of encoding characteristics when the original sampling frequency of wideband audio data is 96 kHz and the signal frequency band is 48 kHz.

FIG. 13 is a diagram illustrating a sub-band division method and a maximum number of allocated bits in a third mode.

FIG. 14 is a diagram showing a sub-band division method and a maximum number of allocated bits in a fourth mode.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 10 Subband division filter 11 Linear prediction encoder 12 Multiplexer 13 Code dither generator 14 Code system control unit 20-24 Subsampling unit 30 Baseband code processing unit 31-34 Redundant run-length compression unit 50 Demultiplexer 51 Linear prediction decoder 52 Subband synthesis filter 54 Decoding system control unit 60 Base band decoding processing unit 61-64 Redundant run-length restoration unit 70-74 Oversampling unit

Claims (21)

(57) [Claims] 1. A inputs the wideband audio data, codes the input information representative of the signal band of the wide band audio data, and controls the entire operation determining the number of divided width based on band splitting the signal bandwidth information A system control unit, a sub-band division filter that divides an input signal into a predetermined number of sub-bands according to a command of the encoding system control unit, a sub-sampling unit that sub-samples a plurality of outputs of the sub-band division filter, A plurality of redundant run-length compression units for detecting a signal peak level at a predetermined time in all or some of the sub-bands, determining a scale factor, and outputting signal data excluding a higher-order redundant run-length And the output data of the plurality of scale factors and the plurality of redundant run length compression units. Lossless coding device that includes a multiplexer for outputting the encoded data by multiplexing, respectively. 2. The lossless coding apparatus according to claim 1 , wherein the scale factor of the subband is fixed at least in a unit of a sound collection program regardless of inside or outside of a predetermined time section. 3. A base band being the lowest band of the plurality of sub-bands, lossless according to claim 1, upper redundant run length excluding process 16 bits to 24-bit linear PCM without by scale factor Encoding device. 4. A inputs the wideband audio data, enter the noise floor information of the wide band audio data, and sub-band dividing filter for dividing the wide band audio data into a predetermined number of sub-bands, the sub-band dividing filter A sub-sampling unit for sub-sampling a plurality of outputs, and a quantization step size in each sub-sampling unit determined based on the noise floor information, and each of the sub-bands is determined by the determined quantization step size. A coding system control unit for setting a quantization step size; and a multiplexer for multiplexing output data of each sub-band and outputting coded data. The noise floor information includes background noise and recording noise of a recording studio.
Information about the noise characteristics of the
Only, or when connecting the digital audio output directly
Is the sampling frequency of the digital audio output and
And information on word length or mixing console
When using a tool or various effectors,
A lossless coding apparatus characterized by including information on noise characteristics. 5. A wideband audio data is input, and noise floor information of the wideband audio data and information indicating a signal band of the wideband audio data are input, and a band division width and a division number are set based on the signal band information. And a coding system control unit that determines the respective quantization step sizes in the respective sub-sampling units based on the noise floor information and controls the overall operation. A sub-band division filter for dividing into bands, a sub-sampling unit for sub-sampling a plurality of outputs of the sub-band division filter and setting the quantization step size, and multiplexing output data and control information of each sub-band, respectively. And a multiplexer for outputting encoded data. , The noise floor information, the recording studio background noise and the sound pickup Mai
Information about the noise characteristics of the
Only, or when connecting the digital audio output directly
Is the sampling frequency of the digital audio output and
And information on word length or mixing console
When using a tool or various effectors,
A lossless coding apparatus characterized by including information on noise characteristics. 6. A code system control unit each subband, the claims for determining the quantization step size of the sub-band to the noise floor determined from information entered a predetermined dB value as low coding quantization noise 4 or
6. The lossless encoding device according to 5 . 7. If the quantization noise of quantization noise and coding of sound pickup is a potentially interfere, of claims 1 to 6 comprising a code dither generator for generating a code dithering to reduce Sprinkle the interference The lossless encoding device according to any one of the above. Lossless coding apparatus according to any one of claims 8 to claims 1 to assign associate the base band and sub-band to each channel of a conventional multi-channel, respectively 7. 9. Enter the output subbands full band or sub-band performs a linear predictive coding, there is no claim 1 comprising a linear predictive encoder for reducing a peak spectrum of the low frequency
9. The lossless encoding device according to any one of 8 . 10. The lossless encoding device according to claim 9 , wherein an ID indicating that a signal has been input via a linear prediction encoder in a subband of the entire band or the partial band is added. 11. A band division width and the number of divisions are determined based on information indicating a signal band of wideband audio data, an input signal is divided into a predetermined number of subbands, and sub-sampling is performed. Alternatively, a signal peak level for a predetermined time is detected in some of the sub-bands, a scale factor is determined, the signal is converted into signal data excluding a higher-order redundant run length, and the plurality of scale factors and the plurality of redundant runs are converted. A lossless recording medium in which code data obtained by multiplexing output data of a length compression unit is recorded. 12. A signal band for wideband audio data.
Broadband audio data is divided into bands using the information
And scale at the peak level of each subband signal.
And each sub-band data with redundant run length compressed
Signal information and additional information such as scale factors are multiplexed
Loss data that inputs encoded data and decodes it.
A demultiplexer for reproducing a plurality of sub-band data, signal band information, and additional information such as a scale factor from the encoded data, and a scale factor for a predetermined time section.
The redundant run length for each subband based on
An original redundant run-length restoring unit, an oversampling unit that determines an oversampling frequency according to signal band information and performs oversampling, and a subband synthesis filter that synthesizes a plurality of subband data and outputs wideband signal data. Lossless decoding device provided. 13. An operation of a redundant run-length restoring unit and an oversampling unit of a sub-band which is not required based on information of the highest signal band or sampling frequency of a reproduction system connected after an output terminal and based on the reproduction signal band information. The lossless decoding device according to claim 12 , further comprising a decoding system control unit that controls not to restore an unnecessary subband by suspending or the like. 14. The specification information of a noise floor of a reproduction system is input, and a quantization step size in a redundant run-length restoring section and / or an oversampling section is determined based on the noise floor specification information. claim If there is a lower bit surplus comprises a decoding system controller for controlling the whole to perform rounding processing 1
3. The lossless decoding device according to 2. 15. A demultiplexer for reproducing a plurality of sub-band data and noise floor information from input encoded data, a redundant run-length restoring unit for recovering data based on the noise floor information for each sub-band, An oversampling unit for oversampling the subband data, a subband synthesis filter for synthesizing a plurality of subband data to output wideband signal data, and a reproduction system connected after the output end.
Input high signal band or sampling frequency information.
Of unnecessary subbands based on reproduction signal bandwidth information
The operation of the run-length restoration unit and oversampling unit
Avoid restoring unnecessary subbands by pausing
A lossless decoding device including a decoding system control unit for controlling . 16. A demultiplexer for reproducing a plurality of sub-band data and noise floor information from input encoded data, a redundant run-length restoring unit for recovering data based on the noise floor information for each sub-band, An oversampling unit for oversampling the subband data, a subband synthesis filter for synthesizing a plurality of subband data to output wideband signal data, and specification information of a noise floor of a reproduction system
And enter the redundant run length based on the noise floor specification information.
Amount in the gust restoration section and / or oversampling section
Determine the child step size and increase the word length of the input data
If there is a surplus of lower bits, perform rounding
A lossless decoding device including a decoding system control unit that controls the whole . 17. A demultiplexer for reproducing a plurality of subband data and additional information such as a scale factor and a time code from input coded data, and a redundant run for each subband based on a scale factor of a predetermined time section. A redundant run-length restoration unit for restoring the length, an oversampling unit for determining an oversampling frequency according to the signal band information and oversampling, and a subband synthesis filter for synthesizing a plurality of subband data and outputting wideband signal data And connect after the output end
Maximum signal bandwidth or sampling frequency of the playback system
Information is input and it is not necessary based on this playback signal bandwidth information
Redundant sub-band run-length restoration unit and oversampling
Unnecessary sub-bands such as pausing the operation of the ring
A lossless decoding device including a decoding system control unit that controls not to perform restoration . 18. A demultiplexer for reproducing a plurality of subband data and additional information such as a scale factor and a time code from input encoded data, and a redundant run for each subband based on a scale factor of a predetermined time section. A redundant run-length restoration unit for restoring the length, an oversampling unit for determining an oversampling frequency according to the signal band information and oversampling, and a subband synthesis filter for synthesizing a plurality of subband data and outputting wideband signal data And the noise of the playback system
Enter the floor specification information and enter this noise floor specification information.
Redundant run-length restoration unit and / or oversampler
Determine the quantization step size in the
Rounding if the word length of the data is large and there is a surplus of lower bits
A lossless decoding device including a decoding system control unit that controls the whole to perform processing . 19. The decoding system control unit according to claim 13, wherein a linear prediction encoding ID in the additional data is recognized, and post-processing for outputting a signal via a linear prediction decoder is performed.
19. The lossless decoding device according to any one of the chairs 18 . 20. The lossless decoding apparatus according to claim 19 , wherein an ID indicating that a signal has been input via a linear predictive encoder in a subband of the entire band or the partial band is added. 21. At the time of encoding, information indicating a signal band such as a sampling frequency of wideband audio data is input by a switch, and a subband division filter for dividing an input signal into a plurality of subbands based on the signal band information. A sub-sampling unit for sub-sampling each of a plurality of outputs of the sub-band division filter, detecting a signal peak level for a predetermined time in each of the sub-bands, determining and outputting the scale factor, and outputting the scale factor A redundant run-length compression unit that outputs signal data excluding a higher-order redundant run length, and a multiplexer that multiplexes the number and scale factor of each of the subbands and the output data of the redundant run-length compression unit. Get the output and at the time of decryption, The connection is changed by a switch, a demultiplexer that reproduces a plurality of subband data and additional information such as a scale factor and a time code from the input encoded data, and a scale factor of a predetermined time section for each subband is provided. A redundant run-length restoring unit for restoring the redundant run-length based on the signal, an over-sampling unit for determining an over-sampling frequency according to the signal band information and over-sampling, and combining a plurality of sub-band data to output wide-band signal data A lossless code decoding device comprising: a sub-band synthesis filter; and configured to extract an output signal from the sub-band synthesis filter as a decoded output.