JP3341448B2 - Efficient encoding method for multi-channel audio data - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a high-efficiency encoding method for multi-channel audio data in compression encoding of multi-channel digital signals. Stereo such as disc players,
The present invention relates to a high-efficiency encoding method of multi-channel audio data suitable for use in a so-called multi-sound sound system.

[0002]

2. Description of the Related Art There are various methods for high-efficiency encoding of audio data, audio signals, and the like. For example, a block frequency division method, so-called transform coding (transform coding), and non-block frequency division System, so-called band division coding (sub-band coding: S
BC).

[0003] The above-mentioned transform coding means that audio data in the time domain is divided into blocks for each unit time, and the data on the time axis for each block is converted into data on the frequency axis, that is, orthogonally transformed. This is a method of dividing into frequency bands and encoding each band. The above-mentioned band division coding is a method of dividing audio data in a time domain and the like into a plurality of frequency bands without dividing them into blocks for each unit time, and coding them.

Further, a high-efficiency coding method combining the above-described transform coding and band division coding has been considered. This high-efficiency encoding, for example, after performing band division by the above-mentioned band division encoding of audio data in the time domain, orthogonally transforms the data for each band into data in the frequency domain, and this orthogonal transformation This is a method of encoding data for each area.

Here, as a filter for band division in the above-mentioned band division coding, for example, there is a filter such as QMF. E. FIG. Croohire Di
digital coding of speech in
Subbands Bell Syst. Tech.
J. Vol 55, No. 8 1976. In addition, ICASP SP 83, BOSTON Po
lyphase Quadrature filter
s-A new subband codingtec
hique Joseph H. Rothweiler
Describes an equal bandwidth filter dividing method and its device.

[0006] As the above-mentioned orthogonal transform, for example, input audio data is converted into a predetermined unit time (frame).
, And a fast Fourier transform (FFT), a cosine transform (DCT), an improved discrete cosine transform, a so-called modified DCT transform (hereinafter referred to as MDC)
T: Modified Discrete Trans
say form. ), Etc., to transform the time axis to the frequency axis. About this MDCT, ICASSP 1987 Subband /
Transform Coding Using Fi
liter BankDesigns Based on
Time Doma AliasingCancel
nation J.L. P. Princen A. B. Br
addUniv. of Surrey Royal
Melbourn Inst. of Tech.

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

As the above-mentioned bit allocation method and its device, the following two methods and its device are known. IEEE
E Transactions of Accout
ics, Speech, and Signal Pro
sessing, vol. ASSP-25, No. 4,
August 1977 describes a method and an apparatus for allocating bits based on the size of data for each band. Also, ICASPSP 1980 The
critical band coder-digit
al encode of the perceptua
l requirements of the aud
itory system M. A. Kransner
MIT describes a method and a device for performing fixed bit allocation by obtaining a required signal-to-noise ratio for each band by using auditory masking.

By the way, for example, in the high-efficiency compression coding method of audio data using the above-mentioned band division coding or the like, the audio data is compressed to about 1/5 using characteristics of human hearing. Method (hereinafter referred to as
It is called 1/5 compression method. ) Is already in practical use.

In addition to ordinary audio equipment, for example, in a stereo system such as a movie film projection system, a high-definition television, a video tape recorder, a video disk player, etc. Audio data of a plurality of eight channels or audio signals are being handled. Even in this case, it is desired to perform high-efficiency coding for reducing the bit rate.

For example, as shown in FIG. 7, the movie film has a video recording area 50 in which a projected image is recorded.
1. Perforation section 502L and perforation section 502 for transporting movie film for projection
R, an analog sound track 503L and an analog sound track 503R provided for enabling reproduction of an audio signal even in a conventional device, a digital sound track 504L and a digital sound track 504 for recording multi-channel digital audio data.
R.

Here, the digital sound track 5
For example, the digital audio data of eight channels of the left channel, the left center channel, the center channel, the right center channel, the right channel, the surround left channel, the surround right channel, and the sub woofer channel are recorded in the 04L and the digital sound track 504R. In such a case, high-efficiency coding for reducing the bit rate is required.

That is, for example, an area in which 16-bit linearly quantized 8-channel audio data can be recorded at a sampling frequency of 44.1 kHz as used in a compact disk (hereinafter, referred to as a CD) or the like. It is difficult to secure it on the movie film.
Therefore, compression of the audio data is required.

The above-mentioned 8 recorded on the above-mentioned movie film
Each channel of the channels is, for example, a left speaker 551, a left center speaker 552, and a center arranged on a screen 550 side where an image reproduced from an image recording area of the movie film is projected by a projector as shown in FIG. The speaker 553, the light center speaker 554, the light speaker 555, the subwoofer speaker 556, and the surround left speaker 558 and the surround right speaker 559 which are arranged so as to surround the audience seat 557, respectively.

Here, the center speaker 553 is
It is arranged at the center on the screen 550 side and outputs a reproduced sound based on the audio data of the center channel. For example, it outputs the most important reproduced sound such as a dialogue of an actor.
The subwoofer speaker 556 outputs a reproduced sound based on the audio data of the subwoofer channel. For example, the subwoofer speaker 556 effectively outputs a sound that is perceived as vibration rather than a low-frequency sound such as an explosion sound. It is often used effectively for scenes and the like.

The left speaker 551 and the right speaker 555 are disposed on the left and right of the screen 550, and output a playback sound based on left channel audio data and a playback sound based on right channel audio data, and exhibit a stereo sound effect. I do. The left center speaker 552 is the left speaker 55
1 and the center speaker 553, and the light center speaker 554 is arranged between the center speaker 553 and the light speaker 555. The left center speaker 552 outputs a reproduction sound based on the left center channel audio data, and the right center speaker 554 outputs a reproduction sound based on the right center channel audio data. Play an auxiliary role. Particularly, in a movie theater or the like having a large screen 550 and a large number of people, the localization of the sound image tends to be unstable depending on the position of the seat. However, the addition of the left center speaker 552 and the right center speaker 554 makes the sound image more realistic. It is effective in creating localization.

Further, the surround left speaker 5
58 and the surround right speaker 559 are arranged so as to surround the audience seats and output a playback sound based on the surround left channel audio data and a playback sound based on the surround right channel audio data. It has the effect of giving an impression. Thereby, a more three-dimensional sound image can be created.

Further, since a medium such as a film is liable to be scratched on its surface, if digital data is recorded as it is, data loss is severe and it is not practical. For this reason, the capability of the error correction code becomes very important, and the data compression needs to be performed to such an extent that the data including the correction code can be recorded in the recording area on the film.

Therefore, the １ ／ compression method is applied as a compression method for compressing the digital audio data of eight channels. That is, by performing optimal bit allocation in consideration of the characteristics of human hearing, 16-bit digital audio data with a sampling frequency of 44.1 kHz, which is recorded on a CD or the like, is compressed to about 1/5 while being compressed. , A high-efficiency coding system that achieves the same sound quality as a CD is applied.

The high-efficiency coding apparatus to which the above-mentioned 1/5 compression system is applied receives, for example, 8-channel digital audio data (hereinafter referred to as audio data) as shown in FIG. Input terminals 101-1
09, encoders 111 to 119 for respectively encoding the audio data of the respective channels input via the input terminals 101 to 109, and audio data of the respective channels encoded by the encoders 111 to 119. The multiplexer 120 includes a multiplexer 120 that collects the data without changing the value, and an output terminal 130 that outputs the data compiled by the multiplexer 120. Here, the encoders 111 to 119 are provided corresponding to audio data of each channel input through the input terminals 101 to 109, and perform encoding for each channel.

The audio data of eight channels input via the input terminals 101 to 109 are encoded by the encoders 111 to 109.
According to 119, encoding is performed for each channel using the １ ／ compression method. Then, the encoded audio data of each channel is output to the multiplexer 120.
Respectively. The multiplexer 120 combines the coded audio data from the encoders 111 to 119 into one data without any change, and combines the combined data into an output terminal 1
Output via 30.

[0022]

[Problems to be solved by the invention]

Here, for example, when the correlation between the data of the two channels is strong and similar but similar data is input, such as stereo data, the encoder for each channel encodes the data. The coded data and its encoded parameter information tend to be similar.

However, the 1/5 compression method is an encoding method for a single channel. Since encoding is performed for each channel, it is necessary to use the 1/5 compression method in encoding multi-channel audio data. However, it has been impossible to perform an effective data encoding process using factors such as data dependence between channels, data characteristics of each channel, and format characteristics.

On the other hand, as an encoding process using the correlation between the data between the channels, when the data of the two channels match, the matched data is made common to represent one channel. There is a method (Japanese Patent Application No. 5-267251). However, this method exerts its effect when the correlation of data between multi-channels is strong. On the other hand, when the correlation is weak, the effect is not sufficiently obtained. Couldn't do it.

In view of the above, the present invention can realize high compression in compression encoding of multi-channel digital audio data regardless of the degree of correlation between digital data between multi-channels. It is another object of the present invention to provide a multi-channel audio data high-efficiency encoding method that can use an existing decoding device.

[0027]

SUMMARY OF THE INVENTION The present invention relates to a highly efficient encoding method for encoding digital audio data of a plurality of channels and obtaining parameter information for the encoding, respectively. A common term detecting step of detecting a common term having a strong correlation from each of the encoded audio data and the parameter information, and a common term requantizing step of requantizing at least a part of the detected common term between the channels And characterized in that:

The present invention also relates to a high-efficiency encoding method for encoding digital audio data of a plurality of channels and obtaining parameter information for the encoding, respectively. A common term detecting step of detecting a common term having a strong correlation from each of the encoded audio data and each of the parameter information; and a common term for requantizing at least a part of the common term between the detected sound frames. Term requantization step.

The present invention is also a high-efficiency encoding method for encoding digital audio data of a plurality of channels and obtaining parameter information of the encoding, wherein the digital audio data of each channel is divided into a plurality of frequency bands. Between adjacent units of each divided band, a common term detecting step of detecting a common term having a strong correlation from each of the encoded audio data and each of the parameter information, at least a common term between the detected units. And a common term requantization step of partially requantizing.

Further, the present invention relates to a high-efficiency encoding method for encoding digital audio data of a plurality of channels and obtaining parameter information of the encoding, respectively. A common term detection step of detecting a common term having a strong correlation between adjacent sound frames and between adjacent units of each band obtained by dividing the digital audio data of the channel into a plurality of frequency bands, based on a result of a predetermined evaluation. A first requantization method for requantizing common terms between channels, a second requantization method for requantizing common terms between temporally adjacent sound frames for each channel, A digital audio data of each channel is divided into a plurality of frequency bands, and a common term between adjacent units of each band is requantized. Select requantization method from the requantization method, and having a common term requantization step of re-quantizing at least a part of the common terms.

[0031]

[0032]

[0033]

[0034] Further, the high efficiency encoding method for multi-channel audio data according to the present invention encodes digital audio data of a plurality of channels, respectively.
A highly efficient encoding method for obtaining parameter information of the encoding, wherein one set of the encoded audio data and parameter information are selected as a reference value for a predetermined unit, and another set of the encoded A difference between the audio data and the parameter information and the reference value is obtained, and at least a part of the another set of encoded audio data and the parameter information is determined according to the corresponding difference value. Or re-quantization for adding a predetermined code to the encoded audio data or parameter information, or re-quantization of the difference value.

[0035]

[0036]

[0037]

[0038]

[0039]

According to the present invention, when digital audio data of each channel is encoded by the existing single channel encoding method, if the correlation between the digital audio data of each channel is strong, parameter information and The encoded digital audio data tends to match. Therefore, while suppressing the redundancy by requantizing at least a part of the common term of the coincident parameter information and the encoded digital audio data,
The parameter information or the channel to which the encoded digital audio data is not similar is determined, and a common item between temporally adjacent sound frames for each channel,
Alternatively, the digital audio data of the channel is divided into a plurality of frequency bands, and at least a part of a common term between adjacent units in each band is requantized, thereby suppressing reduction in sound quality due to requantization of the common term. Like that.

That is, according to the present invention, digital audio data of a plurality of channels are respectively encoded, and
In the high-efficiency coding method for obtaining the parameter information of the coding, a common term having a strong correlation is detected from each of the coded audio data and the parameter information between arbitrary channels, and a common term between the detected channels is detected. Requantize at least some of the terms.

Further, according to the present invention, in a highly efficient encoding method for encoding digital audio data of a plurality of channels and obtaining parameter information of the encoding, a sound frame which is temporally adjacent to each channel is provided. A common term having a strong correlation is detected from the encoded audio data and the parameter information, and at least a part of the detected common term between the sound frames is requantized.

According to the present invention, in a highly efficient encoding method for encoding digital audio data of a plurality of channels and obtaining parameter information of the encoding, the digital audio data of each channel is divided into a plurality of frequency bands. Between adjacent units in each band, a common term having a strong correlation is detected from the encoded audio data and the parameter information, and at least a part of the detected common terms between the units is requantized. .

Further, according to the present invention, in a high-efficiency encoding method for encoding digital audio data of a plurality of channels and obtaining parameter information of the encoding, a plurality of channels are temporally adjacent to each other between arbitrary channels. A common term having a strong correlation between sound frames and between adjacent units of each band obtained by dividing digital audio data of a channel into a plurality of frequency bands is detected, and a common term between channels is detected based on a predetermined evaluation result. A second requantization method for requantizing a common term between sound frames temporally adjacent to each other for each channel, and a digital audio data for each channel. In a third requantization method for requantizing a common term between adjacent units of each band divided into a plurality of frequency bands Select Luo requantization method, re-quantizing at least a part of the common terms.

[0044]

[0045]

[0046]

[0047]

[0048]

[0049]

[0050]

An embodiment of the present invention will be described below with reference to the drawings.

The high-efficiency encoding method for multi-channel digital audio data according to the present invention is realized by, for example, a high-efficiency encoding device having a configuration as shown in FIG.
The high-efficiency coding apparatus adopts the 1/5 compression method, and as shown in FIG. 1, input terminals 1a to 1h to which digital audio data of 8 channels are respectively input.
And encoding of audio data of each channel inputted through the input terminals 1a to 1h, and obtaining parameter information of the encoding, at least a part of a common term of the encoded audio and the parameter information. Channel encoder 2 that requantizes the audio data, the multiplexer 3 that combines the requantized audio data of each channel and the parameter information from the common channel encoder 2 into one data, and the multiplexer 3
Output terminal 4 for outputting data combined into one
It is composed of

As shown in FIG. 2, the common channel encoder 2 divides the audio data input through the input terminals 1a to 1h into a plurality of frequency bands, and divides the audio data into time and frequency signals. Dividing means 21a to 21h for decomposing the data into signal components on a two-dimensional area; encoding the audio data of each channel from the dividing means 21a to 21h; obtaining parameter information of the encoding; A requantizer 22 for requantizing data and a common term of the parameter information; a formatter for assembling the parameter information and the requantized audio data from the requantizer 22 into a bit stream according to a predetermined format 23a ~
23h.

The dividing means 21a to 21h will be described. Here, the dividing units 21a to 21h have the same configuration, and are provided for each channel corresponding to audio data input via the input terminals 1a to 1h.

For example, as shown in FIG. 3, the dividing means 21a has an input terminal 21 to which audio data is input.
1, a band division filter 212 that divides the audio data input via the input terminal 211 into three frequency bands, and an audio data divided into three frequency bands by the band division filter 212.
Performs a CT operation to a two-dimensional area of time and frequency (hereinafter, block floating unit: Block Float)
ing Unit. M) which decomposes into the above signal components
It comprises a DCT operation unit 213 and a unit processing unit 214 for normalizing audio data of each frequency band decomposed into signal components on the block floating unit by the MDCT operation unit 213.

The band division filter 212 converts audio data input via the input terminal 211 from 0 to
It is divided into three frequency bands: a low band of 5.5 kHz, a middle band of 5.5 kHz to 11 kHz, and a high band of 11 kHz to 22 kHz.

The MDCT operation section 213 performs a MDCT operation on the audio data in the low frequency band divided by the band dividing filter 212, and a MDCT circuit 213L for performing an MDCT operation on the audio data in the middle frequency band.
A mid-range MDCT circuit 213M for performing a CT operation, and a high-frequency MCT for performing an MDCT operation on audio data in a high-frequency band.
DCT circuit 213H and low-pass MDCT circuit 213L
And mid-range MDCT circuit 213M and high-range MDCT circuit 213
H to MDCT for audio data of each frequency band
And a block size evaluator 213S for determining a time block length to be operated.

Here, the block size evaluator 213
S shortens the time block length in a portion where audio data of each frequency band changes rapidly, and increases the time block length in a portion where the audio data is steady, and performs MDCT calculation for each frequency band. The time block length is determined, and information indicating the determined time block length is given to the low-band MDCT circuit 213L, the middle-band MDCT circuit 213M, and the high-band MDCT circuit 213H. Thereby, the time resolution is increased in the portion where the audio data changes rapidly, and the effective transmission of the signal component and the quantization noise are controlled in the portion where the audio data is stationary.

The above long time block length is called a long mode, and corresponds to a time of 11.6 ms. Also,
A short time block length is called a short mode, in which the high frequency (11 to 22 kHz) is up to 1.45 ms,
5.5kHz) and mid-range (5.5kHz-11kHz)
In this example, the time resolution is increased to 2.9 ms.

With the above configuration, the MDCT
The arithmetic unit 213 performs a MDCT operation on the audio data divided into the respective frequency bands from the band division filter 212 while controlling a time block length, thereby obtaining a time and frequency block floating unit (hereinafter, referred to as a block floating unit).
A unit is called. ) Decompose into the above signal components.

The unit processing section 214 includes the MDC
A low-frequency normalization circuit 214L that divides the audio data in the low-frequency band decomposed into signal components on the unit by the T operation unit 213 into units and normalizes, that is, normalizes, using a scale factor for each unit; A mid-range normalizing circuit 214M that divides the audio data of the mid-frequency band decomposed into the signal components on the unit into units, and normalizes each of the units using a scale factor; and decomposes the signal components on the unit. A high-frequency normalizing circuit 214H that divides the audio data in the high-frequency band thus obtained into units, and normalizes the audio data using a scale factor for each unit;
4L, a middle-range normalizing circuit 214M, and a high-frequency normalizing circuit 214H. The bit allocator 214S analyzes the components of the audio data supplied from the MDCT calculator 213. ing.

The low-frequency normalization circuit 214L divides the supplied low-frequency band audio data into 20 units, and the middle-frequency normalization circuit 214M converts the supplied middle-frequency band audio data. The high frequency normalization circuit 214H divides the supplied high frequency band audio data into 16 units. That is, each audio data of the low frequency band, the middle frequency band, and the high frequency band is divided into a total of 52 units.

Next, the dividing means 2 constructed as described above
The operation of 1a will be described.

Although not shown, the audio data input through the input terminal 211 is sampled and quantized, and the band division filter 212 is input through the input terminal 211.
Supplied to The band division filter 212 is connected to the input terminal 2
11 is divided into three frequency bands of a low band, a middle band, and a high band, and the audio data divided into the three frequency bands is MD.
It is supplied to the CT operation unit 213.

From the band division filter 212 to the MD
The low frequency band audio data supplied to the CT operation unit 213 is supplied to the low frequency MDCT circuit 213L, and the mid frequency band audio data is supplied to the middle frequency MDCT circuit 213M.
To, high frequency band audio data is high frequency MDCT
In addition to being supplied to the circuit 213H, the audio data of the three frequency bands of the low band, the middle band, and the high band are also supplied to the block size evaluator 213S.

The low-frequency MDCT circuit 213L applies the MDC to the audio data of the low-frequency band from the band division filter 212 while controlling the time block length of the MDCT operation by the block size evaluator 213S.
Perform T operation. The mid-range MDCT circuit 213M adds the MD block to the audio data of the mid-range frequency band from the band division filter 212 while controlling the time block length of the MDCT operation by the block size evaluator 213S.
Perform CT operation.

The high band MDCT circuit 213H adds the MDC to the audio data of the high frequency band from the band division filter 212 while controlling the time block length of the MDCT operation by the block size evaluator 213S.
Perform T operation.

As described above, the low-frequency MDCT circuit 21
3L, mid-range MDCT circuit 213M and high-range MDCT circuit 2
The audio data decomposed into signal components on the unit of time and frequency for each frequency band by 13H is supplied to the unit processing unit 214.

The low frequency band audio data decomposed into the signal components on the unit from the MDCT operation unit 213 is sent to the low frequency normalization circuit 214L, and the audio data in the middle frequency band decomposed into the signal components on the unit is output. The data is supplied to a mid-range normalization circuit 214M, and the audio data in the high frequency band decomposed into signal components on the unit is supplied to a high-range normalization circuit 214H, and the low-, middle-, and high-range audio signals are supplied to the high-frequency normalization circuit 214H. The audio data of the three frequency bands is also supplied to the bit distributor 214S.

The low-frequency normalization circuit 214L includes the MD
The audio data of the low frequency band decomposed into signal components on the unit from the CT operation unit 213 is decomposed into 20 units, and normalized by using a scale factor for each unit. Output. The mid-range normalizing circuit 214M is provided by the MDC
The audio data of the middle frequency band decomposed into the signal components on the unit from the T operation unit 213 is decomposed into 16 units, and normalized by using a scale factor for each unit. Output.

The high-frequency normalization circuit 214H includes the MD
The audio data of the high frequency band decomposed into signal components on the unit from the CT operation unit 213 is decomposed into 16 units, and normalized using a scale factor for each unit. Output. In addition, the bit allocation unit 214S calculates the M
The DCT operation unit 213 analyzes what components each audio data of the low frequency band, the middle frequency band, and the high frequency band decomposed into the signal components on the unit are composed of, and outputs the analysis result. .

At this time, the division units 21b to 21h also perform the same processing on the audio data of the corresponding channel as in the division unit 21a.

Accordingly, the dividing means 21a to 21h output the audio data of the low frequency band, the middle frequency band, and the high frequency band normalized for each unit, and the analysis result of the audio data.

Next, the audio data of the low frequency band, the middle frequency band, and the high frequency band and the analysis result are supplied from the dividing means 21a to 21h, and the requantizer 22 shown in FIG. Give an explanation.

As shown in FIG. 2, the requantizer 22 is connected to the input terminal 2 to which the audio data of the low frequency band is supplied from the dividing means 21a to 21h.
21a to 221h, input terminals 222a to 222h to which the audio data of the middle frequency band from the dividing units 21a to 21h are respectively supplied,
input terminals 223a to 223h to which the audio data of the high frequency band from a to 21h are respectively supplied, input terminals 224a to 224h to which the analysis results from the dividing means 21a to 21h are respectively supplied, The audio data of three frequency bands input via the terminals 221a to 221h, the input terminals 222a to 222h, and the input terminals 223a to 223h are input to the input terminals 224a to 224h.
The primary requantizers 225a to 225 for requantizing the audio data and obtaining the parameter information of the encoding based on the analysis result input via the 224h.
h and the requantization evaluator 226 as a detecting means for detecting a common term from the parameter information from the primary requantizers 225a to 225h and the requantized audio data.
And a secondary requantizer 227 for requantizing at least a part of the common term detected by the requantization evaluator 226.
a to 227h and the secondary requantizers 227a to 227
h and output terminals 228a to 228h for outputting requantized audio data.

The above primary requantizers 225a to 225h
Are based on the analysis results of the components of the audio data input through the input terminals 224a to 224h, and the low-frequency input through the input terminals 221a to 221h, the input terminals 222a to 222h, and the input terminals 223a to 223h. Determine the accuracy with which each unit of the audio data normalized for each unit of the middle band and the high frequency band is coded, that is, determine the parameter information and perform requantization on the audio data. Is what you do. Here, when the parameter information is obtained, bit allocation processing is performed based on the block size mode, the amount of sub-information, and the scale factor to determine the word length, so that the processing can be performed for each channel.

The secondary requantizers 227a to 227h
Although not shown, a first requantization means for requantizing a common term between channels and a second requantizing means for requantizing a common term between temporally adjacent sound frames for each channel. It is composed of quantization means and third requantization means for requantizing a common term between adjacent units in each band obtained by dividing the audio data of the channel into a plurality of frequency bands.

The requantization evaluator 226 detects a common term from the parameter information from the primary requantizers 225a to 225h and each of the requantized audio data.
The means for optimal requantization based on the detection result is determined by the first requantization means, the second requantization means, and the third requantization means of the secondary requantizers 227a to 227h. It is specified by selecting from the following.

Using the requantization means designated by the requantization evaluator 226, the secondary requantizer 227a
227h detect the common term and requantize the common term.

The requantization evaluator 226 will be specifically described.

The requantization evaluator 226 is, for example, as shown in FIG. 4, the primary requantizers 225a to 225
h, inter-channel detection means 51 serving as first common term detection means for detecting common terms between channels for the parameter information and the requantized audio data for the eight-channel audio data supplied from each channel. Inter-sound frame detecting means 52, which is a second common term detecting means for detecting common terms between sound frames temporally adjacent to each other, and between adjacent units in each band obtained by dividing the audio data of the channel into a plurality of frequency bands. Unit-to-unit detecting means 53 which is third common-term detecting means for detecting common terms of
1 and evaluation means 54 for evaluating the detection results of the inter-frame detection means 52 and the inter-unit detection means 53 using a predetermined evaluation function, and means for requantizing based on the evaluation results by the evaluation means 54. Secondary requantizer 22
7a to 227h are constituted by a first requantization means, a second requantization means, and a requantization designation means 55 for selecting and designating among the third requantization means.

The parameter information and the requantized audio data from the primary requantizers 225a to 225h are supplied to the inter-channel detection means 51, the inter-sound frame detection means 52, and the inter-unit detection means 53, respectively. You.

The inter-channel detecting means 51 is provided for
In the parameter information and audio data from the next requantizers 225a to 225h, a common term between channels is detected, and the detection result is supplied to the evaluation means 54. The inter-sound frame detecting means 52 detects, in the parameter information and the audio data from the primary requantizers 225a to 225h, a common term between temporally adjacent sound frames for each channel, and detects the detection result. It is supplied to the evaluation means 54.

The unit-to-unit detecting means 53 uses the parameter information and the audio data from the primary requantizers 225a to 225h to share the common data between adjacent units in each band obtained by dividing the channel audio data into a plurality of frequency bands. The term is detected, and the detection result is supplied to the evaluation means 54.

The evaluation means 54 calculates the common term detection result from the inter-channel detection means 51, the common term detection result from the sound frame detection means 52, and the common term from the inter-unit detection means 53. The detection result is evaluated using a predetermined evaluation function, and the evaluation result is supplied to the requantization specifying means 55.

The requantization designating means 55 includes means for performing requantization based on the evaluation result from the
The secondary requantizers 227a to 227h are selected from the first requantizer, the second requantizer, and the third requantizer, and the result of the selection is selected from the secondary requantizers 227a to 227h. 227
h.

Here, the secondary requantizers 227a to 227a-2
27h is the requantization evaluator 22 having the above-described configuration.
When the common term is detected using the requantization means designated by No. 6 and the common term is requantized, the secondary requantizers 227a to 227h determine whether the common term to be requantized is audio data. , Or even the parameter information is performed by the same requantization method. Thereby, the requantization process can be simplified. Also, the requantization method in the first requantization means, the second requantization means, and the third requantization means is made the same.

This is because the requantization method of the present encoding method is as follows.
This is because this is performed for each piece of recorded data and does not depend on the relationship with the data to be requantized. That is,
In each requantization method, one reference value is determined,
Requantization is performed using the difference from the value.

The quantization method will be described specifically.

For example, as shown in Table 1, when two data values to be requantized are S1 and S2, and S1 is a reference value for comparison, S1 and S2 are the same (S1-S2
= 0) and the difference between S1 and S2 is +1 (S1-
S2 = + 1) or the difference between S1 and S2 is -1 (S1-
If S2 = −1), the amount of information is compressed by quantizing S2 with short bits. That is, the number of bits of data to be recorded at this time is 1 bit or 3 bits.
Bit. Also, even if the difference is the same or +1
Alternatively, if it is not −1, after recording 2 bits for distinction, the original S2 data is recorded. That is, the number of bits of data to be recorded at this time is 2 bits +
This is the number of bits of S2. Since the difference calculation as described above is performed by regarding the values of S1 and S2 as unsigned integers, regardless of how many bits S1 and S2 are, the difference between S1 and S2 is The number of bits may be different. Here, since S1 is a reference value, the difference cannot be quantized.

[0090]

[Table 1]

The first method using the above-described quantization technique
The requantization evaluator 226 selects which requantization means from among the requantization means, the second requantization means, and the third requantization means by the characteristics of the channel or the like. It depends on the usage environment.

For example, when the correlation of audio data between channels is strong, the primary requantizers 225a to 225
In many cases, the data after requantization in h is the same or similar. In this case, S1 shown in Table 1 above is used.
In S2 and S2, for example, when requantization is performed based on the rules in Table 1 above using parameter information of the same unit with different channels or coded audio data of the same frequency spectrum, the data to be recorded is 1 bit, Alternatively, the probability of three bits increases. Therefore, in this case, the requantization evaluator 226 selects and specifies the means for requantizing the common term between channels, which is the first requantization means.

When data has a weak correlation between channels, for example, when there is a tendency that the same sound continues, the correlation between sound frames becomes strong. Therefore, in this case, the requantization evaluator 226 selects and specifies a means for requantizing a common term between adjacent sound frames for each channel, which is a second requantization means.

When there is a tendency such that the voice spectrum is evenly distributed on the frequency axis, the correlation between the units becomes strong. Therefore, in this case, the requantization evaluator 226 The third requantization means is selected and designated as a means for performing common items between adjacent units in each band obtained by dividing audio data of each channel into a plurality of frequency bands.

Here, for example, when re-quantizing a common term between channels, since the above-described re-quantization method always requires a comparison criterion, at least one of the multi-channels used for input needs to be inter-channel. Cannot be requantized. Also, when requantizing common terms between sound frames, the compressed data does not complete the sound frame, so it is not possible to respond to the request for data reproduction from the middle. You need a sound frame that doesn't perform the conversion.

Further, no correlation can be obtained by any of the first requantizing means, the second requantizing means, and the third requantizing means, and the amount of information can be reduced. If not, there is a need for an option not to use this encoding method. In this case, the encoding method is used for a part of the parameter information or a common term of the encoded audio data.

Therefore, the re-quantization evaluator 226 performs a process of selecting a method that can obtain the maximum compression ratio by using a predetermined evaluation function in view of the above.

First-order requantizer 2 configured as described above
25a to 225h, a requantization evaluator 226, and secondary requantizers 227a to 227h.
Will be described.

For example, the primary requantizer 225a receives an analysis result of the audio data inputted from the dividing means 21 via the input terminal 224, and inputs the data from the dividing means 21 via the input terminals 221a to 224a. The parameter information of the audio data normalized for each of the three frequency band units is obtained, the audio data is requantized, and the requantized audio data and the parameter information are requantized and evaluated. Vessels 226 and 2
This is supplied to the next requantizer 227a.

At this time, the primary requantizers 225b to 225
Also in h, the same processing as that of the primary requantizer 225a is performed. Therefore, the primary requantizers 225b to 225b
5h, the parameter information and the requantized audio data are converted by the requantization evaluator 226 and the secondary requantizer 2
27b to 227h.

The requantization evaluator 226 detects common terms from the parameter information from the primary requantizers 225b to 225h and the requantized audio data using three common term detection means. An optimal requantization unit is selected from the three requantization units based on each detection result, and the selection result is supplied to the secondary requantizers 227b to 227h as requantization unit selection information. I do.

The secondary requantizers 227b to 227h
, Based on the selection result from the requantization evaluator 226, the parameter information from the primary requantizers 225a to 225h and the requantized audio data between channels, between sound frames, Alternatively, a common term is detected between units, and the common term is requantized. Then, the secondary requantizers 227b to 227h
Outputs the requantized data to output terminals 228a-22
Output via 8h.

Therefore, as described above, the first, second, and
By selecting the third requantization means using a predetermined evaluation function and performing requantization for each frequency band, it is possible to more effectively compress the amount of information that matches the frequency characteristics of audio data. . That is, this makes it possible to suppress redundant recording.

More specifically, the operation of the common channel encoder 2 shown in FIG. 2 will be described. First, the dividing means 21a to 21h convert the sampled and quantized audio data into three frequency bands. The audio data is decomposed into signal components on a block floating unit of time and frequency, and is normalized using a scale factor for each unit. The dividing means 21a to 21h analyze the audio data, and supply the analysis result and the audio data of three frequency bands normalized for each unit to the requantizer 22.

The requantizer 22 is connected to the dividing means 21
The audio data of the three frequency bands from a to 21h are requantized and parameter information is obtained based on the analysis results from the dividing means 21a to 21h. Further, the requantizer 22 detects a common term from the quantized audio data and the parameter information, selects a unit for requantizing the common term, and selects between channels, between sound frames, or , Re-quantize the common terms between units. Then, the requantizer 22 supplies the parameter information and the requantized spectrum data to the formatters 23a to 23h, respectively.

The formatters 23a to 23h convert the parameter information from the requantizer 22 and the requantized spectrum data into bits for transmission or recording on a recording medium in accordance with a predetermined format. Assemble and output to stream.

Therefore, in the high-efficiency coding apparatus according to the present invention, the common term is requantized in the common channel encoder 2 shown in FIG. Then, the requantized audio data is supplied to the multiplexer 3 and is combined into one data by the multiplexer 3. The combined data is output via the output terminal 4.

Here, the data format output from the output terminal 4 is, for example, as shown in FIG. 5, from the top, 20 bytes of ID data 90 and 212 bytes of data 91-Ch8 of data Ch1-Ch8. 98 and CRC data 99 of an error correction code of 4 bytes.

The encoding processing in the common channel encoder 2 is performed in units of sound frames, and one sound frame is composed of 212 bytes. In this one sound frame, audio reproduction data corresponding to 512 samples and one channel is compression-coded at a sampling grade of 44.1 kHz.

Hereinafter, the data in the sound frame will be described.

The data format in the sound frame is, as shown in FIG.
1, a sub information amount 82, a word length data 83, a scale factor data 84, a spectrum data 85, a redundant scale factor data 86, a redundant word length data 87, a lower sub information amount 88, and a lower block size mode 89. I have.

Here, the redundant scale factor data 86
, Redundant word length data 87, lower sub information amount 88, and lower block size mode 89
Are double writing portions for error correction.

The block size mode 81 is data for recording the evaluation result in the block size evaluator 213S of the MDCT operation unit 213 shown in FIG. For example, as shown in Table 2, in the long mode,
In the low-band and mid-band frequency bands, the low-band MDCT circuit 31 and the mid-band MDCT circuit 3
2 into 128 frequency components, and
In the high frequency band, the high frequency MD of the MDCT
The signal is divided by the CT circuit 33 into 256 frequency components. In the short mode, the low frequency band, the middle frequency band, and the high frequency band correspond to the low frequency MDCT circuit 31 and the medium frequency MDC.
The signal is divided into 32 frequency components by the T circuit 32 and the high frequency MDCT circuit 33, respectively.

[0114]

[Table 2]

The sub information amount 82
Are, as shown in Table 3, the amount 1, the amount 2,
Three pieces of information of the amount 3 are recorded. The amount 1 indicates the number of the recorded word length data 83 and the scale factor data 84, and the amount 2 indicates the double length of the word length data 83.
The amount 3 represents the number of double-written scale factor data 84.

[0116]

[Table 3]

As shown in Table 4, the word length data 83 represents the word length of each unit when requantized.

[0118]

[Table 4]

As shown in Tables 5 and 6, the scale factor data 84 represents a normalized value of each unit.

[0120]

[Table 5]

[0121]

[Table 6]

As described above, in the present invention, in encoding multi-channel audio data, a common term is detected from the encoded audio data and parameter information of the encoding, and at least a part of the common term is detected. Since the re-quantization is performed, a higher compression ratio can be realized than the existing compression coding. Furthermore, the above-mentioned means for requantization is selected using a predetermined evaluation function, and in order to requantize common terms between channels, between sound frames, or between units, the re-quantization is further performed than existing compression coding. High compression ratio can be realized.

Further, since the above-described requantization is performed for each band obtained by dividing the audio data of each channel into a plurality of frequency bands, encoding can be performed in accordance with the frequency characteristics of the audio data.

The data encoded by the above-described high-efficiency encoding apparatus according to the present invention is based on the existing compression encoding scheme in which at least a part of the common information of the parameter information or the encoded data is only requantized. Data. In other words, when reproducing the encoded data, when dividing the data collected at the time of recording for each channel, only the data of one common item is allocated to the corresponding channel, and the decoding by the existing decoding device is performed. Is possible.

[0125]

In other words, according to the present invention, in a highly efficient encoding method for encoding digital audio data of a plurality of channels and obtaining parameter information of the encoding, the above-described encoding is performed between arbitrary channels. By detecting a common term having a strong correlation from each audio data and each of the parameter information and requantizing at least a part of the detected common term between the channels, the correlation between the channels of the digital audio data is used. High compression can be realized. In addition, when reproducing data encoded by the above-described high-efficiency encoding method, an existing decoding device can be used.

Further, according to the present invention, in a high-efficiency encoding method for encoding digital audio data of a plurality of channels and obtaining parameter information of the encoding, a sound frame which is temporally adjacent to each channel may be encoded. By detecting a common term having a strong correlation from each of the encoded audio data and each of the parameter information and requantizing at least a part of the common term between the detected sound frames, a correlation between sound frames is obtained. High compression can be realized using the relationship. Therefore,
High compression can be achieved irrespective of the degree of correlation of digital data between multiple channels. In addition, when reproducing data encoded by the above-described high-efficiency encoding method, an existing decoding device can be used.

According to the present invention, digital audio data of each channel is divided into a plurality of frequency bands in a highly efficient encoding method for encoding digital audio data of a plurality of channels and obtaining parameter information of the encoding. Between adjacent units in each band, a common term having a strong correlation is detected from the encoded audio data and the parameter information, and at least a part of the detected common terms between the units is requantized. Thus, high compression can be realized using the correlation between units. Therefore, high compression can be realized irrespective of the degree of correlation of digital data between multiple channels. In addition, when reproducing data encoded by the above-described high-efficiency encoding method, an existing decoding device can be used.

Further, according to the present invention, in a high-efficiency encoding method for encoding digital audio data of a plurality of channels and obtaining parameter information of the encoding, a plurality of channels are temporally adjacent to each other between arbitrary channels. A common term having a strong correlation between sound frames and between adjacent units of each band obtained by dividing digital audio data of a channel into a plurality of frequency bands is detected, and a common term between channels is detected based on a predetermined evaluation result. A second requantization method for requantizing a common term between sound frames temporally adjacent to each other for each channel, and a digital audio data for each channel. In a third requantization method for requantizing a common term between adjacent units of each band divided into a plurality of frequency bands If the requantization method is selected from the above and the requantization is performed by the first quantization method by requantizing at least a part of the common term, the correlation of the digital audio data between the channels is determined. When the second quantization method is used to perform re-quantization, high compression can be realized using the correlation between digital audio data between sound frames. When the requantization is performed by the third quantization method, high compression can be realized by utilizing the correlation of digital audio data between units. Therefore, high compression can be realized irrespective of the degree of correlation of digital data between multiple channels. In addition, when reproducing data encoded by the above-described high-efficiency encoding method, an existing decoding device can be used.

[0129]

[0130]

[0131]

[0132]

[0133]

[0134]

[Brief description of the drawings]

FIG. 1 is a diagram showing a schematic configuration of a high-efficiency encoding device according to the present invention.

FIG. 2 is a diagram showing a configuration of a common channel encoder shown in FIG.

FIG. 3 is a diagram illustrating a configuration of a dividing unit illustrated in FIG. 2;

FIG. 4 is a diagram showing a configuration of a secondary requantizer shown in FIG. 2;

FIG. 5 is a diagram illustrating a state in which data obtained by encoding 8-channel audio data by a ５ compression method is recorded on a recording medium.

FIG. 6 is a diagram for explaining a state of recording encoded data in a sound frame.

FIG. 7 is a plan view schematically showing the configuration of a movie film.

FIG. 8 is a diagram schematically illustrating a configuration of an 8-channel digital sound system.

FIG. 9 is a diagram showing a configuration for performing a process of collecting 8-channel encoded audio data into one data without changing it.

[Explanation of symbols]

 1a-1h Input terminal 2 Common channel encoder 3 Multiplexer 4 Output terminal 21a-21h Dividing means 22 Requantizer 23a-23h Formatter 51 Inter-channel detecting means 52 Inter-subframe detecting means 53 Inter-unit detecting means 54 Evaluation means 55 Re Quantization designation unit 212 Band division filter 213 MDCT operation unit 213L Low band MDCT circuit 213M Middle band MDCT circuit 213H High band MDCT circuit 213S Block size evaluator 214 Unit processing unit 214L Low band normalization circuit 214M Middle band normalization circuit 214H High Range normalizing circuit 214S Bit allocator 225a-225h Primary requantizer 226 Requantization evaluator 227a-227h Secondary requantizer

Continuation of the front page (56) References JP-A-4-360331 (JP, A) JP-A-5-227039 (JP, A) JP-A-7-123008 (JP, A) JP-A-1-318327 (JP) , A) (58) Field surveyed (Int. Cl. ⁷ , DB name) H03M 7/30 G10L 11/00 G11B 20/10 301

Claims (5)

(57) [Claims] 1. A high-efficiency encoding method for encoding digital audio data of a plurality of channels and obtaining parameter information of the encoding, respectively, comprising the steps of: A common term detecting step of detecting a common term having a strong correlation from each of the parameter information; and a common term requantizing step of requantizing at least a part of the detected common terms between the channels. A highly efficient encoding method for multi-channel audio data. 2. A highly efficient encoding method for encoding digital audio data of a plurality of channels and obtaining parameter information of the encoding, wherein the encoding is performed between sound frames temporally adjacent to each other for each channel. Common term detection step of detecting a common term having a strong correlation from each of the converted audio data and the parameter information, and common term requantization for requantizing at least a part of the common term between the detected sound frames. And a step for efficiently encoding multi-channel audio data. 3. A high-efficiency encoding method for encoding digital audio data of a plurality of channels and obtaining parameter information of the encoding, wherein each of the digital audio data of each channel is divided into a plurality of frequency bands. between the adjacent units, a common term detection step of detecting a strong common term correlation from each audio data and the respective parameter information is the encoded, at least a part of the common section between the detected above unit re A common term requantization step of quantizing the multi-channel audio data. 4. A high-efficiency encoding method for encoding digital audio data of a plurality of channels and obtaining parameter information of the encoding, wherein the sound frames are temporally adjacent to each other between channels. during, and a common term detection step of detecting each strong common term correlation between the adjacent unit of each band obtained by dividing the digital audio data of the channel into a plurality frequency bands, based on a result of a predetermined evaluation, between the channels A first requantization method for requantizing a common term of the first and second channels, a second requantization method for requantizing a common term between temporally adjacent sound frames for each channel, and a digital method for each channel. Third requantization method for requantizing common terms between adjacent units in each band obtained by dividing audio data into a plurality of frequency bands Requantization method is selected, high-efficiency encoding method of the multi-channel audio data; and a common term requantization step of re-quantizing at least a portion of the common section from within. 5. The common-term requantization step selects a part of the common terms detected in the common-term detection step as a reference value, and requantizes each common term other than the reference value. 5. The method according to claim 4, wherein the encoding is performed based on a difference value between the value and the common term.