Summary of the invention
The objective of the invention is to based on existing DRA audio coding gordian technique, but the DRA audio coding frame structure of design layering can be used in some technical fields that need hierarchical coding the DRA technology.
The present invention at first provides a kind of hierarchical audio coding method, and it comprises: A) by multi-channel audio coding algorithm coding input multichannel audio data; B) data that are encoded are divided into a plurality of ancillary chunks and a plurality of frequency spectrum blocks; And the spectral line sequence number value size according to contained data correspondence in a plurality of frequency spectrum blocks is arranged each frequency spectrum blocks.Preferably, the multi-channel audio coding algorithm is a DRA multi-channel audio coding algorithm.
Further, described method also comprises: C) part relevant with frequency spectrum in the data of a plurality of frequency spectrum blocks is aligned to respectively in each the spectral coefficient part in a plurality of frequency spectrum blocks; And D) will be aligned to each remainder in a plurality of frequency spectrum blocks respectively with the irrelevant part of frequency spectrum in the data of a plurality of frequency spectrum blocks, just in the side information part.
Further, in each of a plurality of frequency spectrum blocks, E) each spectral coefficient part is all at first according to the less sound channel of big minispread numbering of the corresponding spectral line sequence number value of institute, and then according to the big minispread of corresponding spectral line sequence number value number bigger sound channel.
Further, in each of a plurality of frequency spectrum blocks, E) each spectral coefficient part all at first has the coded data of less spectral line sequence number value according to the big minispread of sound channel numbering, and then has the coded data of big spectral line sequence number value according to the big minispread of sound channel numbering.
Further, according to encoding characteristics, select above-mentioned E respectively) and F) one of dual mode each spectral coefficient is partly arranged.
Preferably, encoding characteristics comprise only use E) or only use F) carry out all spectral coefficients and partly arrange the total code efficiency that is obtained.
Further, the length that corresponding fixed bit rate ground is divided and divided frequency spectrum blocks based on the length of corresponding DRA code book section be divided, be similar to the length of each frequency spectrum blocks can in accurate corresponding fixed bit rate ground.
Further, can by preceding and after each frequency spectrum blocks recover current frequency spectrum piece.Preferably, frequency spectrum blocks recover to comprise use preceding and after the coefficient of each frequency spectrum blocks carry out linear interpolation.
Further, the computing method of the layering number of frequency spectrum blocks are: the layering number of frequency spectrum blocks=(the pairing frame length-frame head of code check-side information)/minute layer bandwidth.
Further, the border of frequency spectrum blocks all replaces DRA code book section boundary.
Further, when the distance of the border of frequency spectrum blocks and DRA code book section boundary during, adopt DRA code book section boundary to replace the border of frequency spectrum blocks less than certain predetermined threshold; When the distance of the border of frequency spectrum blocks and DRA code book section boundary during, adopt the border of frequency spectrum blocks to replace DRA code book section boundary greater than certain predetermined threshold.
The present invention also provides a kind of hierarchical audio coding system, comprises G) audio coder, scrambler uses the multi-channel audio coding algorithm; H) dispenser comprises and cuts apart module, and the data that are encoded are divided into a plurality of ancillary chunks and a plurality of frequency spectrum blocks; Each frequency spectrum blocks is arranged according to the spectral line sequence number value size of contained data correspondence in a plurality of frequency spectrum blocks by this system.
Preferably, the multi-channel audio coding algorithm is a DRA multi-channel audio coding algorithm.
Further, dispenser also comprises the arrangement module, and it is used for I) part that the data of a plurality of frequency spectrum blocks are relevant with frequency spectrum is aligned in each the spectral coefficient part in a plurality of frequency spectrum blocks; And J) will be aligned to each side information part in a plurality of frequency spectrum blocks with the irrelevant part of frequency spectrum in the data of a plurality of frequency spectrum blocks.
Further, in each of a plurality of frequency spectrum blocks, E) each spectral coefficient part is all at first according to the less sound channel of big minispread numbering of the corresponding spectral line sequence number value of institute, and then according to the big minispread of corresponding spectral line sequence number value number bigger sound channel.
Further, in each of a plurality of frequency spectrum blocks, F) each spectral coefficient part all at first has the coded data of less spectral line sequence number value according to the big minispread of sound channel numbering, and then has the coded data of big spectral line sequence number value according to the big minispread of sound channel numbering.
Further, according to encoding characteristics, select above-mentioned E respectively) and F) one of dual mode each spectral coefficient is partly arranged.
Preferably, encoding characteristics comprise only use E) or only use F) carry out all spectral coefficients and partly arrange the total code efficiency that is obtained.
Further, the length that corresponding fixed bit rate ground is divided and divided frequency spectrum blocks based on the length of corresponding DRA code book section be divided, be similar to the length of each frequency spectrum blocks can in accurate corresponding fixed bit rate ground.
Further, can by preceding and after each frequency spectrum blocks recover current frequency spectrum piece.Preferably, frequency spectrum blocks recover to comprise use preceding and after the coefficient of each frequency spectrum blocks carry out linear interpolation.
Further, the computing method of the layering number of frequency spectrum blocks are: the layering number of frequency spectrum blocks=(the pairing frame length-frame head of code check-side information)/minute layer bandwidth.
Further, the border of frequency spectrum blocks all replaces DRA code book section boundary.
Further, when the distance of the border of frequency spectrum blocks and DRA code book section boundary during, adopt DRA code book section boundary to replace the border of frequency spectrum blocks less than certain predetermined threshold; When the distance of the border of frequency spectrum blocks and DRA code book section boundary during greater than certain predetermined threshold, the border of sampling frequency spectrum blocks replaces DRA code book section boundary.
The present invention also provides a kind of multi-sound channel digital audio frame structure of hierarchical coding, comprising: a plurality of ancillary chunks and a plurality of frequency spectrum blocks; And the spectral line sequence number value size according to contained data correspondence in a plurality of frequency spectrum blocks is arranged each frequency spectrum blocks.
What preferably, frame structure was preserved is through the data behind the DRA multi-channel audio coding algorithm coding.
Further, each of a plurality of frequency spectrum blocks comprises respectively: arranged in the data of a plurality of frequency spectrum blocks the spectral coefficient part with the frequency spectrum relative section; And the message part with the irrelevant part of frequency spectrum has been arranged in the data of a plurality of frequency spectrum blocks on the limit.
Further, in each of a plurality of frequency spectrum blocks, E) each spectral coefficient part is all at first according to the less sound channel of big minispread numbering of the corresponding spectral line sequence number value of institute, and then according to the big minispread of corresponding frequency spectrum number bigger sound channel.
Further, in each of a plurality of frequency spectrum blocks, E) each spectral coefficient part all at first has the coded data of less spectral line sequence number value according to the big minispread of sound channel numbering, and then has the coded data of big spectral line sequence number value according to the big minispread of sound channel numbering.
Further, according to encoding characteristics, select above-mentioned E respectively) and F) one of dual mode each spectral coefficient is partly arranged.
Preferably, encoding characteristics comprise only use E) or only use F) carry out all spectral coefficients and partly arrange the total code efficiency that is obtained.
Further, the length that corresponding fixed bit rate ground is divided and divided frequency spectrum blocks based on the length of corresponding DRA code book section be divided, be similar to the length of each frequency spectrum blocks can in accurate corresponding fixed bit rate ground.
Further, by preceding and after each frequency spectrum blocks can recover the current frequency spectrum piece.Preferably, frequency spectrum blocks recover to comprise use preceding and after the coefficient of each frequency spectrum blocks carry out linear interpolation.
Further, the computing method of the layering number of frequency spectrum blocks are: the layering number of frequency spectrum blocks=(the pairing frame length-frame head of code check-side information)/minute layer bandwidth.
Further, the border of frequency spectrum blocks all replaces DRA code book section boundary.
Further, when the distance of the border of frequency spectrum blocks and DRA code book section boundary during, adopt DRA code book section boundary to replace the border of frequency spectrum blocks less than certain predetermined threshold; When the distance of the border of frequency spectrum blocks and DRA code book section boundary during, adopt the border of frequency spectrum blocks all to replace DRA code book section boundary greater than certain predetermined threshold.
Based on technique scheme, on the basis of existing DRA coding techniques, handle, but design the DRA audio coding frame structure of layering by the data that are encoded being carried out hierarchy.Description of drawings
Embodiment
By describing the preferred embodiments of the present invention hereinafter by accompanying drawing.Unnecessary details in the following description, function or the structure that becomes prior art will be described in detail, because will cause the ambiguous of introducing of the present invention.
Typical DRA audio coder 100 has been shown in Figure 1A, and it can be realized by hardware, software and/or firmware.In brief, the related technology of DRA standard is exactly with a plurality of technology modules source sound (for example, input PCM sample) to be carried out signal Processing, to reach the almost purpose of lossless compress source sound.Above-mentioned a plurality of technology modules includes but not limited to: transient analysis module 120, multiresolution bank of filters module 122, linear scalar quantization module 130, quantification index coding module 132, code table are selected module 134, human auditory system model module 140, overall Bit Allocation in Discrete module 142 and multiplexing module 150.According to the relevant regulations of DRA standard, above-mentioned technology modules is essential module, and promptly standard compliant DRA output code flow (that is DRA standard code stream) must be through the code stream after the above-mentioned resume module.With it accordingly, typical DRA audio decoder has been shown among Figure 1B, it is used to receive the code stream by after the DRA coder processes, and by carrying out the inverse process of encoding encoding code stream is reduced to the output of PCM sample.
DRA audio frame structure after Fig. 1 C has schematically shown and handled through the DRA audio coder comprises 5 major parts such as information, filling bit, additional information of synchronization character, frame head information, every sound channel.
The DRA technology is a kind of high-quality multi-channel audio coding algorithm, and it can support cbr (constant bit rate) and variable bit rate audio coding, for the DRA technology is laid a good foundation in the enforcement and the popularization of different field.But the DRA technology can't be supported hierarchical coding (that is, scalable coding) at present, and this just makes it can't have the advantage that hierarchical coding possesses.
In order to address the above problem, the invention provides audio frequency hierarchical coding frame structure based on the DRA coding techniques.
In accompanying drawing and explanation subsequently, frame head information (is seen Fig. 1 C, is used to store and transmit the used part supplementary of decoding; In order more clearly to represent the effect of this part, hereinafter also be referred to as " frame head side information ") comprise window sequence information, the frame head type information in the former DRA frame and differ from information such as stereo and intensity stereo master switch; And (hereinafter be referred to as the frequency spectrum blocks side information at the coding side information of each frequency spectrum blocks, see Fig. 7 A and 7B) in, then comprised the information such as spectral coefficient quantization step when Huffman code book subscript and code book range of application, spectral coefficient quantize subscript, spectral coefficient quantization step and the stereo switch of difference and carry out intensity stereo.But those skilled in the art can be understood that after reading this instructions: the particular content that is comprised in frame head side information and the frequency spectrum blocks side information is not limited by the present invention.For example, can comprise and differ from stereo equally in the frequency spectrum blocks side information and information such as intensity stereo master switch.
Fig. 2 A shows hierarchical audio coding method 10A in accordance with a preferred embodiment of the present invention.As shown in the figure, method 10A starts from step 11; Subsequently, in step 12, receive multichannel audio code stream from external device (ED).Next, in step 13, capable the cutting apart of code stream to receiving, they are divided into a plurality of ancillary chunks and a plurality of frequency spectrum blocks.Then, in step 14A, each frequency spectrum blocks in a plurality of frequency spectrum blocks is divided into side information part and spectral coefficient, and partly (example as shown in Figure 7A, frequency spectrum blocks 0 all is divided into the entropy coding data of coding side information and the frequency spectrum blocks X of two parts: frequency spectrum blocks X to each of frequency spectrum blocks N-1, X value from 0 to N-1), the storage part relevant in the spectral coefficient part with spectral coefficient.Next, in step 15A, the data in each spectral coefficient part are all at first numbered less sound channel according to the big minispread of the corresponding spectral line sequence number value of institute.Then, in step 16A, again according to the bigger sound channel of big minispread numbering of corresponding spectral line sequence number value.At last, shown in step 17, the side information part of the data after the arrangement of step 16A generation, step 14A generation and a plurality of ancillary chunks that step 13 produces are packed together, form the DRA frame (shown in Fig. 7 A, hereinafter also will describe in detail) of hierarchical coding.Subsequently, finish cataloged procedure in step 18.Preferably, be the audio code stream of handling through the DRA encryption algorithm at the multichannel audio code stream that step 12 received.
Fig. 2 B shows hierarchical audio coding method 10B according to a further advantageous embodiment of the invention.(in conjunction with Fig. 2 A) as shown in the figure, the step 11-13 among step 11-13 among the method 10B and the method 10A is identical, does not repeat them here.The main difference point of method 10B and 10A is: when having multi-channel audio signal, reach the purpose of hierarchical coding by at first the frequency domain spectra coefficient of each sound channel being interweaved.Specifically, at first in step 14B, each frequency spectrum blocks in a plurality of frequency spectrum blocks is divided into side information part and spectral coefficient part (see Fig. 7 B for details, hereinafter also will describe in detail); Then, in step 15B, each spectral coefficient part all at first has the coded data of less spectral line sequence number value according to the big minispread of sound channel numbering; Next, in step 16B, the big minispread according to the sound channel numbering has the coded data of big spectral line sequence number value again; At last, shown in step 17, the side information part of the data after the arrangement of step 16B generation, step 14B generation and a plurality of ancillary chunks that step 13 produces are packed together, form the DRA frame (shown in Fig. 7 B, hereinafter also will describe in detail) of hierarchical coding.Subsequently, finish cataloged procedure in step 18.Preferably, be the audio code stream of handling through the DRA encryption algorithm at the multichannel audio code stream that step 12 received.
Fig. 2 C shows the hierarchical audio coding method 10C according to another preferred embodiment of the present invention.(in conjunction with Fig. 2 A, 2B) as shown in the figure, the step among the method 10C among step 11-13 and the method 10A is identical.Next, in step 19, judge the aligning method that the spectral coefficient part in a plurality of frequency spectrum blocks is adopted, concrete determination methods will be described in more detail below.If the judgement symbol position is 0 in step 19, the step 14A-16A among the manner of execution 10A then; Otherwise, the step 14B-16B among the manner of execution 10B.At last, shown in step 17, the side information part of the data after step 16A or the step 16B arrangement, step 14A or step 14B generation and a plurality of ancillary chunks that step 13 produces are packed together, form the DRA frame (shown in Fig. 7 C, hereinafter also will describe in detail) of hierarchical coding.Subsequently, finish cataloged procedure in step 18.Preferably, be the audio code stream of handling through the DRA encryption algorithm at the multichannel audio code stream that step 12 received.
Specifically, in step 19, zone bit determines according to encoding characteristics.Encoding characteristics for example comprises: only using method 10A or only using method 10B carry out all spectral coefficients and partly arrange the total code efficiency that is obtained.The judgement that step 19 is carried out is actual selects one bigger among both exactly from pairing efficient of method 10A and the pairing efficient of method 10B, export judged result then: the sort of coding method of efficiency of selection the higher person correspondence.
In subsequently accompanying drawing and explanation, label n represents the spectral line sequence number value, and BL represents the fixed bit rate (can be distributed by the user) be scheduled to, B
nThe total bitrate of 0-n spectral line of expression, L
0The length of expression frequency spectrum blocks 0.
Figure 3 illustrates length L according to the accurate division frequency spectrum blocks 0 in the embodiment of the invention
0Method 30.Those skilled in the art can analogize the division methods that obtains other frequency spectrum blocks by reading this instructions.Method 30 starts from step 31, and spectral line sequence number value initial value is set to 0.Next, reference method 10A partly arranges the spectral coefficient in the frequency spectrum blocks 0 in step 32.Next in step 33, compare B
nSize with fixed bit rate BL.If B
nLess than BL, then make n add 1, proceed step 32-33; Otherwise, enter determining step 34: work as B
nWhen equaling BL, forward step 35 to, make L
0Value n also exports L
0Otherwise, forward step 36 to, make L
0Value n-1 also exports L
0
According to another embodiment of the invention, also may partly arrange the spectral coefficient in the frequency spectrum blocks, and then judge and output L according to step subsequently according to method 10B or 10C in step 32
0, above-mentioned distortion to step 32 is apparently to those skilled in the art, does not repeat them here.
According to method 30, as the L that selects
0Make the gross bit rate B of each frequency spectrum blocks
nWhen being lower than fixed bit rate BL, at L
0Filling bit ' 0 ' after the pairing data of individual spectral line is till gross bit rate reaches fixed bit rate BL, to guarantee accurately to mate the predetermined bit rate.
Fig. 4 shows the length L according to the approximate division frequency spectrum blocks 0 of the embodiment of the invention
0Method 40.Hereinafter will introduce the approximate division methods of frequency spectrum blocks 0 in detail, those skilled in the art can analogize the division methods that obtains other frequency spectrum blocks by reading this instructions.This method starts from step 41, and spectral line sequence number value initial value is set to 0.Next in step 42,10A partly arranges spectral coefficient according to method.Next, in step 43, compare B
nSize with fixed bit rate BL.If B
nGreater than BL, then enter step 44 (describing subsequently); Otherwise, make n add 1, proceed step 42-43.In step 44, judge B
N-1The absolute value of-BL and B
nMagnitude relationship between the absolute value of-BL, and judge following manner of execution 40 according to this: if B
N-1The absolute value of-BL is more than or equal to B
nThe absolute value of-BL then enters step 46, makes L
0=n also exports L
0Otherwise, enter step 45, make L
0=n-1 also exports L
0
According to another embodiment of the invention, also may partly arrange the spectral coefficient in the frequency spectrum blocks, and then judge and output L according to step subsequently according to method 10B or 10C in step 42
0, above-mentioned distortion to step 42 is apparently to those skilled in the art, does not repeat them here.
According to method 40, the L that chooses
0Guarantee the gross bit rate B of frequency spectrum blocks 0
0With fixed bit rate BL the most near (but may be greater than or less than BL).Work as L
0Pairing bit rate B
0When exceeding fixed bit rate BL, do not give up last spectral line; Work as L
0Pairing bit rate B
0When being lower than fixed bit rate BL, at also filling bit ' 0 ' not thereafter.Choose L with said method
0The result can only be approximate match predetermined bit rate.
According to yet another embodiment of the invention, the invention also discloses by preceding and after the method for each frequency spectrum blocks current frequency spectrum blocks of recovering to lose.Specifically, be used in preceding and after the coefficient of each frequency spectrum blocks carry out linear interpolation and recover current frequency spectrum blocks.
According to another embodiment of the invention, the invention also discloses the computing formula of dividing the frequency spectrum blocks number: the layering number of frequency spectrum blocks=(the pairing frame length-frame head of code check-side information)/minute layer bandwidth.Particularly, the code check of supposing this frame is 128kbps, and the shared code check of frame head and side information is 16kbps, and the branch layer bandwidth is 4kbps, according to computing formula, can draw the layering number of frequency spectrum blocks=(128kbps-16kbps)/4kbps=28.
In Fig. 5 A-5C, show method according to yet another embodiment of the invention, that confirm frequency spectrum blocks border and DRA code book segment boundary.The length L of frequency spectrum blocks n
nExpression.In an example of the present invention, choosing the frequency spectrum blocks number is 3, also can be taken as other numerical value certainly.As shown in Figure 5, frequency spectrum blocks is respectively LB
0, LB
1, LB
2, the border is respectively a, b, c, d.DRA code book section is respectively R
0, R
1, R
2, the border is respectively a ', b ', c ', d '.
In Fig. 5 A, show the method that replaces DRA code book segment boundary with the border of frequency spectrum blocks.Promptly when determining frequency spectrum blocks LB
0, LB
1, LB
2Border a, b, c, during d, the code book section is reclassified as R
0, R
1, R
2, border a ', b ', c ', d ' and frequency spectrum blocks LB
0, LB
1, LB
2Border a, b, c, d are corresponding fully.That is, the border of frequency spectrum blocks has replaced DRA code book section boundary correspondingly.
In Fig. 5 B, show the another method of determining frequency spectrum blocks border and DRA code book segment boundary.Particularly, be example: at first be to determine a threshold value TI, judge the distance between frequency spectrum blocks border b and the DRA code book segment boundary b ' with border b.In Fig. 5 B, the distance of frequency spectrum blocks border b and DRA code book segment boundary b ' is judged and is adopted DRA code book segment boundary b ' to replace frequency spectrum blocks border b less than predetermined threshold TI.That is, the border of frequency spectrum blocks is reclassified as b '.
In Fig. 5 C, show a method again of determining frequency spectrum blocks border and DRA code book segment boundary.Shown in Fig. 5 C, dotted line is original DRA code book section R
0, the border is b '.Under the situation of Fig. 5 C, the distance of frequency spectrum blocks border b and DRA code book segment boundary b ' is during greater than certain predetermined threshold TI, and this moment is opposite with method among Fig. 5 B, adopts the border b of frequency spectrum blocks to replace DRA code book section boundary b '.That is, reaffirm DRA code book segment boundary b ', adopt the border b of frequency spectrum blocks to replace DRA code book section boundary b '.
In Fig. 6, show hierarchical audio coding system 60 according to the embodiment of the invention.As shown in the figure, system 60 comprises audio coder 62 and dispenser 63.Wherein audio coder 62 receives voice datas 61, and for example adopt DRA multi-channel audio coding algorithm to voice data 61 in addition audio coding handle.Dispenser 63 receives the data after audio coder 62 is handled, and with the packing data of exporting 64.Dispenser 63 comprises cuts apart the module (not shown), and the data that are used for being encoded are divided into a plurality of ancillary chunks and a plurality of frequency spectrum blocks, and arranges each frequency spectrum blocks with the spectral line sequence number value size of contained data correspondence in a plurality of frequency spectrum blocks.
In addition, also can further comprise in the dispenser 63 and arrange the module (not shown), be used for the part that the data of a plurality of frequency spectrum blocks are relevant with frequency spectrum and be aligned in each the spectral coefficient part in a plurality of frequency spectrum blocks; And will be aligned to each side information part in a plurality of frequency spectrum blocks with the irrelevant part of frequency spectrum in the data of a plurality of frequency spectrum blocks.
Arrangement module in the dispenser 63 can adopt first kind of mode to come each spectral coefficient is partly arranged: to each the spectral coefficient part in each frequency spectrum blocks, at first according to the less sound channel of big minispread numbering of corresponding spectral line sequence number value, and then number bigger sound channel according to the big minispread of the corresponding spectral line sequence number value of institute.
Arrangement module in the dispenser 63 can also adopt the second way to come each spectral coefficient is partly arranged: to each the spectral coefficient part in each frequency spectrum blocks, at first the big minispread according to the sound channel numbering has the coded data of less spectral line sequence number value, and then has the coded data of big spectral line sequence number value according to the big minispread of sound channel numbering.
Can further include judge module in the arrangement module in the dispenser 63, judge respectively and select one of above-mentioned dual mode to come each spectral coefficient is partly arranged according to encoding characteristics.Wherein, encoding characteristics comprises and only uses first kind of mode or only use the second way to come each spectral coefficient is partly arranged the total code efficiency that is obtained; Judge module is used for selecting the mode that total higher the sort of spectral coefficient of code efficiency is partly arranged in the above-mentioned dual mode.
The length that corresponding fixed bit rate is divided and divided frequency spectrum blocks based on the length of corresponding DRA code book section be divided and be similar to the length of cutting apart the Module Division frequency spectrum blocks in the dispenser 63 can by accurate corresponding fixed bit rate.Wherein, the method that accurate corresponding fixed bit rate is divided and approximate corresponding fixed bit rate is divided is introduced in the method 40 in conjunction with the method 30 of Fig. 3 and Fig. 4 respectively in front in detail, does not repeat them here.And divide the length and the method for frequency spectrum blocks based on the length of corresponding DRA code book section, be apparently to those skilled in the art.
Cut apart frequency spectrum blocks after the Module Division in the dispenser 63, can by preceding and after each frequency spectrum blocks recover the current frequency spectrum blocks of losing.Specifically, be used in preceding and after the coefficient of each frequency spectrum blocks carry out linear interpolation and recover current frequency spectrum blocks.
According to another embodiment of the invention, the invention also discloses the computing formula of dividing the frequency spectrum blocks number: the layering number of frequency spectrum blocks=(the pairing frame length-frame head of code check-side information)/minute layer bandwidth.Particularly, the code check of supposing this frame is 128kbps, and the shared code check of frame head and side information is 16kbps, and the branch layer bandwidth is 4kbps, according to computing formula, can draw the layering number of frequency spectrum blocks=(128kbps-16kbps)/4kbps=28.
In the dispenser 62 cut apart the Module Division frequency spectrum blocks time, according to a preferred embodiment of the present invention,, dual mode is arranged to determining of frequency spectrum blocks border.A kind of is all to replace DRA code book section boundary with the frequency spectrum blocks border.That is, wherein DRA code book segment boundary is corresponding one by one with the frequency spectrum blocks border.
Realize by certain predetermined threshold is set on another kind of setting means DRA code book segment boundary and frequency spectrum blocks border.When the distance of frequency spectrum blocks border and DRA code book segment boundary during, adopt DRA code book segment boundary to replace the frequency spectrum blocks border less than certain predetermined threshold; When the distance of frequency spectrum blocks border and DRA code book segment boundary during, adopt the frequency spectrum blocks border to replace DRA code book segment boundary greater than certain predetermined threshold.Detailed is described in the front in conjunction with introducing among Fig. 5 A, Fig. 5 B and Fig. 5 C, at this not in burden.
In Fig. 7 A, 7B and Fig. 7 C, describe the audio frame structure that hierarchy encoding method obtained in detail according to the embodiment of the invention; Wherein, S[block] three-dimensional coordinate among [ch] [bin] represents piece number, Taoist monastic name and spectral coefficient successively.
Shown in Fig. 7 A, Fig. 7 B and Fig. 7 C, the multi-sound channel digital audio frame structure comprises a plurality of ancillary chunks and a plurality of frequency spectrum blocks.Particularly, a plurality of ancillary chunks comprise synchronization character, frame head information and data such as general side information, filling bit and additional data.Further, arrange each frequency spectrum blocks according to the spectral line sequence number value size of contained data correspondence in a plurality of frequency spectrum blocks.Particularly, frequency spectrum blocks is divided into the N piece, is labeled as frequency spectrum blocks 0, frequency spectrum blocks 1... frequency spectrum blocks N-1 respectively.Further, each frequency spectrum blocks in a plurality of frequency spectrum blocks is divided into spectral coefficient part and side information part, wherein, spectral coefficient has partly been arranged the part relevant with frequency spectrum in the data behind the coding, and side information has partly been arranged the part that has nothing to do with frequency spectrum in the data behind the coding.
Shown in Fig. 7 A, to the big minispread numbering less sound channel of the part of each spectral coefficient in each frequency spectrum blocks according to the corresponding spectral line sequence number value of institute, and then according to the big minispread of corresponding spectral line sequence number value number bigger sound channel.
Particularly, the spectral coefficient of frequency spectrum blocks 0 is arranged as S[0] [0] [0]~S[0] [0] [L
0-1], S[0] [1] [0]~S[0] [1] [L
0-1] ... .S[0] [nch-1] [0]~S[0] [nch-1] [L
0-1].
The spectral coefficient of frequency spectrum blocks 1 is S[1] [0] [L
0]~S[1] [0] [L
0+ L
1-1], S[1] [1] [L
0]~S[1] [1] [L
0+ L
1-1] ... .S[1] [nch-1] [L
0]~S[1] [nch-1] [L
0+ L
1-1] spectral coefficient of frequency spectrum blocks N-1 is
As shown in Fig. 7 B, each spectral coefficient part in each frequency spectrum blocks is all at first had the coded data of less spectral line sequence number value according to the big minispread of sound channel numbering, and then have the coded data of big spectral line sequence number value according to the big minispread of sound channel numbering.Particularly, the spectral coefficient of frequency spectrum blocks 0 is arranged as S[0] [0] [0]~S[0] [nch-1] [0], S[0] [0] [1]~S[0] [nch-1] [1] ... .S[0] [0] [L
0-1]~and S[0] [nch-1] [L
0-1].The spectral coefficient of frequency spectrum blocks 1 is S[1] [0] [L
0]~S[1] [nch-1] [L
0], S[1] [0] [L
0+ 1]~and S[1] [nch-1] [L
0+ 1] ... .S[1] [0] [L
0+ L
1-1]~and S[1] [nch-1] [L
0+ L
1-1].The spectral coefficient of frequency spectrum blocks N-1 is
As shown in Fig. 7 C, in frame head information, increase a bit as the structure mark position.Particularly, according to encoding characteristics, the structure mark position can be made as " 0 " or " 1 ".When the structure mark position was " 0 ", the spectral coefficient of this frame intermediate frequency spectrum piece was partly used the aligning method among Fig. 7 A,, used the audio frame structure among Fig. 7 A that is; When the structure position was " 1 ", the spectral coefficient of this frame intermediate frequency spectrum piece was partly used the aligning method among Fig. 7 B,, used the audio frame structure among Fig. 7 B that is.
Preferably, encoding characteristics comprises and only uses Fig. 7 A sound intermediate frequency frame structure or only use Fig. 7 B sound intermediate frequency frame structure to carry out all spectral coefficients and partly arrange the total code efficiency that is obtained.
Though described the present invention in conjunction with being considered to most realistic and optimum embodiment at present, but those skilled in the art are to be understood that and the invention is not restricted to the disclosed embodiments, on the contrary, the present invention is intended to cover various modifications and the equivalent construction that comprises within the spirit of claims and the category.Those skilled in the art can be understood that: can various deformation and/or improvement be used the present invention as being shown in specific embodiment ground, and this does not break away from the spirit or scope of the present invention of describing in broad mode.Therefore, to be considered to be descriptive but not determinate to the embodiment of this paper in all fields.