CN101283403B

CN101283403B - Method and apparatus for signal processing and encoding and decoding method, and apparatus thereof

Info

Publication number: CN101283403B
Application number: CN2006800372934A
Authority: CN
Inventors: 房熙锡; 金东秀; 林宰显; 吴贤午; 郑亮源; 金孝镇
Original assignee: LG Electronics Inc
Current assignee: LG Electronics Inc
Priority date: 2005-10-05
Filing date: 2006-10-09
Publication date: 2013-03-27
Anticipated expiration: 2026-10-09
Also published as: CN101283397A; CN101283254B; CN101283253A; CN101283402A; CN101283397B; CN101283403A; CN101283252A; CN101283254A; CN101283402B; CN101283252B; CN101283398A; CN101283405A; CN101283405B; CN101283398B; CN101283253B

Abstract

An apparatus for processing a signal and method thereof are disclosed. Data coding and entropy coding are performed with interconnection, and grouping is used to enhance coding efficiency. The present invention includes the steps of obtaining a group reference value corresponding to a plurality of data included in one group through first grouping and external grouping for the first grouping and a difference value corresponding to the group reference value and obtaining the data using the group reference value and the difference value.

Description

Method and apparatus and Code And Decode method and device thereof that signal is processed

Technical field

The present invention relates to the devices and methods therefor of processing signals, relate in particular to the devices and methods therefor of decoding data.

Background technology

Generally speaking, many technology for signal compression and recovery have been introduced up to now.And the applicable target of corresponding technology is the various data that comprise voice data, video data etc.In addition, signal compression or recovery technology are towards the direction evolution that improves the audio or video quality with high compression ratio.In addition, doing a lot of work aspect the transfer efficiency that improves the various communication environments of adaptation.

Yet, believe still to have the leeway of improving transfer efficiency.So, need to do a large amount of work and come to make the maximise transmission efficiency of signal in very complicated communication environment by the exploitation of new signal processing scheme.

Summary of the invention

Therefore, the present invention relates to a kind of devices and methods therefor of basically having eliminated the processing signals of one or more limitations owing to relevant technologies and the problem that shortcoming causes.

An object of the present invention is to provide a kind of devices and methods therefor for the treatment of signal, the transfer efficiency that can optimize signal of mat.

Another object of the present invention provides a kind of devices and methods therefor of decoding data, the efficiently decoding data of mat.

Another object of the present invention provides a kind of devices and methods therefor for encoding/decoding data, the control data transfer maximizing efficiency that can be used in Audio recovery of mat.

Another object of the present invention provides a kind of medium that comprises encoded data.

Another object of the present invention provides a kind of data structure, mat can transmit efficiently encoded data.

Another purpose of the present invention provides a kind of system that comprises decoding device.

In order to realize these or other advantage and according to purpose of the present invention, as embody and broadly described, the method for processing signals may further comprise the steps according to one embodiment of present invention: obtain with by comprising that the first marshalling and the second marshalling of organizing into groups are included in the corresponding group reference value of a plurality of data in one group and first difference value corresponding with this group reference value; And utilization group reference value and the first difference value obtain data.

The first marshalling and marshalling comprise at least one that is selected from lower group: inner marshalling, outside marshalling, territory marshalling and data marshalling.The data marshalling can be carried out in the territory group.The time domain that is included in the group of territory comprises one that is selected from lower group: time slot territory, parameter set territory or data set territory.The frequency domain that is included in the group of territory can comprise one that is selected from lower group: sampling territory, subband domain, hybrid domain, parameter band territory, frequency ranges of data territory or sound channel territory.

The group reference value can be one of the first guiding reference value and difference reference value.

The method also comprises at least one step of decoding in group reference value and the first difference value.The first guiding reference value is determined by group.

The method is further comprising the steps of: obtain corresponding to the second guiding reference value of a plurality of the first guiding reference values with corresponding to the second difference value of the second guiding reference value, and utilize the second guiding reference value and the second difference value to obtain the first guiding reference value.

And the second marshalling can be the outside marshalling for the first marshalling.Perhaps, the second marshalling can be the inside marshalling for the first marshalling.

And, at least one that marshalling can be in time domain and frequency domain data are carried out.And marshalling can be at least one territory of the organizing into groups marshalling in time domain and the frequency domain.

Time domain comprises one that is selected from lower group: time slot territory, parameter set territory or data set territory.Frequency domain can comprise one that is selected from lower group: sampling territory, subband domain, hybrid domain, parameter band territory, frequency ranges of data territory or sound channel territory.The marshalling data can or index or parameter.

The method also comprises utilizing by the first marshalling and is included in the step that an indicated entropy table of the index in the group carries out the entropy decoding to the first difference value, and wherein data are utilization group reference values and obtain through the first difference value of entropy coding.

The method also comprises utilizing by the first marshalling and is included in the step that the indicated entropy table of index in the group carries out the entropy decoding to the first difference value and group reference value, and wherein said data are to utilize through the group reference value of entropy decoding with through the first difference value of entropy decoding to obtain.

For realizing these and other advantage and according to purpose of the present invention, a kind of device of processing signals comprise obtain with by comprising that the first marshalling and second is grouped in interior marshalling and is included in the corresponding group reference value of a plurality of data in the group and obtains parts corresponding to the value of the difference value of this group reference value, and utilization group reference value and difference value obtain the data acquisition section of data.

For realizing these and other advantage and according to purpose of the present invention, a kind of method of processing signals may further comprise the steps: utilize with by comprising that the first marshalling and second is grouped in interior marshalling and is included in a plurality of data in the group corresponding group reference value and described data and generates a difference value, and transmit the difference value that generates.

For realizing these and other advantage and according to purpose of the present invention, a kind of device of processing signals comprises: use with by comprising that the marshalling of the first marshalling and second in being grouped in is included in the value generation parts that the corresponding group reference value of a plurality of data in the group and described data generate a difference value, and the output block that transmits the difference value that generates.

Beneficial effect

Therefore, the present invention can realize efficient data decoding and entropy decoding, thus data compression and recovery that realization has high-transmission efficient.

The accompanying drawing summary

Fig. 1 and Fig. 2 are the block diagrams according to system of the present invention;

Fig. 3 and Fig. 4 are the figure that explains according to PBC decoding of the present invention;

Fig. 5 is the figure that explains according to DIFF decoding of the present invention;

Fig. 6 to 8 is the figure that use the example of DIFF decoding scheme;

Fig. 9 is the block diagram of the relation when explaining according to a kind of at least three kinds of decoding schemes of selection of the present invention;

Figure 10 is the block diagram of the relation when explaining according to a kind of at least three kinds of decoding schemes of selection of correlation technique;

Figure 11 and Figure 12 are respectively the process flow diagrams according to data decoding selection scheme of the present invention;

Figure 13 is the figure that explains according to inner marshalling of the present invention;

Figure 14 is the figure that explains according to outside marshalling of the present invention;

Figure 15 is the figure that explains according to the multiple marshalling in inside of the present invention;

Figure 16 and Figure 17 are respectively the diagrams of explaining mixing marshalling according to another embodiment of the invention;

Figure 18 is the exemplary plot according to 1D of the present invention and 2D entropy table;

Figure 19 is the exemplary plot according to two kinds of methods for 2D entropy decoding of the present invention;

Figure 20 is the figure according to the entropy decoding scheme for the PBC decode results of the present invention;

Figure 21 is the figure according to the entropy decoding scheme for the DIFF decode results of the present invention;

Figure 22 is the figure that explains according to the method for the selective entropy table of the present invention;

Figure 23 is the slice map according to data structure of the present invention;

Figure 24 is the block diagram of the device for audio compression and recovery according to an embodiment of the invention;

Figure 25 is the detailed diagram of spatial information addressable part according to an embodiment of the invention; And

Figure 26 is the detailed diagram of spatial information decoding parts according to an embodiment of the invention.

Embodiment

Now will be in detail with reference to preferred embodiment of the present invention, its example is shown in the drawings.

General terms current and that generally use is selected as the term that uses among the present invention.And, apply in addition the term that artificial special circumstances arbitrariness ground is selected, its detailed implication is explained in the description of preferred embodiment of the present invention in detail.Therefore, should not understand the present invention from the literal of term, and should understand with the implication of term.

In the present invention, the implication of " decoding " comprises cataloged procedure and decode procedure.Yet, it should be apparent to those skilled in the art that specific decode procedure is only applicable to cataloged procedure or decode procedure, this will distinguish in following description to corresponding component.And " decoding " also can be called as " encoding and decoding ".

In the present invention, the step of decoded signal should be deciphered to explain by being divided into data decoding and entropy.Yet, between data decoding and entropy decoding, there is correlativity, this will explain after a while in detail.

In the present invention, the whole bag of tricks of data being organized into groups for expeditiously executing data decoding and entropy decipher will be explained.Grouping method has the effective technological thought of independence that has nothing to do with concrete data or entropy decoding scheme.

In the present invention, explanation had the audio coding scheme (for example, " ISO/IEC 23003, MPEG Surround ") of spatial information as the specific example that adopts the decoding of data decoding and entropy.

Fig. 1 and Fig. 2 are the figure according to system of the present invention.Fig. 1 illustrates code device 1, and Fig. 2 illustrates decoding device 2.

With reference to figure 1, code device 1 according to the present invention comprises at least one in data marshalling parts 10, the first data encoding parts 20, the second data encoding parts 31, the 3rd data encoding parts 32, entropy addressable part 40 and the bit stream multiplexing components 50.

Optionally, the second and the 3rd

data encoding parts

31 and 32 can be integrated into a data addressable part 30.For example, carry out variable length code by 40 pairs of entropy addressable parts through the second and the 3rd

data encoding parts

31 and 32 coded datas.Said elements will be explained as follows in detail.

Data marshalling parts 10 are according to specifying unit binding input signal to improve data-handling efficiency.

For example, data marshalling parts 10 are distinguished data according to data type.And the data of distinguishing are encoded by one in the

data encoding parts

20,31 and 32.For the purpose of data-handling efficiency, data marshalling parts 10 are distinguished at least one group with the part in the data.And the data after the marshalling are encoded by one in the

data encoding parts

20,31 and 32.In addition, will after a while with reference to figures 13 to 17 explain in detail comprising data marshalling parts 10 operation according to grouping method of the present invention.

Data encoding parts

20,31 and 32 all come coded input data according to corresponding encoding scheme separately.

Data encoding parts

20,31 and 32 adopt at least one in PCM (pulse-code modulation) scheme and the differential decoding scheme separately.Particularly, for example, the first data encoding parts 20 adopt the PCM scheme, and the second data encoding parts 31 adopt the first differential decoding scheme of utilizing the guiding reference value, and the 3rd data encoding parts 32 adopt the second differential decoding scheme of the difference of utilizing adjacent data.

Hereinafter,, the first differential decoding scheme is called " based on the decoding (PBC) of guiding " for the ease of explaining, and the second differential decoding scheme is called " differential decoding (DIFF) ".And

data encoding parts

20,31 and 32 operation will explained with reference to figure 3 to 8 after a while in detail.

Simultaneously, entropy addressable part 40 is carried out variable length code according to the statistical properties of data with reference to entropy table 41.And, explain in detail after a while the operation of entropy addressable part 40 with reference to Figure 18 to 22.

Then 50 layouts of bit stream multiplexing components and/or the data of conversion through deciphering will be transmitted with the form of bit stream through the data of layout/conversion to make it corresponding to transmission specification.Yet, do not use bit stream multiplexing components 50 if use concrete system of the present invention, to those skilled in the art obvious be this system configuration can be become without bit stream multiplexing components 50.

Simultaneously, decoding device 2 is configured to the code device 1 corresponding to above explanation.

For example, with reference to figure 2, bit stream multichannel Knock-Down Component 60 receives the bit stream of input, and according to initialize format to being included in various information in the bit stream that receives and translating and classifying.

Entropy decoding parts 70 utilize entropy table 71 that data are reverted to entropy coding raw data before.In this case, obviously the entropy table 41 of entropy table 71 and front code device 1 shown in Figure 1 disposes identical.

The first data decode parts 80, the second data decode parts 91 and the 3rd data decode parts 92 are carried out respectively and the first to the 3rd above-mentioned

data encoding parts

20,31 and 32 corresponding decodings.

Particularly, carry out in the situation of differential decoding at the second and the 3rd data decode

parts

91 and 92, overlapping decode procedure can be integrated in a decode procedure and process.

Data reconstruction parts 95 are with the raw data before the data of data decode

parts

80,91 and 92 decodings are recovered or are reconstructed into data encoding.Sometimes, the data through decoding can be resumed into the data that obtain by conversion or modification raw data.

By the way, for the efficient execution of data decoding, the present invention uses together to two kinds of decoding schemes of major general, and is intended to utilize the correlativity between the decoding scheme that efficient decoding scheme is provided.

And the present invention aims to provide the various data marshalling schemes for the efficient execution of data decoding.

In addition, the present invention aims to provide the data structure that comprises feature of the present invention.

When technological thought of the present invention is applied to various system, to those skilled in the art aobvious and easy be when using element illustrated in figures 1 and 2, to use various other configurations.For example, the controller that needs executing data quantification or needs control said process.

[data decoding]

PCM (pulse-code modulation), the PBC (based on the decoding of guiding) and the DIFF (differential decoding) that can be used as data decoding scheme of the present invention will be explained as follows in detail.In addition, subsequently also with high efficiency selected and the correlativity of decryption decoding scheme.

(1.PCM pulse-code modulation)

PCM is a kind of decoding scheme that analog signal conversion is become digital signal.Then PCM quantizes corresponding result with the preset interval sampled analogue signals.PCM is disadvantageous aspect decoding efficiency, but can effectively be used for being not suitable for the data of PBC or DIFF decoding scheme, and this will explain after a while.

In the present invention, PCM uses with PBC or DIFF decoding scheme in executing data decoding, and this will explain with reference to figure 9 to 12 after a while.

(2.PBC based on the decoding of guiding)

2-1.PBC concept

PBC is a kind of concrete benchmark of determining in the data group of distinguishing, and the decoding scheme that data and the relation between the determined benchmark as decoding target are used.

The value that becomes the benchmark of using PBC can be defined as " reference value ", " guiding ", " guiding reference value " or " guiding value ".Hereinafter, for the ease of explaining, be referred to as " guiding reference value ".

And, the difference value between the data in guiding reference value and the group can be defined as " difference " or " guiding difference ".

In addition, used the final group of concrete marshalling scheme by above-mentioned data marshalling parts 10 as the data group indication of the unit that uses PBC.The data marshalling can be carried out by variety of way, and this will explain after a while in detail.

In the present invention, being defined as " parameter " by the data of organizing into groups to have specific meanings with upper type explains.This only is for the ease of explaining, can replacing with different terms.

PBC process according to the present invention comprises at least two following steps.

At first, selection is corresponding to the guiding reference value of a plurality of parameters.In this case, the guiding reference value is to determine with reference to the parameter that becomes the PBC target.

For example, the guiding reference value be set to from the parameter that becomes the PBC target mean value, become target parameter mean value approximate value, corresponding to the intermediate value of the by-level of the parameter that becomes target and become the parameter of target the value of selecting in the value of the most normal use.And the guiding reference value also can be set to preset default value.In addition, guiding value can be definite by selecting in presetting table.

Perhaps, in the present invention, guide reference value temporarily and be set to guiding reference value by at least two kinds of selections in the system of selection of multiple guiding reference value, calculate the decoding efficiency of every kind of situation, so be selected as final guiding reference value corresponding to the interim guiding reference value of the situation with optimal decoding efficient.

When mean value was P, the approximate value of mean value was Ceil[P] or Floor[P].In this case, Ceil[x] be the maximum integer that is no more than " x ", and Floor[x] be the smallest positive integral that surpasses " x ".

Yet, also can select arbitrariness fixing default value and not with reference to the parameter that becomes the PBC target.

Again for example, mentioned as described above, after having selected randomly and multiplely some values that are elected to be guiding, the value that shows optimal decoding efficient can be elected to be optimal guidance.

Secondly, draw difference value between the parameter in selected guiding and the group.For example, difference value can calculate by deduct the guiding reference value from the parameter value that becomes the PBC target.This is explained as follows with reference to Fig. 2 value Fig. 4.

Fig. 3 and Fig. 4 are the figure that explains according to PBC decoding of the present invention.

For example, suppose to exist in the group a plurality of parameters (for example, 10 parameters), have respectively following parameter value, X[n]={ 11,12,9,12,10,8,12,9,10,9}.

If select the PBC scheme to be used for the parameter of this group of coding, then at first should select to guide reference value.In the present example, can see that will guide reference value in Fig. 4 is made as " 10 ".

Mentioned as described above, can select to guide reference value by the method for various selection guiding reference values.

Calculate the difference value of PBC according to formula 1.

[formula 1]

D[n]=x[n]-P, n=0 wherein, 1 ..., 9.

In this case, " P " indication guiding reference value (=10), and x[n] be the target component of data decoding.According to the result of the PBC of formula 1 corresponding to d[n]=1,2 ,-1,2,0 ,-2,2 ,-1,0 ,-1}.That is, the result of PBC decoding comprises selected guiding reference value and the d[n that calculates].And these values will become the target of the entropy decoding that will explain after a while.In addition, PBC is more effective in the generally less situation of the deviation of target component value.

2-2.PBC object

The target of PBC decoding is not defined as one.Can decipher by PBC the numerical data of various signals.For example, it is applicable to the audio coding that will explain after a while.In the present invention, will be explained in detail as the PBC decoding target with the additional control data that voice data is processed.

The control data are transmitted after the transmission down-mix audio signal, then use it for and rebuild this audio frequency.In the following description, will control data definition and be " spatial information or spatial parameter ".

Spatial information comprises various spatial parameters, such as levels of channels poor (hereinafter referred to as CLD), inter-channel coherence (hereinafter referred to as ICC), sound channel predictive coefficient (hereinafter referred to as CPC) etc.

Particularly, CLD is the parameter of energy difference between two different sound channels of indication.For example, CLD have between-15 and+value of scope between 15.ICC is the parameter of correlativity between two different sound channels of indication.For example, ICC has the value of the scope between 0 and 7.And CPC is the parameter that indication is used for generating from two sound channels the predictive coefficient of three sound channels.For example, CPC has the value of the scope between-20 and 30.

As the target of PBC decoding, can comprise a yield value that is used for the conditioning signal gain, for example, ADG (gain of arbitrariness multi-channel audio).

And the ATD (arbitrariness tree data) that is applied to the arbitrariness sound channel transform frame of down-mix audio signal can become the PBC decoding target.Particularly, ADG is the parameter different from CLD, ICC or CPC.That is, ADG is different from the spatial information that extracts such as CLD, ICC CPC etc. corresponding to the parameter of regulating audio gain from the sound signal sound channel.Yet, for using example, can process ADG or ATD to improve the efficient of audio coding by the mode identical with above-mentioned CLD.

As another target of PBC decoding, can consider " partial parameters ".In the present invention, the part of " partial parameters " expression parameter.

For example, suppose special parameter is expressed as the n bit, the n bit is divided at least two parts.And, two parts can be defined as respectively the first and second partial parameters.In the situation of attempting execution PBC decoding, can draw the difference value between first's parameter value and the guiding reference value.Yet the second portion parameter value of getting rid of outside difference value calculates should be as independent value transmission.

More specifically, for example, in the situation by n bit indication parameter value, least significant bit (LSB) (LSB) is defined as the second portion parameter, and will be defined as first's parameter with all the other (n-1) the individual high-order parameter value that makes up.In this case, can only carry out PBC to first's parameter.This is because owing to each improves with the little decoding efficiency that makes of deviation between (n-1) individual high-order first's parameter value that makes up.

The second portion parameter of eliminating outside Difference Calculation transmitted separately, then considers by decoding parts reconstruct final argument the time.Perhaps, also can obtain the second portion parameter rather than transmit individually this second portion parameter by predetermined scheme.

Use the PBC decoding of the characteristic of utilizing partial parameters according to the characteristic limitations ground of target component.

For example, mention that the deviation between each first's parameter should be less in as described above.If this deviation is large, then there is no need to use partial parameters.It in addition may reduce decoding efficiency.

According to experimental result, the CPC parameter of above-mentioned spatial information is suitable for the application of PBC scheme.Yet it is not first-selection that CPC is applied to the rudenss quantization scheme.In the rough situation of quantization scheme, the deviation between first's parameter increases.

In addition, utilize the data decoding of partial parameters applicable to DIFF scheme and PBC scheme.

The partial parameters concept is being applied in the situation of CPC parameter, the signal processing method and the device that are used for rebuilding are explained as follows.

For example, the method of partial parameters processing signals of utilizing according to the present invention may further comprise the steps: utilize corresponding to the reference value of first's parameter with corresponding to the difference value of this reference value and obtain first's parameter, and utilize this first's parameter and second portion parameter to determine a parameter.

In this case, reference value or guiding reference value or difference reference value.And first's parameter comprises the part position of parameter, and the second portion parameter comprises all the other positions of parameter.In addition, the second portion parameter comprises the least significant bit (LSB) of parameter.

This signal processing method also comprises the step of utilizing determined Reconstruction sound signal.

This parameter is at least one the spatial information that comprises among CLD, ICC, CPC and the ADG.

If be not rough if parameter is the quantization scale of CPC and parameter, then can obtain the second portion parameter.

And final argument also doubly takes advantage of this result and second portion parameter phase Calais to determine by this partial parameters doubly being multiply by two.

The device that utilizes the partial parameters processing signals according to the present invention comprises utilizing corresponding to the reference value of first's parameter with corresponding to the difference value of this reference value and obtains first's parameter acquiring parts of first's parameter and utilize first's parameter and the second portion parameter is determined the parameter-determining means of a parameter.

This signal processing apparatus also comprises by receiving the second parameter acquiring parts of second portion parameter acquiring second portion parameter.

And the first parameter acquiring parts, parameter-determining means and second portion parameter acquiring parts are included among above-mentioned data decode

parts

91 or 92.

A kind of signal processing method of partial parameters that utilizes according to the present invention may further comprise the steps: a parameter is divided into first's parameter and second portion parameter, and utilizes reference value and first's parameter corresponding to first's parameter to generate difference value.

And this signal processing method also comprises the step of this difference value of transmission and second portion parameter.

The device that utilizes the partial parameters processing signals according to the present invention comprises that the difference value that a parameter is divided into the parameter partition member of first's parameter and second portion parameter and utilizes reference value and this first's parameter corresponding to first's parameter to generate difference value generates parts.

And this signal processing apparatus also comprises the parameter output block of this difference value of transmission and second portion parameter.

In addition, parameter partition member and difference value generation parts are included within above-mentioned

data encoding parts

31 or 32.

2-3.PBC condition

PBC of the present invention decoding select independent guiding reference value, then selected guiding reference value is included in the bit stream aspect, the transfer efficiency of PBC decoding probably is lower than the DIFF decoding scheme that will explain after a while.

So the present invention aims to provide the optimal conditions of carrying out PBC decoding.

If experimentally becoming the number of the data of data decoding target in a group is 3 or more at least, then PBC decoding is suitable for.This is corresponding to the result who considers data decoding efficient.This means if only have two data in group then DIFF or PCM decoding are more efficient than PBC decoding.

Although PBC decoding is applicable at least three data or more data, preferably PBC decoding is applied to exist in the group situation of at least 5 data.In other words, the situation of the most efficient applicable PBC decoding is to exist at least five to become the data of data decoding target and the little situation of deviation between this at least five data.And the data number that is applicable to carry out the minimum of PBC decoding will be determined according to system and decoding environment.

Provide the data that become the data decoding target for each frequency ranges of data.This will explain by the marshalling process that will describe after a while.So for example, the present invention proposes around using PBC decoding in deciphering at least five data frequency bands need to be arranged at the mpeg audio that will explain after a while.

Hereinafter, utilize the signal processing method of PBC executive condition and device to be explained as follows.

In signal processing method according to an embodiment of the invention, if satisfied the condition that presets if obtained corresponding to the number of data of guiding reference value and the number of frequency ranges of data, then obtain the guiding reference value and corresponding to the guiding difference value of this guiding reference value.Subsequently, utilize this guiding reference value and guiding difference value to obtain data.Particularly, the number utilization of data is obtained comprising the frequency ranges of data number that data are arranged.

In signal processing method according to another embodiment of the invention, utilize the number of data to decide in a plurality of data decoding schemes one, and according to determined data decoding scheme this data of decoding.A plurality of data decoding schemes comprise the guiding decoding scheme at least.If the number of data satisfies the condition that presets, then be defined as guiding decoding scheme with the data decoding scheme.

And the data decode process may further comprise the steps: obtain corresponding to the guiding reference value of a plurality of data with corresponding to the guiding difference value of this guiding reference value, and utilize this guiding reference value and guiding difference value to obtain data.

In addition, in this signal processing method, data are parameters.And sound signal utilizes these parameters to recover.In this signal processing method, receive the identification information corresponding to number of parameters, and utilize the identification information that receives to generate the parameter of this number.By considering the data number, indicate the identification information of a plurality of data decoding schemes to be extracted by layering.

In the step of extracting identification information, extract the first identification information of indication the first data decoding scheme, then utilize the second identification information of the number extraction indication second data decoding scheme of the first identification information and data.In this case, the first identification information indicates whether it is the DIFF decoding scheme.And the indication of the second identification information is guiding decoding scheme or PCM marshalling scheme.

In signal processing method according to another embodiment of the invention, if the number of a plurality of data satisfies the condition that presets, then utilize guiding reference value and these data corresponding to a plurality of data to generate the guiding difference value.Then transmit the guiding difference value that generates.In this signal processing method, transmission be the guiding reference value.

In signal processing method according to still another embodiment of the invention, according to the number specified data decoding scheme of a plurality of data.Then according to determined data decoding scheme data are encoded.In this case, a plurality of data decoding schemes comprise the guiding decoding scheme at least.If the number of data satisfies the condition that presets, then the data decoding scheme is defined as guiding decoding scheme.

A kind ofly comprise that for the treatment of the device of signal the number that obtains corresponding to the number of the data of guiding reference value obtains parts, obtains the guiding reference value and obtain parts and utilize this guiding reference value and guide the data acquisition section that difference value obtains data corresponding to the value of the guiding difference value of this guiding reference value when the number of data satisfies the condition that presets according to one embodiment of present invention.In this case, number obtains parts, value obtains parts and data acquisition section is included among the above-mentioned data decode

parts

91 or 92.

A kind of according to another embodiment of the invention for the treatment of the device of signal comprise according to the number of a plurality of data determine in a plurality of data decoding schemes one scheme determining means and according to the decoding parts of determined data decoding scheme to decoding data.In this case, a plurality of data decoding schemes comprise the guiding decoding scheme at least.

The number that a kind of device of according to still another embodiment of the invention processing signals is included in a plurality of data utilizes guiding reference value and these data corresponding to a plurality of data to generate the guiding difference values when satisfying the condition preset value generates the output block of the guiding difference value that parts and transmission generate.In this case, value being generated parts is included within the above-mentioned

data encoding parts

31 or 32.

A kind ofly comprise according to the scheme determining means of the number specified data decoding scheme of a plurality of data and the addressable part of data being encoded according to the established data decoding scheme for the treatment of the device of signal according to still a further embodiment.In this case, a plurality of data decoding schemes comprise the guiding decoding scheme at least.

2-4.PBC signal processing method

Signal processing method and device according to the PBC of utilization decoding feature of the present invention are explained as follows.

In signal processing method according to an embodiment of the invention, obtain corresponding to the guiding reference value of a plurality of data with corresponding to the guiding difference value of this guiding reference value.Subsequently, utilize this guiding reference value and guiding difference value to obtain data.And the method also comprises at least one step of decoding in guiding difference value and the guiding reference value.In this case, the data of application PBC are parameters.And the method also can comprise the step of the Reconstruction sound signal that utilization is obtained.

A kind of device of according to one embodiment of present invention processing signals comprises corresponding to the guiding reference value of a plurality of data with corresponding to the guiding difference value of this guiding reference value and utilize this guiding reference value and the guiding difference value obtains the data acquisition section of data.In this case, value is obtained parts and data acquisition section is included among the above-mentioned

data decoding parts

91 or 92.

A kind of signal processing method according to another embodiment of the invention may further comprise the steps: utilize guiding reference value and these data corresponding to a plurality of data to generate the guiding difference value and export the guiding difference value that generates.

A kind of signal processing apparatus according to another embodiment of the invention comprises guiding reference value and the value generation parts of these data generation guiding difference value and the output block of the guiding difference value that output generates that utilizes corresponding to a plurality of data.

Signal processing method according to still another embodiment of the invention may further comprise the steps: obtain corresponding to the guiding reference value of a plurality of gains with corresponding to the guiding difference value of this guiding reference value, and utilize this guiding reference value and guiding difference value to obtain gain.And the method also comprises at least one step of decoding in guiding difference value and the guiding reference value.In addition, the method also comprises the step of utilizing the gain reconstructed audio signals of obtaining.

In this case, the guiding reference value can be the value of normal use, a value being set to default value or extracting from table in the average intermediate value of the mean value of a plurality of gains, a plurality of gains, a plurality of gain.And the method also can comprise having that gain selection of high coding efficiency as the step of finally guiding reference value after in a plurality of gains each being arranged the guiding reference value.

A kind of comprise obtaining corresponding to the guiding reference value of a plurality of gains with corresponding to the value of the guiding difference value of guiding reference value for the treatment of the device of signal according to still another embodiment of the invention obtain parts and utilize this guiding reference value and guide the gain that difference value obtains gain and obtain parts.

A kind of signal processing method according to still another embodiment of the invention may further comprise the steps: utilize guiding reference value and this a plurality of Gain generating guiding difference values corresponding to a plurality of gains, and export the guiding difference value that generates.

And signal processing apparatus according to still a further embodiment comprises guiding reference value and the value calculating unit of a plurality of Gain generating guiding difference value and the output block of the guiding difference value that output generates that utilizes corresponding to a plurality of gains.

(3.DIFF differential decoding)

DIFF decoding is the decoding scheme that a kind of utilization is present in the relation between a plurality of data in the data group of distinguishing, and it also can be called as " differential decoding ".In this case, refer to used by above-mentioned data marshalling parts 10 the final group of specific marshalling scheme as the data group of the unit that uses DIFF.The data with specific meanings of organizing into groups in top mode in the present invention, are defined as " parameter " to be explained.And this is identical with the explanation for PBC.

Particularly, the DIFF decoding scheme is to utilize difference value between the parameter be present in same group, the decoding scheme of the difference value between the adjacent parameter more specifically.

Type and the detailed applications example of explaining in detail the DIFF decoding scheme with reference to figure 5 to 8 are as follows.

3-1.DIFF type

Fig. 5 is the figure that explains DIFF decoding type according to the present invention.DIFF decoding is distinguished according to the direction that draws difference value from adjacent parameter.

For example, DIFF decoding type can be classified as DIFF (hereinafter referred to as " DIFF_FREQ " or " DF ") on the frequency direction and the DIFF (hereinafter referred to as " DIFF_TIME " or " DT ") on the time orientation.

With reference to figure 5, group-1 indication is calculated the DIFF (DF) of difference value at frequency axis, and organize-2 or group-3 calculate difference value at time shaft.

As can see from Figure 5, again distinguish the DIFF (DT) that calculates difference value at time shaft according to the direction that draws difference value at time shaft.

For example, be applied to organize-2 DIFF (DT) corresponding to the scheme that between the parameter value of current time and the parameter value of previous moment (for example, group-1), draws difference value.This is called as backward time D IFF (DT) (hereinafter referred to as " DT-BACKWARD ").

For example, be applied to organize-3 DIFF (DT) corresponding to the scheme that between the parameter value of current time and next parameter value (for example, group-4) constantly, draws difference value.This is called as forward direction time D IFF (DT) (hereinafter referred to as " DT-FORWARD ").

Therefore, as shown in Figure 5, group the-1st, DIFF (DF) decoding scheme, and organize the-the 2nd, DIFF (DT-BACKWARD) decoding scheme, and organize the-the 3rd, DIFF (DT-FORWARD) decoding scheme.Yet the decoding scheme of group-4 is not determined.

In the present invention, although the DIFF on frequency axis only is defined as a decoding scheme (for example, DIFF (DF)), also can makes definition and it be divided into " DIFF (DF-TOP) " and " DIFF (DF-BOTTOM) ".

3-2.DIFF the example of using

Fig. 6 to 8 is the figure that use the example of DIFF decoding scheme.

In Fig. 6, for the ease of explaining, with the group shown in Fig. 5-1 with organize-2 as an example.Group-1 is followed DIFF (DF) decoding scheme, and its parameter value is x[n]={ 11,12,9,12,10,8,12,9,10,9}.Group-2 is followed DIFF (DF-BACKWARD) decoding scheme, and its parameter value is y[n]={ 10,13,8,11,10,7,14,8,10,8}.

Fig. 7 illustrates the result of the difference value of calculating group-1.Because group the-1st is deciphered by DIFF (DF) decoding scheme, so can calculate difference value by formula 2.Formula 2 is illustrated in that draw on the frequency axis and difference value last parameter.

[formula 2]

d[0]＝x[0]

D[n]=x[n]-x[n-1], n=1 wherein, 2 ..., 9.

Particularly, DIFF (DF) result of the group-1 that obtains by formula 2 is d[n]=11,1 ,-3,3 ,-2 ,-2,4 ,-3,1 ,-1}.

Fig. 8 illustrates the result of the difference value of calculating group-2.Because group-2 is deciphered by DIFF (DF-BACKWARD) decoding scheme, so can calculate difference value by formula 3.Formula 3 is illustrated in the difference value that draws on the time shaft with last parameter.

[formula 3]

D[n]=y[n]-x[n], n=1 wherein, 2 ..., 9.

Particularly, DIFF (DF-BACKWARD) result of the group-2 that obtains by formula 3 is d[n]=1,1 ,-1 ,-1,0,01,2 ,-1,0 ,-1}.

4. the selection of data decoding scheme

The invention is characterized in by mixing various data decoding schemes and compress or data reconstruction.So, in decoding during particular group, need to from least three kinds or more select a kind of decoding scheme in the data decoding scheme.And, the identification information of selected decoding scheme should be delivered to the decoding parts by bit stream.

According to the method for selection data decoding scheme of the present invention and utilize the interpretation method of the method and device to be explained as follows.

The method of processing signals may further comprise the steps according to one embodiment of present invention: obtain the data decoding identification information, and according to the data decoding scheme of data decoding identification information indication data are carried out data decode.

In this case, the data decoding scheme comprises the PBC decoding scheme at least.And the PBC decoding scheme utilizes corresponding to the guiding reference value of a plurality of data and guides difference value to decoding data.And the guiding difference value utilizes described data and guiding reference value to generate.

The data decoding scheme also comprises the DIFF decoding scheme.The DIFF decoding scheme is corresponding to one of DIFF-DF scheme and DIFF-DT scheme.And the DIFF-DT scheme is corresponding to one of forward direction time D IFF-DT (FORWARD) scheme and backward time D IFF-DT (BACKWARD).

This signal processing method is further comprising the steps of: obtain entropy decoding identification information, and utilize the entropy decoding scheme of this entropy decoding identification information indication that data are carried out the entropy decoding.

In the data decode step, carry out data decode through the data of entropy decoding by the data decoding scheme.

And this data processing method also comprises the step of described data being come decoded audio signal as parameter.

Device for the treatment of signal comprises according to one embodiment of present invention:

The identification information that obtains the data decoding identification information obtains parts and according to the decoding parts that data carried out data decode by the data decoding scheme of data decoding identification information indication.

The method of processing signals may further comprise the steps according to another embodiment of the invention: according to the data decoding scheme data are carried out data encoding, and generate the data decoding identification information with transmission designation data decoding scheme.

In this case, the data decoding scheme comprises the PBC decoding scheme at least.The utilization of PBC decoding scheme is encoded to data corresponding to guiding reference value and the guiding difference value of a plurality of data.And the guiding difference value utilizes described data and guiding reference value to generate.

Carry out the addressable part of data encoding to data and generate output block with the data decoding identification information of transmission designation data decoding scheme a kind of comprising according to the data decoding scheme for the treatment of the device of signal according to another embodiment of the invention.

Select the method for data decoding scheme and transmission decoding to select the method for identification information to be explained as follows by optimum transmission efficiency.

4-1. consider the data decoding identification method of usage frequency

Fig. 9 explains according to the present invention selecting for the moment block diagram of relation of at least three kinds of decoding schemes.

With reference to figure 9, suppose: have the first to the 3rd

data encoding parts

53,52 and 51, wherein the usage frequency of the first data encoding parts 53 is minimum, and the usage frequency of the 3rd data encoding parts 51 is the highest.

For the ease of explaining, for sum " 100 ", suppose that the usage frequency of the first data decode parts 53 is " 10 ", the usage frequency of the second data encoding parts 52 is " 30 ", and the usage frequency of the 3rd data encoding parts 51 is " 60 ".Particularly, for 100 data groups, can think that the PCM scheme is employed 10 times, the PBC scheme is employed 30 times and the DIFF scheme is employed 60 times.

Based on above hypothesis, the required bit number of identification information of three kinds of decoding schemes of sign calculates in the following manner.

For example, according to Fig. 9, because use the first information of 1 bit, so use 100 bits as sign 100 groups the first information of decoding scheme altogether.The 3rd data encoding parts 51 of high usage frequency identify by this 100 bit because have, so the second information of all the other 1 bits can only utilize 40 bits to distinguish the first data encoding parts 53 and the second data encoding parts 52.

Therefore, as the result of " first information (100 bit)+the second information (40 bit) ", for the identification information of the decoding type of every group of 100 data group selection altogether needs altogether 140 bits.

Figure 10 is the block diagram of explaining the relation when selecting a kind of at least three kinds of decoding schemes according to correlation technique.

As shown in Figure 9, for the ease of explaining, for sum " 100 ", suppose that the usage frequency of the first data decode parts 53 is " 10 ", the usage frequency of the second data encoding parts 52 is " 30 ", and the usage frequency of the 3rd data encoding parts 51 is " 60 ".

In Figure 10, the required bit number of identification information of three kinds of decoding scheme types of sign calculates in the following manner.

At first, according to Figure 10, because use the first information of 1 bit, so use 100 bits as sign 100 groups the first information of decoding scheme altogether.

The first data encoding parts 53 with minimum usage frequency are preferentially identified by this 100 bit.So, the second informational needs of all the other 1 bits altogether 90 bits with on distinguish the second data encoding parts 52 and the 3rd data encoding parts 51.

Therefore, as the result of " first information (100 bit)+the second information (90 bit) ", for the identification information of the decoding type of every group of 100 data group selection altogether needs altogether 190 bits.

Situation shown in Figure 9 and situation shown in Figure 10 are compared, can find out that data decoding selection identification information shown in Figure 9 more has superiority aspect transfer efficiency.

That is, exist in the situation of at least three kinds or more kinds of data decoding schemes, the invention is characterized in utilize different identification informations rather than by same identification information differentiation in similar two kinds of decoding scheme types each other aspect the usage frequency.

For example, as shown in figure 10, be classified in the situation of same identification information at the first addressable part 51 and the second addressable part 52, the data transmission bit increases, thereby reduces transfer efficiency.

Exist in the situation of at least three kinds of data decoding types, the invention is characterized in to distinguish by the first information to have the data decoding scheme of high usage frequency.So by the second information, all the other the two kinds of decoding schemes with low usage frequency are distinguished respectively.

Figure 11 and Figure 12 are respectively the process flow diagrams according to data decoding selection scheme of the present invention.

In Figure 11, suppose that DIFF decoding is to have the data decoding scheme of high usage frequency.In Figure 12, suppose that PBC decoding is to have the data decoding scheme of high usage frequency.

With reference to Figure 11, check that still there be not (S10) in the PCM decoding existence with minimum usage frequency.As mentioned above, this checks according to the first information that is used for sign and carries out.

As the result who checks, if PCM decoding checks then whether PBC deciphers (S20).This is according to the second information and executing that is used for sign.

In the usage frequency of DIFF decoding in 100 times is being 60 times situation altogether, the identification information that is used for every group decoding type selecting for 100 identical data groups needs altogether 140 bits, i.e. " first information (100 bit)+the second information (40 bit) ".

With reference to Figure 12, be similar to Figure 11, check that still there be not (S30) in the PCM decoding existence with minimum usage frequency.As mentioned above, this checks according to the first information that is used for sign and carries out.

As the result who checks, if PCM decoding checks then whether DIFF deciphers (S40).This is by the second information and executing that is used for sign.

In the usage frequency of DIFF decoding in 100 times is being 80 times situation altogether, the identification information that is used for every group decoding type selecting for 100 identical data groups needs altogether 120 bits, i.e. " first information (100 bit)+the second information (20 bit) ".

The method of a plurality of data decoding schemes of sign according to the present invention and utilize the signal processing method of the method and device to be explained as follows.

The method of processing signals may further comprise the steps according to one embodiment of present invention: layering extract a plurality of data decoding schemes of indication identification information and according to corresponding to the data decoding scheme of identification information to decoding data.

In this case, to be included in the identification information of PBC decoding scheme in a plurality of data decoding schemes and DIFF decoding scheme be to extract from different layers in indication.

In decoding step, obtain data corresponding to the reference value of a plurality of data with the difference value that utilizes these data to generate according to the utilization of data decoding scheme.In this case, reference value is guiding reference value or difference reference value.

The method of processing signals comprises the step of the identification information of the three or more at least data decoding schemes of layering extraction indication according to another embodiment of the invention.In this case, the identification information that has two kinds of decoding schemes of high usage frequency in the sign information is to extract from different layers.

The method of processing signals may further comprise the steps according to still another embodiment of the invention: identification information is extracted in the usage frequency layering according to the identification information of designation data decoding scheme, and according to corresponding to the data decode scheme of identification information to decoding data.

In this case, the mode of extracting the first identification information and the second identification information with layering is extracted identification information.The first identification information indicates whether it is the first data decoding scheme, and the second identification information indicates whether it is the second data decoding scheme.

The first identification information indicates whether it is the DIFF decoding scheme.And the indication of the second identification information is guiding decoding scheme or PCM marshalling scheme.

The first data decoding scheme can be the PCM decoding scheme.And the second data decoding scheme can be PBC decoding scheme or DIFF decoding scheme.

Described data are parameters, and this data processing scheme also comprises the step of utilizing all Reconstruction sound signals.

The device of processing signals comprises that layering extract to distinguish that the identifier of the identification information of a plurality of data decoding schemes extracts parts (for example, " 710 " among Figure 13) and according to corresponding to the data decoding scheme of this identification information decoding parts to decoding data according to one embodiment of present invention.

The method of processing signals may further comprise the steps according to another embodiment of the invention: according to the data decoding scheme data are encoded, and generate the identification information of distinguishing the mutually different data decoding scheme of usage frequency of using when data are encoded.

In this case, identification information is distinguished PCM decoding scheme and PBC decoding scheme mutually.Particularly, identification information is distinguished PCM decoding scheme and DIFF decoding scheme.

And, the device of processing signals according to still another embodiment of the invention comprises that the addressable part of data being encoded according to the data decoding scheme and the identification information that generates the identification information of distinguishing the mutually different data decoding scheme of usage frequency of using when data are encoded generate parts (for example, " 400 " among Figure 11).

4-2. the relation between data decoding

At first, between PCM of the present invention, PBC and DIFF, there is separate and/or relevant relation.For example, can be freely a kind of in three kinds of decoding of each group selection type that becomes the data decoding target.So conceptual data decoding has brought the result that three kinds of decoding scheme types are bonded to each other and use.Yet, by considering the usage frequency of three kinds of decodings, at first select to have a kind of in the DIFF decoding scheme of optimum usage frequency and all the other two kinds of decoding schemes (for example, PCM and PBC).Subsequently, next selects a kind of among PCM and the PBC.Yet as mentioned above, this is the transfer efficiency that will consider identification information, and does not ascribe the similarity of actual decoding scheme to.

Aspect the similarity of decoding scheme, PBC and DIFF calculate aspect the difference value similar each other.So the decode procedure of PBC and DIFF is significantly overlapping each other.Particularly, the step by difference value reconstruction initial parameter when decoding is defined as " Δ decoding ", and can be designed to process in same step.

In the process of carrying out PBC or DIFF decoding, may there be the parameter that departs from its scope.In this case, need by independent PCM decoding and parameter corresponding to transmission.

[marshalling]

1. the concept of organizing into groups

The present invention proposes to bundle deal with data in order to " marshalling " of efficient decoding by the data with appointment.Particularly, in the situation of PBC decoding, select for unit by group because guide reference value, so the marshalling process must be the step that will finish before the PBC decoding carrying out.To organize into groups and be applied in an identical manner DIFF decoding.And according to some also is applicable to entropy decoding in the marshalling scheme of the present invention, this will explain in the corresponding description part of back.

Can marshalling type of the present invention be classified into reference to the manner of execution of marshalling " outside marshalling " and " inner marshalling ".

Perhaps, can marshalling type of the present invention be classified into " territory marshalling ", " data marshalling " and " sound channel marshalling " with reference to the marshalling target.

Perhaps, can marshalling type of the present invention be classified into reference to the marshalling execution sequence " the first marshalling ", " the second marshalling " and " the 3rd marshalling ".

Perhaps, can carry out counting with reference to marshalling and marshalling type of the present invention is classified into " single marshalling " and " repeatedly marshalling ".

Yet above marshalling classification is carried out for the ease of passing on concept of the present invention, the term of its use is not limited.

Marshalling according to the present invention be with various marshalling schemes with overlapping each other with or the mode of the usefulness of interosculating finish.

In the following description, explain according to marshalling of the present invention with outside marshalling by marshalling being distinguished into inner marshalling.Subsequently, the wherein multiple marshalling of multiple marshalling type coexistence will be explained.And the concept with domain of interpretation marshalling and data marshalling.

2. inner marshalling

The execution that inner marshalling refers to organize into groups is at internal implementation.If carry out on the whole inner marshalling, then internally again marshalling of previous group is to generate new group or cut apart group.

Figure 13 is the figure that explains inner marshalling according to the present invention.

With reference to Figure 13, for example, inner marshalling according to the present invention is implemented by frequency domain unit's (hereinafter referred to as " frequency band ").So inner marshalling scheme can be organized into groups corresponding to a class field sometimes.

If sampled data is by specific filter, for example, QMF (quadrature mirror filter) then generates a plurality of subbands.In the subband pattern, carry out first frequency and organize into groups to generate the first group of frequency band that can be described as parameter band.The first frequency marshalling can be by bundling to generate parameter band with subband brokenly.So, can dispose the parameter band that differs in size.Yet, according to decoding purpose, comparably configuration parameter frequency band.And the step that generates subband can be classified as class marshalling.

Subsequently, carry out the second frequency marshalling in the parameter band that generates, can be described as second group of frequency band of frequency ranges of data with generation.The second frequency marshalling can generate frequency ranges of data by the parameter band unification with consistent number (uniform number).

Finish the rear purpose according to decoding of marshalling, can be by carrying out decoding corresponding to the parameter band unit of first group of frequency band or by the frequency ranges of data unit corresponding to second group of frequency band.

For example, when using above-mentioned PBC decoding, parameter band that can be by will marshalling is as a group or by selecting to guide reference value (a class group reference value) with the frequency ranges of data of marshalling as a group.PBC utilizes selected guiding reference value to realize, that explains in the detailed operation of PBC and the above description is identical.

Again for example, when using above-mentioned DIFF decoding, by the parameter band of marshalling is determined the group reference value as a group, then calculate difference value.Perhaps, also can be by the frequency ranges of data of marshalling be determined the group reference value and is calculated difference value as a group.And that explains in the detailed operation of DIFF and the above description is identical.

If with first and/or frequency marshalling use and actual decoding, then must information corresponding to transmission, explain with reference to Figure 23 after a while.

3. outside marshalling

The situation that the execution that outside marshalling refers to organize into groups is externally implemented.If implement generally outside marshalling, then externally again marshalling of previous group is to generate new group or combination group.

Figure 14 is the diagram of the interprets external marshalling according to the present invention.

With reference to Figure 14, for example, outside marshalling according to the present invention is implemented by time domain unit's (hereinafter referred to as " time slot ").So outside marshalling scheme can be organized into groups corresponding to a class field sometimes.

The frame that comprises sampled data is carried out very first time marshalling, to generate first group of time slot.Figure 14 exemplarily illustrates and generates eight time slots.Very first time marshalling also has the meaning that a frame is divided into equal-sized a plurality of time slots.

At least one in the time slot that generates organized into groups in selection by the very first time.Figure 14 illustrates

time slot

1,4,5 and 8 the situation selected.According to decoding scheme, in selecting step, can select whole time slots.Then with selected

time slot

1,4,5 with 8 again layout become

time slot

1,2,3 and 4.Yet, according to the target of decoding, can be partly the selected time slot 1 of layout, 4,5 and 8 again.In this case, because the time slot of getting rid of outside layout again is excluded outside final group forms, so got rid of from PBC or DIFF decoding target.

Selected time slot is carried out the marshalling of the second time, to be formed in the group of processing together on the final time shaft.

For example,

time slot

1 and 2 or

time slot

3 and 4 can consist of a group, this group is called as time slot pair.Again for example,

time slot

1,2 and 3 can consist of a group, and this group is called as time slot three restructuring (timeslot triple).And the single time slot with other time slot formation group can not exist.

Marshalling is applied in the situation of actual decoding in the first and second times, needs the corresponding information of transmission, explains with reference to Figure 23 after a while.

4. multiple marshalling

Multiple marshalling refers to inside marshalling, outside marshalling and various other marshalling are mixed to generate the marshalling scheme of final group.As mentioned in above description, can use by overlapping each other or interosculating according to each marshalling scheme of the present invention.And, multiple marshalling is used as a kind of scheme that improves the efficient of various decoding schemes.

4-1. blend interior marshalling and outside marshalling

Figure 15 is the figure that explains according to multiple marshalling of the present invention, wherein inside marshalling and outside marshalling is mixed.

With reference to Figure 15, the frequency band 64 of final marshalling generates finish inner marshalling in frequency domain after.And

final time slot

61,62 and 63 generates finish outside marshalling in time domain after.

A single time slot that obtains after marshalling is finished is called as data set.In Figure 15, Reference numeral 61a, 61b, 61a, 62b and 63 be the designation data collection respectively.

Especially, two

data set

61a and 61b or two data set 62a and 62b can consist of a pair of by outside marshalling in addition.A pair of data set is called as " data to ".

After multiple marshalling is finished, carry out PBC or DIFF decoding and use.

For example, in the situation of carrying out PBC decoding, guiding reference value P1, P2 or P3 are selected for the data finally finished to 61 or 62 or each right data set 63 of composition data not.Then utilize selected guiding reference value to carry out PBC decoding.

For example, in the situation of carrying out DIFF decoding, for data set 61a, 61b, 62a, 62b, 63 determine DIFF decoding type separately.Mentioned as described above, should be wherein each data set and determine a DIFF direction, and it can be defined as one of " DIFF-DF " and " DIFF-DT ".Mentioned identical in the process of carrying out DIFF decoding according to determined DIFF decoding scheme and the above description.

In order to come composition data pair by carrying out outside marshalling, each right data set of reply composition data is carried out the inside marshalling that is equal in multiple marshalling.

For example, the right data set 61a of composition data has identical frequency ranges of data number separately with 61b.And the right data set 62a of composition data has identical frequency ranges of data number separately with 62b.Yet it is no problem belonging to respectively the frequency ranges of data number of the right data set (for example, 61a and 62a) of different data different.

This means and the marshalling of different inside can be applied to each data pair.

In the right situation of configuration data, can carry out the first marshalling and can carry out the second marshalling by outside marshalling by inner marshalling.

For example, the frequency ranges of data number that obtains after the second marshalling is done to specify corresponding to the frequency ranges of data number that obtains after the first marshalling and is doubly taken advantage of.This is because each right data set of composition data has identical frequency ranges of data number.

4-2. blend interior marshalling and inner marshalling

Figure 16 and Figure 17 are respectively the figure that explains mixing marshalling according to another embodiment of the invention.Particularly, Figure 16 and Figure 17 concentrated area illustrate the mixing of inner marshalling.So, obviously in Figure 16 or Figure 17, carried out and maybe can carry out outside marshalling.

For example, again carry out the situation of inner marshalling after Figure 16 is illustrated in second frequency marshalling and finishes in the situation of generated data frequency band.Particularly, the frequency ranges of data that generates by the second frequency marshalling is divided into low-frequency band and high frequency band.In the situation of specific decoding, must utilize dividually low-frequency band or high frequency band.Particularly, separately utilize the situation of low-frequency band and high frequency band to be called as " bimodulus ".

So in the situation of bimodulus, data decoding is to carry out as a group by the low or high frequency band that will finally generate.For example, be respectively low and high frequency band generation guiding reference value P1 and P2, then in corresponding frequency band, carry out PBC decoding.

Bimodulus can be used according to the characteristic of every sound channel.So, can be referred to as " sound channel marshalling ".And, also can differentially use bimodulus according to data type.

For example, Figure 17 shows the situation of again carrying out inner marshalling in the situation of generated data frequency band after the marshalling of above-mentioned second frequency is finished.That is the frequency ranges of data that, generates by the second frequency marshalling is divided into low-frequency band and high frequency band.In the situation of specific decoding, only utilize low-frequency band, and need to abandon high frequency band.Concrete, only marshalling utilizes the situation of low-frequency band to be called as " low-frequency channel (LFE) pattern ".

In low-frequency channel (LFE) pattern, come executing data decoding by the low-frequency band that will finally generate as a group.

For example, for low-frequency band generates guiding reference value P1, then in corresponding low-frequency band, carry out PBC decoding.Yet, can generate new frequency ranges of data by carry out inner marshalling in selected low-frequency band.This is the low-frequency band that will show for the concentrated area marshalling.

And low-frequency channel (LFE) pattern is used according to the low-frequency channel characteristic, and can be referred to as " sound channel marshalling ".

5. territory marshalling and data marshalling

Can marshalling be classified into territory marshalling and data marshalling with reference to the target of marshalling.

The territory marshalling refers to the scheme in the upper marshalling of special domain (for example, frequency domain or time domain) territory unit.And, can carry out the territory marshalling by above-mentioned inside marshalling and/or outside marshalling.

And the data marshalling refers to organize into groups the scheme of data itself.The data marshalling can be carried out by above-mentioned inside marshalling and/or outside marshalling.

In the data marshalling in particular cases, can carry out marshalling, so that it is available in entropy decoding.For example, the data marshalling can be used for the entropy decoding to real data under the marshalling state of finally finishing shown in Figure 15.That is, the processed mode of data is to make in two data adjacent one another are on the frequency direction or on one of the time orientation to be bundled in together.

Yet in the situation of in the above described manner implementation data marshalling, the data in final group are partly organized into groups again.So PBC or DIFF not only are applied to the group (for example, two data) through the data marshalling.In addition, will explain after a while the entropy decoding scheme of organizing into groups corresponding to data.

6. use the signal processing method of marshalling

6-1. use at least the signal processing method of inner marshalling

Signal processing method and the device of the above-mentioned marshalling scheme of a kind of use according to the present invention are explained as follows.

The method of processing signals according to an embodiment of the invention may further comprise the steps: obtain corresponding to group reference value corresponding to a plurality of data that are included in a group by the first marshalling and inside marshalling of being used for this first marshalling and the difference value corresponding with this group reference value, and utilize this group reference value and difference value to obtain data.

The invention is characterized in that the data number of organizing into groups by the first marshalling processing is greater than the data number of organizing into groups by inside.Organizing in this case reference value can be guiding reference value or difference reference value.

Method according to an embodiment of the invention also comprises at least one step of decoding in group reference value and the difference value.In this case, the guiding reference value is determined by group.

And, preset respectively the number that is included in the data in the inner group by the inside marshalling.In this case, it is different to be included in the number of the data in the inner group.

On frequency domain, data are carried out the first marshalling and inner marshalling.In this case, frequency domain can be corresponding in hybrid domain, parameter band territory, frequency ranges of data territory and the sound channel territory.

And, the invention is characterized in via the first marshalling obtain first group to comprise the inside group that obtains via the inside marshalling.

Frequency domain of the present invention is distinguished by frequency band.Frequency band becomes sub-band by the inside marshalling.Sub-band becomes parameter band by the inside marshalling.Parameter band becomes frequency ranges of data by the inside marshalling.In this case, the number of parameter band can be limited to maximum 28.And, parameter band can be organized into a data frequency band by 2,5 or 10.

The device of processing signals according to an embodiment of the invention comprises and obtains the group reference value corresponding with a plurality of data that are included in a group by the first marshalling and inside marshalling of being used for this first marshalling and obtain parts with the value of difference value corresponding to this group reference value, and utilizes this group reference value and difference value to obtain the data acquisition section of data.

The method of processing signals according to another embodiment of the invention may further comprise the steps: utilize corresponding to organizing into groups and be used for group reference value and these data of organizing into groups a plurality of data that are included in a group in this first inside of organizing into groups by first and generate a difference value, and transmit the difference value that generates.

And, the device of processing signals according to another embodiment of the invention comprises utilizing corresponding to the group reference value of a plurality of data that are included in a group by the first marshalling and inside marshalling of being used for this first marshalling and the value that these data generate a difference value and generates parts, and the output block of the difference value that generates of transmission.

6-2. utilize the signal processing method of multiple marshalling

A kind of signal processing method and device according to above-mentioned marshalling scheme of the present invention is explained as follows.

A kind of method of processing signals according to an embodiment of the invention may further comprise the steps: obtain be included in a group by marshalling in the corresponding group reference value of a plurality of data and corresponding to the difference value of this group reference value, and utilization group reference value and difference value obtain data.

In this case, the group reference value can be one of guiding reference value and difference reference value.

And marshalling can be corresponding to one of outside marshalling and inner marshalling.

In addition, marshalling can be corresponding to one of territory marshalling and data marshalling.

The data marshalling can be carried out in the territory group.And, be included in the territory marshalling time domain in processing and comprise in time slot territory, parameter set territory and the data set territory at least one.

Be included in the territory marshalling frequency domain in processing and comprise in sampling territory, subband domain, hybrid domain, parameter band territory, frequency ranges of data territory and the sound channel territory at least one.

Can a plurality of data from be included in group set a difference reference value.And, determine at least one during whether organize into groups counting, marshalling scope and marshalling exists.

A kind of device of processing signals according to an embodiment of the invention comprise obtain be included in a group by marshalling in the corresponding group reference value of a plurality of data and obtain parts corresponding to the value of the difference value of this group reference value, and utilization group reference value and difference value obtain the data acquisition section of data.

A kind of method of processing signals according to another embodiment of the invention may further comprise the steps: group reference value and these data that utilization is corresponding with a plurality of data be included in a group by marshalling in generate a difference value, and transmit the difference value that generates.

The value that the group reference value that a plurality of data during a kind of device of processing signals according to another embodiment of the invention comprises utilization and is included in a group by marshalling are corresponding and these data generate a difference value generates parts, and the output block that transmits the difference value that generates.

A kind of method of processing signals according to another embodiment of the invention may further comprise the steps: obtain with by comprising that the first marshalling and second is grouped in that interior marshalling is included in the corresponding group reference value of a plurality of data in the group and corresponding to the first difference value of this group reference value, and utilization group reference value and the first difference value obtain data.

In this case, the group reference value can comprise guiding reference value or difference reference value.

The method also comprises at least one step of decoding in group reference value and the first difference value.And the first guiding reference value is determined by group.

In this case, the second marshalling can comprise the outside or inner marshalling for the first marshalling.

On in time domain and frequency domain at least one data are organized into groups.Particularly, marshalling is at least one territory of the organizing into groups marshalling in time domain and the frequency domain.

Time domain can comprise time slot territory, parameter set territory or data set territory.

Frequency domain can comprise sampling territory, subband domain, hybrid domain, parameter band territory, frequency ranges of data territory or sound channel territory.And the data of marshalling (grouped data) are index or parameter.

Utilization is included in an indicated entropy table of the index in the group by the first marshalling the first difference value is carried out the entropy decoding.And utilization group reference value and first difference value of decoding through entropy obtain data.

Utilization is included in an indicated entropy table of the index in the group by the first marshalling the first difference value and group reference value is carried out the entropy decoding.And, utilize through the group reference value of entropy decoding and first difference value of decoding through entropy and obtain data.

A kind of device of processing signals according to another embodiment of the invention comprise obtain with by comprising that the first marshalling and second is grouped in interior marshalling and is included in the corresponding group reference value of a plurality of data in the group and obtains parts corresponding to the value of the difference value of this group reference value, and utilization group reference value and difference value obtain the data acquisition section of data.

A kind of method of processing signals according to another embodiment of the invention may further comprise the steps: utilize with by comprising that the marshalling of the first marshalling and second in being grouped in is included in the corresponding group reference value of a plurality of data in the group and generates a difference value with these data, and transmit the difference value that generates.

A kind of device of processing signals according to another embodiment of the invention comprise utilize with by comprising that the marshalling of the first marshalling and second in being grouped in is included in the value generation parts that the corresponding group reference value of a plurality of data in the group and these data generate a difference value, and the output block that transmits the difference value that generates.

A kind of method of processing signals according to another embodiment of the invention may further comprise the steps: obtain the group reference value corresponding with a plurality of data that are included in a group by the first marshalling and outside marshalling of being used for this first marshalling and corresponding to the difference value of this group reference value, and utilization group reference value and difference value obtain data.

In this case, the first data number corresponding with the number of processing data organize into groups by the first marshalling less than with corresponding second the keeping count of of the number of the data of organizing into groups by outside marshalling processing.And, have the relation of doubly taking advantage of between the first data number and the second data number.

The group reference value can comprise guiding reference value or difference reference value.

The method also comprises at least one step of decoding in group reference value and the difference value.

By group the guiding reference value is decoded.

On in time domain and frequency domain at least one data are organized into groups.Time domain can comprise time slot territory, parameter set territory or data set territory.And frequency domain can comprise sampling territory, subband domain, hybrid domain, parameter band territory, frequency ranges of data territory or sound channel territory.

The method comprises that also the data that will obtain are as the step of Reconstruction sound signal.And outside marshalling can comprise paired parameter.

A kind of device of processing signals according to another embodiment of the invention comprises and obtains the group reference value corresponding with a plurality of data that are included in a group by the first marshalling and outside marshalling of being used for this first marshalling and obtain parts corresponding to the value of the difference value of this group reference value, and utilization group reference value and difference value obtain the data acquisition section of data.

A kind of method of processing signals according to still another embodiment of the invention may further comprise the steps: utilize group reference value and these data corresponding with a plurality of data that are included in a group by the first marshalling and outside marshalling of being used for this first marshalling to generate difference value, and transmit the difference value that generates.

And, device according to another embodiment processing signals of the present invention comprises: the value of utilizing the group reference value corresponding with a plurality of data that are included in a group by the first marshalling and outside marshalling of being used for this first marshalling and these data to generate a difference value generates parts, and the output block that transmits the difference value that generates.

6.3. the signal processing method of usage data marshalling at least

A kind of method of processing signals according to an embodiment of the invention may further comprise the steps: obtain the group reference value corresponding with a plurality of data that are included in a group by data marshallings and the inside marshalling that is used for the data marshalling and corresponding to the difference value of this group reference value, and utilization group reference value and difference value obtain data.

In this case, be included in the number of the data in the inner marshalling less than the number that is included in the data in the data marshalling.And data are corresponding to parameter.

Inner marshalling is intactly to carry out in a plurality of data through the data marshalling.In this case, can carry out inner marshalling by parameter band.

Inner marshalling can partly be carried out in a plurality of data through the data marshalling.In this case, can carry out inner marshalling by the sound channel of a plurality of each in the data of data marshalling.

The method also comprises at least one step of decoding in group reference value and the difference value.In this case, the guiding reference value is determined by group.

Can be to the marshalling of data executing data and inner marshalling on frequency domain.

Frequency domain can comprise in sampling territory, subband domain, hybrid domain, parameter band territory, frequency ranges of data territory and the sound channel territory.When obtaining data, the marshalling information of at least one in usage data marshalling and the inner marshalling.

Marshalling information comprise position, each group of each group number, by the existence of group set of applications reference value whether, the coding and decoding scheme of the number of group reference value, group reference value and obtain in the whether existence of group reference value at least one.

A kind of device of processing signals according to an embodiment of the invention comprises and obtains the group reference value corresponding with a plurality of data that are included in a group by data marshallings and the inside marshalling that is used for the data marshalling and obtain parts corresponding to the value of the difference value of this group reference value, and utilization group reference value and difference value obtain the data acquisition section of data.

A kind of method of processing signals according to another embodiment of the invention may further comprise the steps: utilize group reference value and these data corresponding with a plurality of data that are included in a group by data marshalling and the inside marshalling that is used for the data marshalling to generate a difference value, and transmit the difference value that generates.

And, the device of processing signals comprises that the value of utilizing the group reference value corresponding with a plurality of data that are included in a group by data marshalling and the inside marshalling that is used for the data marshalling and these data to generate a difference value generates parts according to another embodiment of the invention, and the output block that transmits the difference value that generates.

[entropy decoding]

1. the concept of entropy decoding

The result that entropy decoding according to the present invention refers at data decoding carries out the process of variable-length decoding.

Generally speaking, the probability of occurrence of concrete data is processed in entropy decoding with statistical.For example, by improving on the whole transfer efficiency to the less bit of data allocations that has the high frequency of occurrences at probability to the more bit of the data allocations that has the low frequency of occurrences at probability.

And the present invention is intended to propose a kind of and general entropy decoding high-level efficiency entropy interpretation method different, that interknit with PBC decoding and DIFF decoding.

1-1. entropy table

At first, decoding is necessary to predetermined entropy table for entropy.The entropy table is defined as code book.And addressable part and decoding parts use same entropy table.

The present invention proposes a kind of entropy interpretation method and uniqueness entropy table to process expeditiously various data decoding results.

1-2. entropy decoding type (1D/2D)

Entropy decoding of the present invention is classified as two types.One type is to derive an index (index 1) by an entropy table, and another type is to derive two index (index 1 and index 2) in succession by an entropy table.The former is called as " decoding of 1D (one dimension) entropy " and the latter is called as " decoding of 2D (bidimensional) entropy ".

Figure 18 is the exemplary plot according to 1D of the present invention and 2D entropy table.With reference to Figure 18, entropy table of the present invention comprises index field, length field and code word field substantially.

For example, if concrete data (for example, guiding reference value, difference value etc.) are to calculate by above-mentioned data decoding, then corresponding data (corresponding to " index ") have the code word by the appointment of entropy table.Code word is transformed into bit stream and then is transferred to the decoding parts.

The entropy decoding parts that received code word are identified for the entropy table of corresponding data, then utilize the bit length of this code word of configuration in corresponding code word and the determined table to derive index value.In this case, the present invention is shown as sexadecimal with codeword table.

Positive sign (+) or the negative sign (-) of being deciphered the index value of deriving by 1D or 2D entropy are omitted.So, need to after the decoding of 1D or 2D entropy is finished, compose sign.

In the present invention, differentially give sign symbol according to 1D or 2D.

For example, in the situation of 1D entropy decoding, if corresponding index is not " 0 ", then distributes and transmit 1 independent bit symbol position (for example, " bsSign ").

In the situation of 2D entropy decoding, because in succession extract two index, so be that mode by the relation between two index that extract of programming determines whether to distribute the sign symbol position.In this case, this program use between index index and value, two extractions of two extractions difference and corresponding to the maximum value in the entropy table (lav).Compare with the situation of the sign symbol position being distributed to each index in the situation of simple 2D, this can reduce the transmitted bit number.

Index is that the 1D entropy table of deriving one by one can be used for all data decoding results therein.Yet, derive therein two index 2D entropy tables at every turn and use limited for particular case.

For example, if data decoding be not by above-mentioned marshalling process obtain to (pair), then 2D entropy matrix section uses limited.And the use of 2D entropy table is limited to the guiding reference value that calculates as the PBC decode results.

Therefore, as described in the above description, the mode that entropy decoding of the present invention is characterised in that entropy is deciphered and the result of data decoding connects each other is utilized the most efficient entropy decoding scheme.Explain in details as follows.

1-3.2D method (time pairing/frequency pairing)

Figure 19 is the exemplary plot that is used for two kinds of methods of 2D entropy decoding according to the present invention.The decoding of 2D entropy is for the process that derives two index adjacent one another are.So, can be according to these two the in succession direction differentiation 2D entropy decodings of index.

For example, two index situation adjacent one another are on frequency direction is called " 2D frequency pairing (hereinafter referred to as 2D-FP) ".And, two index situation adjacent one another are on time orientation is called " 2D time match (hereinafter referred to as 2D-TP) ".

With reference to Figure 19,2D-FP can dispose respectively different concordance lists with 2D-TP.Scrambler must be determined according to the result of data decode the entropy decoding scheme of full blast.

Explain in the following description the method for determining efficiently entropy decoding with data decoding with connecting each other.

1-4. entropy decoded signal disposal route

A kind of method of entropy decoding processing signals of utilizing according to the present invention is explained as follows.

In a kind of method of processing signals according to an embodiment of the invention, obtain corresponding to the reference value of a plurality of data with corresponding to the difference value of this reference value.Subsequently, difference value is carried out the entropy decoding.Then the difference value that uses reference value and decode through entropy obtains data.

The method also comprises the step of reference value being carried out the entropy decoding.And the method also can comprise uses the reference value of decoding through entropy and the step of obtaining data through the difference value of entropy decoding.

The method also comprises the step of obtaining entropy decoding identification information.And entropy decoding is according to being carried out by the indicated entropy decoding scheme of entropy decoding identification information.

In this case, the entropy decoding scheme is a kind of in 1D decoding scheme and the multidimensional decoding scheme (for example, 2D decoding scheme).And the multidimensional decoding scheme is frequency to a kind of in (TP) decoding scheme of (FP) decoding scheme and time.

Reference value can comprise one of guiding reference value and difference reference value.

And this signal processing method also can comprise the step of utilizing data to come reconstructed audio signals as parameter.

A kind of device for the treatment of signal according to an embodiment of the invention comprises obtaining corresponding to the reference value of a plurality of data with corresponding to the value of the difference value of this reference value and obtains parts, difference value is carried out the entropy decoding parts of entropy decoding and utilizes reference value and obtain the data acquisition section of data through the difference value of entropy decoding.

In this case, value is obtained parts and is included in the above-mentioned bit stream multichannel Knock-Down Component 60, and data acquisition section is included in the above-mentioned data decode

parts

91 or 92.

A kind of method of processing signals according to another embodiment of the invention may further comprise the steps: use reference value and these data corresponding to a plurality of data to generate a difference value, the difference value that generates is carried out the entropy coding, and output is through the difference value of entropy coding.

In this case, reference value is carried out the entropy coding.Transmission is through the reference value of entropy coding.

The method also comprises the step that generates the entropy decoding scheme that is used for the entropy coding.And, the entropy decoding scheme that transmission generates.

A kind of device for the treatment of signal according to another embodiment of the invention comprises that use generates parts, the difference value that generates is carried out the entropy addressable part of entropy coding and the output block of exporting the difference value of encoding through entropy with the value that these data generate a difference value corresponding to the reference value of a plurality of data.

In this case, value generation parts are included in the above-mentioned data encoding parts 31 or 32.And output block is included in the above-mentioned bit stream multiplexing components 50.

A kind of method of processing signals according to another embodiment of the invention may further comprise the steps: obtain the data corresponding to a plurality of data decoding schemes, use determines to be included in guiding reference value and guiding at least one used entropy table in the difference value in these data for the entropy table identifier of data decoding scheme uniqueness, and uses this entropy table that in guiding reference value and the guiding difference value at least one carried out the entropy decoding.

In this case, the entropy table identifier is uniqueness for one in guiding decoding scheme, frequency differential decoding scheme and the time difference decoding scheme.

And each in the difference value is uniqueness to the entropy table identifier for guiding reference value and guiding.

The entropy table is uniqueness for the entropy table identifier, and comprises in directory, difference on the frequency submeter and the mistiming submeter one.

Perhaps, the entropy table is not uniqueness for the entropy table identifier, and in difference on the frequency submeter and the mistiming submeter one can be shared.

But the entropy table frequency of utilization table of difference corresponding to the guiding reference value.In this case, by 1D entropy decoding scheme the guiding reference value is carried out the entropy decoding.

The entropy decoding scheme comprises 1D entropy decoding scheme and 2D entropy decoding scheme.Particularly, 2D entropy decoding scheme comprise frequency to (2D-FP) decoding scheme and time to (2D-TP) decoding scheme.

And, but this method usage data comes reconstructed audio signals as parameter.

A kind of device of processing signals according to another embodiment of the invention comprises and obtaining corresponding to the guiding reference value of a plurality of data and the entropy decoding parts that obtain parts and the guiding difference value is carried out the entropy decoding corresponding to the value of the guiding difference value of this guiding reference value.And this device also comprises the data acquisition section of using the guiding reference value and obtaining data through the guiding difference value of entropy decoding.

A kind of method of processing signals according to still another embodiment of the invention may further comprise the steps: use guiding reference value and these data corresponding to a plurality of data to generate the guiding difference value, the guiding difference value that generates is carried out the entropy coding, and transmission is through the guiding difference value of entropy coding.

In this case, the table for the entropy coding can comprise the guiding special table.

The method also comprises the step of the guiding reference value being carried out the entropy coding.And transmission is through the guiding reference value of entropy coding.

The method also comprises the entropy decoding scheme that generates for the entropy coding.And, the entropy decoding scheme that transmission generates.

A kind of device of processing signals according to still another embodiment of the invention comprises that the value of using guiding reference value and these data corresponding to a plurality of data to generate the guiding difference values generates parts, the guiding difference value that generates is carried out the entropy addressable part of entropy coding and transmission through the output block of the guiding difference value of entropy coding.

2. with the relation of data decoding

As described above described, the present invention proposes three kinds of data decoding schemes.Yet, for do not carry out entropy decoding according to the data of PCM scheme.Explain respectively in the following description relation between the decoding of PBC decoding and entropy and the relation between DIF decoding and the entropy decoding.

2-1.PBC decoding and entropy decoding

Figure 20 is the figure according to the entropy decoding scheme for the PBC decode results of the present invention.

As mentioned above, after finishing PBC decoding, calculate a guiding reference value and a plurality of difference value.And guiding reference value and difference value all become the target of entropy decoding.

For example, according to above-mentioned grouping method, determine to use the group of PBC decoding.In Figure 20, for the ease of explaining, with on the situation of paired group (pair) on the time shaft and the time shaft the situation of group (non-pair) is not as an example in pairs.Entropy decoding explanation after PBC decoding is finished is as follows.

At first, explain in the situation 83 of not organizing in pairs execution PBC decoding.A guiding reference value that becomes the entropy decoding target is carried out the decoding of 1D entropy, and can carry out the decoding of 1D entropy or the decoding of 2D entropy to remaining difference value.

Particularly, because in the situation of not organizing in pairs, just have one group for a data set on the time shaft, so can not carry out the decoding of 2D-TP entropy.Even carry out 2D-FP, also tackle and after the derivation index is to (pairs of indexes), fail to consist of the parameter value execution 1D entropy decoding among the right last frequency band 81a.Behind the entropy decoding scheme of one definite every data, namely use corresponding entropy table to generate code word.

Because the present invention relates to for example is the situation that a group generates a guiding reference value, should carry out the decoding of 1D entropy.Yet, in another embodiment of the present invention, if from a group, generate at least two guiding reference values, can carry out the decoding of 2D entropy in succession guiding reference value.

Secondly, the situation 84 in pairs group execution PBC decoding is explained as follows.

A guiding reference value that becomes the entropy decoding target is carried out the decoding of 1D entropy, can carry out the decoding of 1D entropy, the decoding of 2D-FP entropy or the decoding of 2D-TP entropy to remaining difference value.

Particularly, because organizing in pairs in the situation of (pair), there is a group for two data sets adjacent one another are on the time shaft, so can carry out the decoding of 2D-TP entropy.Even carry out 2D-FP, also tackle derive index to after fail to consist of a right last frequency band 81b or the parameter value among the 81c is carried out the decoding of 1D entropy.Yet, such as among Figure 20 confirmation, in the situation of using the decoding of 2D-TP entropy, do not have last frequency band and fail to consist of a right situation.

2-2.DIFF decoding and entropy decoding

Figure 21 is the figure according to the entropy decoding scheme for the DIFF decode results of the present invention.

As mentioned above, after finishing DIFF decoding, a guiding reference value and a plurality of difference value have been calculated.And guiding reference value and difference value all become the target of entropy decoding.Yet in the situation of DIFF-DT, reference value may not exist.

For example, according to above-mentioned grouping method, determine to use the group of DIFF decoding.In Figure 21, for the ease of explaining, with on the situation of paired group (pair) on the time shaft and the time shaft the situation of group (non-pair) is not as an example in pairs.And Figure 21 illustrates according to DIFF decoding direction will divide into DIFF-DT on the time-axis direction and the situation of the DIFF-DF on the frequency axis direction as the data set of data decoding unit.

Entropy decoding explanation after DIFF decoding is finished is as follows.

At first, explain in the situation of not organizing in pairs execution DIFF decoding.In situation about not organizing in pairs, a data set is present on the time shaft.And this data set can be deciphered direction according to DIF becomes DIFF-DF or DIFF-DT.

For example, if a data set of not organizing in pairs is DIFF-DF (85), then reference value becomes the parameter value among the first frequency band 82a.Reference value is carried out the decoding of 1D entropy, and can carry out the decoding of 1D entropy or the decoding of 2D-FP entropy to remaining difference value.

That is, reach in the situation about not organizing in pairs at DIFF-DF, on time shaft, have a group for a data set.So, can not carry out the decoding of 2D-TP entropy.Even carry out 2D-FP, deriving index to rear, also reply is failed to consist of a parameter value among the right last frequency band 83a and is carried out the decoding of 1D entropy.One has determined decoding scheme for each data, then uses corresponding entropy table to generate code word.

For example, be in the situation of DIFF-DT (86) at a data set of not organizing in pairs, because do not have reference value in corresponding data centralization, process so do not carry out " the first frequency band ".So, can carry out the decoding of 1D entropy or the decoding of 2D-FP entropy to difference value.

At DIFF-DT and be in the situation about not organizing in pairs, the data set that draw difference value may be to fail to consist of adjacent data collection or a data set in another audio frame that data are right.

That is, at DIFF-DT and be not organize in pairs in the situation of (86), on time shaft, there is a group for a data set.So, can not carry out the decoding of 2D-TP entropy.Even carry out 2D-FP, deriving index to rear, also reply is failed to consist of a parameter value in the right outcome parameter frequency band and is carried out the decoding of 1D entropy.Yet Figure 21 illustrates the situation of failing to consist of a right last frequency band that for example do not exist just.

Once determined decoding scheme for each data, then use corresponding entropy table to generate code word.

Secondly, explain in the situation of organizing in pairs (pair) execution DIFF decoding.In the situation of organizing in pairs executing data decoding, two data sets consist of a group on time shaft.And the data set in each group becomes DIFF-DF or DIFF-DT according to DIFF decoding direction separately.So, it can be classified as situation that to consist of two a pair of data sets all be DIFF-DF (87), consist of two a pair of data sets and all be the situation of DIFF-DT and consist of the situation that two a pair of data sets have respectively different decoding directions (for example, DIFF-DF/DT or DIFF-DT/DF) (88).

For example, all be in the situation of DIFF-DF (being DIFF-DF/DF) (87) consisting of two a pair of data sets, if each of data centralization is not paired (non-paired) and is DIFF-DF, then all available entropy decoding schemes all can be carried out.

For example, become parameter value among the first frequency band 82b or the 82c in each reference value of corresponding data centralization, and reference value is carried out the decoding of 1D entropy.And, can carry out the decoding of 1D entropy or the decoding of 2D-FP entropy to remaining difference value.

Even in corresponding data set, carry out 2D-FP, deriving index to rear, also reply fails to consist of a right last frequency band 83b or the decoding of the execution of the parameter value among 83c 1D entropy.Because two one of data set formations are right, can carry out the decoding of 2D-TP entropy.In this case, sequentially carry out the decoding of 2D-TP entropy on the frequency band in the scope of the extremely last frequency band of next frequency band that in corresponding data set, does not comprise the

first frequency band

82b or 82c.

If carry out the decoding of 2D-TP entropy, then do not generate and fail to consist of a right last frequency band.

One determines the entropy decoding scheme of every data, namely uses corresponding entropy table to generate code word.

For example, all be that DIFF-DT (that is, in situation DIFF-DT/DT) (89), because there is not reference value in the corresponding data set, does not process so do not carry out the first frequency band consisting of two right data sets.And, can carry out the decoding of 1D entropy or the decoding of 2D-FP entropy by all difference values in each data set.

Even in corresponding data set, carry out 2D-FP, deriving index to rear, also tackle the parameter value of failing to consist of in the right last frequency band and carry out the decoding of 1D entropy.Yet, being and not having the example of failing to consist of a right last frequency band shown in Figure 21.

Because two one of data set formations are right, the decoding of 2D-TP entropy is executable.In this case, in corresponding parameter set on the frequency band in from the first frequency band to last frequency band range order carry out the decoding of 2D-TP entropy.

For example, may exist two right data sets of formation to have respectively the situation (that is, DIFF-DF/DT or DIFF-DT/DF) (88) of different decoding directions.Figure 21 illustrates the example of DIFF-DF/DT.All entropy decoding schemes that can be suitable for the corresponding decoding type of each data set executive basis basically in this case.

For example, among the DIFF-DF data set in the middle of two right data sets of formation, with the reference value in the corresponding data set (DIFF-DF) parameter value in the first frequency band 82d is carried out the decoding of 1D entropy.And, can carry out the decoding of 1D entropy or the decoding of 2D-FP entropy to remaining difference value.

Even in corresponding data set (DIFF-DF), carry out 2D-FP, deriving index to rear, also reply fails to consist of the parameter value execution 1D entropy decoding in the right last frequency band 83d.

For example, among the DIFF-DT data group in the middle of two right data sets of formation, because reference value does not exist, process so do not carry out the first frequency band.And, can carry out the decoding of 1D entropy or the decoding of 2D-FP entropy to all difference values in the corresponding data set (DIFF-DT).

Even in corresponding data set (DIFF-DT), carry out 2D-FP, deriving index to rear, also reply fails to consist of the parameter value execution 1D entropy decoding in the right last frequency band.Yet, being and not having the example of failing to consist of a right last frequency band shown in Figure 21.

Has respectively mutually different decoding direction because consist of two right data sets, so can carry out the decoding of 2D-TP entropy.In this case, sequentially carry out the decoding of 2D-TP entropy on the frequency band in the frequency band range of the extremely last frequency band of next frequency band behind the first frequency band that comprises the first frequency band 82d from eliminating.

2-3. entropy decoding and marshalling

As mentioned above, in the situation of 2D-FP or 2D-TP, use a code word to extract two index.So, this means to be entropy decoding execution marshalling scheme.And, can be referred to as " time marshalling " or " frequency marshalling ".

For example, addressable part is organized into groups two index that extract in the data decoding step in frequency or time orientation.

Subsequently, addressable part uses the selection of entropy table to represent a code word of the index (grouped indexes) of two marshallings, then transmits selected code word in the bit stream by it is included in.

The decoding parts receive and are included in the code word that two index in the bit stream obtain by marshalling, and use applied entropy table to extract two index values.

2-4. the signal processing method by data decoding and entropy decoding Relations Among

Characteristic explain according to the signal processing method by the relation between the relation between PBC decoding and the entropy decoding and DIFF decoding and the entropy decoding of the present invention is as follows.

A kind of method of processing signals according to an embodiment of the invention may further comprise the steps: obtain difference information, according to comprising that time marshalling and the frequency entropy decoding scheme in being grouped in carries out entropy to this difference information and decodes, and according to comprising that the data decode scheme of guiding difference, time difference and frequency differential carries out data decode to difference information.And that explains in the detailed relation between the decoding of data decoding and entropy and the above description is identical.

A kind of method of processing signals according to another embodiment of the invention may further comprise the steps: obtain digital signal, according to the entropy decoding scheme this digital signal is carried out entropy decoding, and according to comprising that at least one of a plurality of data decoding schemes of guide decoding scheme carry out data decode to the digital signal of decoding through entropy.In this case, the entropy decoding scheme can be determined according to the data decoding scheme.

A kind of device of processing signals according to another embodiment of the invention comprises the signal acquisition parts that obtain digital signal, this digital signal is carried out the entropy decoding parts of entropy decoding and according to the data decode parts that comprise that at least one of a plurality of data decoding schemes of guide decoding scheme are carried out data decode to the digital signal of decoding through entropy according to the entropy decoding scheme.

A kind of method of processing signals according to still another embodiment of the invention may further comprise the steps: by the data decoding scheme digital signal is carried out data encoding, encode to carry out entropy through the digital signal of data encoding by the entropy decoding scheme, and transmission is through the digital signal of entropy coding.In this case, the entropy decoding scheme can be determined according to the data decoding scheme.

And, according to still another embodiment of the invention the device of processing signals comprise by the data decoding scheme to digital signal carry out the data encoding parts of data encoding, by the entropy decoding scheme to carry out the entropy addressable part of entropy coding through the digital signal of data encoding.And this device comprises that also transmission is through the output block of the digital signal of entropy coding.

3. the selection of entropy table

The entropy table that is used for entropy decoding is determined automatically according to data decoding scheme and the data type that becomes the entropy decoding target.

For example, if if data type is CLD parameter and entropy decoding target is the guiding reference value, the 1D entropy table that then table name is called " hcodPilot_CLD " is used for entropy decoding.

For example, if data type is the CPC parameter, if data decoding is DIFF-DF, and if the entropy decoding target be the first frequency band values, the 1D entropy table that then table name is called " hcodFirstband_CPC " is used for entropy decoding.

For example, if data type is the ICC parameter, if the data decoding scheme is PBC, and if entropy decoding carried out by 2D-TP, the 2D-PC/TP entropy table that then table name is called " hcod2D_ICC_PC_TP_LL " is used for entropy decoding.In this case, the maximum value in " LL " indicating gauge during the 2D table name claims (hereinafter referred to as " LAV ").And, will explain after a while maximum value (LAV).

For example, if data type is the ICC parameter, if the data decoding scheme is DIF-DF, and if entropy decoding carried out by 2D-FP, the 2D-FP entropy table that then table name is called " hcod2D_ICC_DF_FP_LL " is used for entropy decoding.

That is it is very important, determining to use which the execution entropy decoding in a plurality of entropy tables.And, preferably will be for the characteristic of each data that becomes each entropy target and stark suitable entropy table is configured to independently.

Yet, be used for having each other similarly the entropy table of the data of attribute and can share use.For a representational example, if data type is " ADG " or " ATD ", then can use CLD entropy table.And, " the first frequency band " entropy table can be applied to the guiding reference value that PBC deciphers.

Below explain in detail the method for using maximum value (LAV) selective entropy table.

3-1. the maximum value of entropy table (LAV)

Figure 22 is the figure that explains according to the method for selective entropy table of the present invention.

A plurality of entropy tables shown in Figure 22 (a) are at the table of selective entropy table shown in (b) of Figure 22.

As mentioned above, there are a plurality of entropy tables according to data decoding and data type.

For example, the entropy table can comprise be applicable to data type be " xxx " situation the entropy table (for example, table 1 is to 4), be applicable to data type be " yyy " situation the entropy table (for example, table 5 is to 8), the special-purpose entropy table of PBC (for example, table k to k+1), effusion entropy table (for example, table n-2～n-1) and LAV index entropy table (for example, table n).

Particularly, come allocation list although preferably give a code word by each index that may occur in corresponding data, if so, the size of table significantly increases.And, manage unwanted or almost absent variable index be inconvenient.In the situation of 2D entropy table, because too many index may occur, these problems have been brought more inconvenience.In order to address these problems, will use maximum value (LAV).

For example, if the scope of the index value of specific data type " X～+ X " (X=15) between, then in this scope, select at least one LAV that has the high frequency of occurrences at probability, and it be configured in the independent table.

For example, when configuration CLD entropy table, can provide the table of " LAV=3 ", the table of " LAV=5 ", the table of " LAV=7 " or the table of " LAV=9 ".

For example, in (a) of Figure 22, table-1 91a can be arranged to the CLD table of " LAV=3 ", will show the CLD table that-2 91b are arranged to " LAV=5 ", to show the CLD table that-3 91c are arranged to " LAV=7 ", table-4 91d arrange the CLD table to " LAV=9 ".

The index that departs from the LAV scope in the LAV table is by effusion entropy table (for example, table n-2～n-1) process.

For example, when the CLD table 91c that uses " LAV=7 " carries out decoding, if there is the index that departs from maximal value " 7 " (for example, 8,9 ..., 15), then corresponding index (is for example shown n-2～n-1) process by effusion entropy table individually.

Equally, can be that another data type (for example, ICC, CPC etc.) arranges the LAV table by the mode identical with the CLD table.Yet the LAV of each data has different values, because the scope of each data type is different.

For example, when configuration ICC entropy table, for example, can provide the table of " LAV=1 ", the table of " LAV=3 ", the table of " LAV=5 " or the table of " LAV=7 ".When configuration CPC entropy table, for example, can provide the table of " LAV=3 " table, " LAV=6 ", the table of " LAV=9 " and the table of " LAV=12 ".

3-2. be used for the entropy table of LAV index

The present invention adopts the LAV index to utilize LAV selective entropy table.That is, shown in Figure 22 (b), the LAV value of each data type is distinguished by the LAV index.

Particularly, in order to select the last entropy table that uses, then the LAV index of confirming the data type that each is corresponding confirms the LAV corresponding to the LAV index." LL " in the formation that the final LAV value of confirming claims corresponding to above-mentioned entropy table name.

For example, if data type is the CLD parameter, if the data decoding scheme is DIFF-DF, if entropy decoding carried out by 2D-FP, and if " LAV=3 ", then table name is called that " the entropy table of hcod2D_CLD_DF_FP_03 is used for entropy decoding.”

When confirming the LAV index of each data type, the invention is characterized in the LAV index is used the entropy table individually.This means that LAV index self processes as the target of entropy decoding.

For example, the table-n in Figure 22 (a) is as LAV index entropy table 91e.It is expressed as table 1.

[table 1]

LavIdx	Bit length	Code word [sexadecimal/scale-of-two]
			0	1	0x0(0b)

1	2	0x2(10b)
			2	3	0x6(110b)
3	3	0x7(111b)

This table expression LAV index value itself is from statistically seeing different aspect the usage frequency.

For example, because " LAV Index=0 " has the highest usage frequency, distribute 1 bit to it.And, distribute 2 bits for " LAV Index=1 " with second high usage frequency.At last, distribute 3 bits for " LAV=2 or 3 " with low usage frequency.

In the situation of not using LAV index entropy table 91e, should transmit 2 bit identification information, when using LAV entropy table, to distinguish four kinds of LAV index at every turn.

Yet, if use LAV index entropy table 91e of the present invention, for example for have at least 60% usage frequency " it is just enough that LAV Index=0 situation is transmitted 1 bit codewords.So the present invention can make transfer efficiency be higher than the transfer efficiency of related art method.

In this case, the LAV index entropy table 91e in the table 1 is applied to the situation of four kinds of LAV index.And, if obviously more LAV index is arranged then transfer efficiency will be higher.

3-3. the signal processing method that uses the entropy table to select

Below explain signal processing method and the device that utilizes above-mentioned entropy table to select.

The method of processing signals may further comprise the steps according to one embodiment of present invention: obtain index information, index information is carried out the entropy decoding, and sign is corresponding to the content of the index information of decoding through entropy.

In this case, index information is the information about the index with usage frequency probability nature.

As mentioned above, make the special-purpose entropy table of index of reference 91e carry out the entropy decoding to index information.

Content is sorted out according to data type, and is used for data decode.And content can become marshalling information (grouping information).

Marshalling information is about the information with a plurality of data marshallings.

And the index of entropy table is the maximum value (LAV) in the index that is included in the entropy table.

In addition, the entropy table uses when parameter is carried out the decoding of 2D entropy.

A kind of device of processing signals according to an embodiment of the invention comprises the acquisition of information parts that obtain index information, index information is carried out the decoding parts of entropy decoding and sign corresponding to the mark component through the content of the index information of entropy decoding.

A kind of method of according to another embodiment of the invention processing signals may further comprise the steps: generating indexes information index information is carried out the entropy coding, and transmission is through the index information of entropy coding with sign content.

A kind of device of processing signals according to another embodiment of the invention comprises that generating indexes information is with the Information generation parts of sign content, carry out entropy fgs encoder parts and transmission through the information output parts of the index information of entropy coding to index information.

A kind of method of processing signals according to another embodiment of the invention may further comprise the steps: obtain difference value and index information, index information is carried out the entropy decoding, sign is corresponding to the entropy table of the index information of decoding through entropy, and the entropy table that utilization identifies carries out the entropy decoding to difference value.

Subsequently, will be used for obtaining data corresponding to reference value and the difference value through decoding of a plurality of data.In this case, reference value can comprise guiding reference value and difference reference value.

Make the special-purpose entropy table of index of reference carry out the entropy decoding to index information.And, according to each the type in a plurality of data the entropy table is sorted out.

Data are parameter, and the method also comprises the step of utilizing the Reconstruction sound signal.

In the situation of difference value being carried out the entropy decoding, use the entropy table that difference value is carried out the decoding of 2D entropy.

In addition, the method is further comprising the steps of: obtain reference value, and use the entropy table that is exclusively used in this reference value that reference value is carried out the entropy decoding.

A kind of device of processing signals according to another embodiment of the invention comprises the input block that obtains difference value and index information, carry out the entropy decoding to index information index decoding parts, sign are corresponding to through the table mark component of the entropy table of the index information of entropy decoding and the data decode parts that use the entropy table that identifies difference value to be carried out the entropy decoding.

This device also comprises and uses corresponding to the reference value of a plurality of data and obtain the data acquisition section of data through the difference value of decoding.

A kind of method of processing signals according to still another embodiment of the invention may further comprise the steps: use reference value and these data corresponding to a plurality of data to generate a difference value, use the entropy table that difference value is carried out the entropy coding, and generating indexes information is to identify this entropy table.

And this method is further comprising the steps of: index information is carried out the entropy coding, and transmission is through index information and the difference value of entropy coding.

And, the device of processing signals according to still another embodiment of the invention comprises: the value of using reference value and these data corresponding to a plurality of data to generate a difference value generates parts, use the entropy table difference value to be carried out the value addressable part of entropy coding, generating indexes information is identifying the Information generation parts of this entropy table, and the index addressable part that index information is carried out the entropy coding.And this device also comprises the information output parts of transmission entropy code index information and difference value.

[data structure]

Below explain the data structure that comprises the various information that are associated with above-mentioned data decoding, marshalling and entropy decoding according to of the present invention.

Figure 23 is the slice map according to data structure of the present invention.

With reference to Figure 23, data structure according to the present invention comprises head 100 and a plurality of frame 101 and 102.The configuration information that is common to subordinate's

frame

101 and 102 comprises in head 100.And configuration information comprises the marshalling information for above-mentioned marshalling.

For example, marshalling information comprises very first time marshalling information 100a, first frequency marshalling information 100b and sound channel marshalling information 100c.

In addition, the configuration information in the head 100 is called as " main configuration information ", and the message part that is recorded in the frame is called as " payload (payload) ".

Particularly, for example explain in the following description the situation that data structure of the present invention is applied to audio space information.

At first, the very first time marshalling information 100a in the head 100 becomes " bsFrameLength " field of specifying the timeslot number in the frame.

First frequency marshalling information 100b becomes " bsFreqRes " field of specifying a frame intrinsic parameter frequency band number.

Sound channel marshalling information 100c represents " OttmodeLFE-bsOttBands " field and " bsTttDualrnode-bsTttBandsLow " field." OttmodeLFE-bsOttBands " field is the information of specifying the parameter band number that is applied to the LFE sound channel.And " bsTttDualmode-bsTttBandsLow " field is the information of specifying the parameter band number of the bimodulus medium and low frequency band with low-frequency band and high frequency band.Yet, also can " bsTttDualmode-bsTttBandsLow " field not classified as sound channel marshalling information but classify as frequency and organize into groups information.

Comprise separately frame information (Frame Info) 101a that is common to all groups in the frame and a plurality of groups of 101b and 101c in the

frame

101 and 102.

Frame information 101a comprises selection of time information 103a, the second time marshalling information 103b and second frequency marshalling information 103c.In addition, frame information 101a is called as " the sub-configuration information " that is applied to each frame.

At length, for example, explain in the following description the situation that data structure of the present invention is applied to audio space information.

Selection of time information 103a in the frame information 101a comprises " bsNumParamset " field, " bsParamslot " field and " bsDataMode " field.

" bsNumParamset " field is the information that indication is present in the number of the parameter set in the whole frame.

And " bsParamslot " field is the information of specifying the position of the time slot that has parameter set.

In addition, " bsDataMode " field is the information of specifying the Code And Decode disposal route of each parameter set.

For example, in the situation of " bsDataMode=0 " of design parameter collection (for example, default mode), the decoding parts are replaced corresponding parameter set with default value.

In the situation of " bsDataMode=1 " of design parameter collection (for example, preceding mode), the decoding parts are kept the decode value of previous parameter set.

In the situation of " bsDataMode=2 " of design parameter collection (for example, the interpolation pattern), the decoding parts calculate corresponding parameter set by the interpolation between parameter set.

At last, the design parameter collection " in the situation of bsDataMode=3 (for example, read mode), the decoding data that is used for corresponding parameter set has been transmitted in expression.So a plurality of groups of 101b and 101c in the frame are the groups that is configured to the transmission of data in the situation of " bsDataMode=3 " (for example, read mode).Therefore, the decoding type information in addressable part is organized with reference to each is to decoding data

Below explain in detail signal processing method and the device of use according to an embodiment of the invention " bsDataMode " field.

The signal processing method of a kind of use according to an embodiment of the invention " bsDataMode " field may further comprise the steps: obtaining mode information, according to the data attribute by pattern information indication obtain corresponding to the guiding reference value of a plurality of data and and corresponding to the guiding difference value of this guiding reference value, and use the guiding reference value and guide difference value and obtain data.

In this case, data are parameters, and the method also comprises the step of operation parameter reconstructed audio signals.

If pattern information indication read mode then obtains the guiding difference value.

Pattern information also comprises at least one in default mode, preceding mode and the interpolation pattern.

And the guiding difference value obtains by the group frequency band.

In addition, this signal processing method use the first parameter (for example, dataset) sign read mode number and use the second parameter (for example, setidx) to obtain the guiding difference value based on the first variable.

The signal processing apparatus of a kind of use according to an embodiment of the invention " bsDataMode " field comprises the acquisition of information parts of obtaining mode information, obtains corresponding to the guiding reference value of a plurality of data and corresponding to the value of the guiding difference value of guiding reference value according to the data attribute by the pattern information indication and obtain the data acquisition section that parts and use guiding reference value and guiding difference value obtain data.

And, acquisition of information parts, value are obtained parts and data acquisition section is arranged in above-mentioned data decode

parts

91 or 92.

A kind of signal processing method of use " bsDataMode " field according to another embodiment of the invention may further comprise the steps: the pattern information, the use that generate the designation data attribute generate the guiding difference value and transmit the difference value that generates corresponding to guiding reference value and these data of a plurality of data.And the method also comprises the step that the difference value that generates is encoded.

A kind of signal processing apparatus of use " bsDataMode " field according to another embodiment of the invention comprises that the Information generation parts of the pattern information of generation designation data attribute, use generate the value generation parts of guiding difference value and the output block that transmits the difference value that generates corresponding to guiding reference value and these data of a plurality of data.And value generates parts and is arranged in the above-mentioned

data encoding parts

31 or 32.

The second time marshalling information 103b among the frame information 101a comprises " bsDatapair " field." bsDatapair " field is to specify by whether there being the right information of data set between the data set of " bsDataMode=3 " appointment.Particularly, two data sets are organized into a group by " bsDatapair " field.

Second frequency marshalling information in the frame information 101a comprises " bsFreqResStride " field." bsFreqResStride " field is to doing for the second time information of marshalling by the parameter of at first having organized into groups as first frequency marshalling information 100b " bsFreqRes " field.That is, frequency ranges of data adds up to the parameter equal by the span of " bsFreqResStride " field appointment to generate by binding.So, provide parameter value by frequency ranges of data.

Group

101b and 101c comprise data decoding type information 104a, entropy decoding type information 104b, code word 104 and auxiliary data 104d separately.

At length, for example, below explain the situation that data structure of the present invention is used for audio space information.

At first, the data decoding type information 104a in each among group 101b and the 101c comprises " bsPCMCoding " field, " bsPilotCoding " field, " bsDiffType " field and " bdDifftimeDirection " field.

" bsPCMCoding " field is that the data decoding of the corresponding group of sign is the information of PCM scheme or DIFF scheme.

Only have when " bsPCMCoding " field is specified the PCM scheme, just specify the PBC scheme whether to exist by " bsPilotCoding " field.

" bsDiffType " field is the information of specifying the decoding direction in the situation of using the DIFF scheme.And " bsDiffType " field is specified " DF:DIFF-FREQ " or " DT:DIFF-TIME ".

And " bdDifftimeDirection " field is to be that decoding direction on the axle of lower fixed time of situation of " DT " is the information of " FORWARD " or " BACKWARD " in " bsDiffType " field.

Entropy decoding type information 104b among group 101b and the 101c each comprises " bsCodingScheme " field and " bsPairing " field.

" bsCodingScheme " field is that appointment entropy coding is 1D or the information of 2D.

And " bsPairing " field is that the direction of two index of extraction in the situation of " bsCodingScheme " field appointment " 2D " is frequency direction (FP: frequency is matched) or time orientation (TP: the information time pairing).

Code word 104c among group 101b and the 101c each comprises " bsCodeW " field.And " bsCodeW " field is specified the code word on the table that is applied to entropy decoding.So the major part of above-mentioned data becomes the target of entropy decoding.In this case, they can transmit by " bsCodeW " field.For example, become the LAV index value of the guiding reference value of target of entropy decoding and PBC decoding by the transmission of " bsCodeW " field.

Auxiliary data 104d among group 101b and the 101c each comprises " bsLsb " field and " bsSign " field.Particularly, auxiliary data 104d comprises " bsLsb " field and " bsSign " field, and is decoded into other data of not transmitting by " bsCodeW " field by entropy.

" bsLsb " field is the field that is applied to above-mentioned partial parameters, and is only to be the supplementary of just transmitting in " CPC " and the situation that belongs to non-rudenss quantization in data type.

And " bsSign " field is the information that specifies in the sign symbol of the index that extracts in the situation of using the decoding of 1D entropy.

In addition, the data by the transmission of PCM scheme are included among the auxiliary data 104d.

Below explain the feature according to signal handle data structures of the present invention.

At first, signal handle data structures according to the present invention comprises: have the data decoding information of the guiding decoding information that comprises at least frame by frame meter and at least one the payload part in the entropy decoding information, and have the head part for the main configuration information of payload part.

Main configuration information comprises the very first time message part and the first frequency message part that has for the frequency information of all frames that has for the temporal information of all frames.

And main configuration information also comprises the first inner marshalling message part that has for frame by frame the random groups that comprises a plurality of data being carried out the information of inner marshalling.

Frame comprise have in data decoding information and the entropy decoding information at least one the first data division and have frame information part for the sub-configuration information of the first data division.

Sub-configuration information comprises the second temporal information part that has for the temporal information of all groups.And sub-configuration information also comprises outside marshalling message part, and it has the information of the random groups that comprises a plurality of data being carried out outside marshalling by group.In addition, sub-configuration information also comprises the second inner marshalling message part, and it has the information of the random groups that comprises a plurality of data being carried out inner marshalling.

At last, group comprises the data decoding information that has about the information of data decoding scheme, have entropy decoding information about the information of entropy decoding scheme,, corresponding to the reference value of a plurality of data and have the second data division that uses the difference value that reference value and these data generate.

[to the application of audio coding (MPEG around)]

Below explain the example of unified concept of the present invention and feature.

Figure 24 is the block diagram of the device for audio compression and recovery according to an embodiment of the invention.

With reference to Figure 24, the device for audio compression and recovery according to an embodiment of the invention comprises audio compression parts 105～400 and Audio recovery parts 500～800.

Audio compression parts 105～400 comprise reduction audio mixing unit 105, core decoding unit 200, spatial information decoding unit 300 and multiplexing components 400.

And reduction audio mixing unit 105 comprises that multi-channel audio parts 110 and spatial information generate parts 120.

In reduction audio mixing unit 105, the input of multi-channel audio parts 110 is multichannel X that the N sound channel is arranged ₁, X ₂..., X _N) sound signal and this sound signal.

The output of multi-channel audio parts 110 is that the reduction audio mixing is to the signal of number of channels less than the number of input sound channel.

The output of reduction audio mixing unit 105 is reduced audio mixing and is realized to one or two sound channel, according to the sound channel of the concrete number of independent reduction audio mixing order or according to system the sound channel of the concrete number that presets.

Core decoding unit 200 is in the output of multi-channel audio parts 110---and namely reduce the audio mixing sound signal and carry out core decoding.In this case, realize core decoding in the mode of using the compression inputs such as various conversion scheme such as discrete transform scheme.

Spatial information generates parts 120 and extract spatial information from multi-channel audio signal.Spatial information generates parts 120 and then the spatial information that extracts is transferred to spatial information decoding unit 300.

The spatial information executing data decoding of 300 pairs of inputs of spatial information decoding unit and entropy decoding.Spatial information decoding unit 300 is carried out at least a among PCM, PBC and the DIFF.In some cases, spatial information decoding unit 300 is also carried out entropy decoding.The decoding scheme of spatial information decoding parts 700 can be according to spatial information decoding unit 300 usefulness any data decoding scheme determine.And, after a while with reference to the detailed version space decoding of information of Figure 25 parts 300.

The output of the output of core decoding unit 200 and spatial information decoding unit 300 is transfused to multiplexing components 400.

Multiplexing components 400 is multiplexed into bit stream with these two inputs, then bit stream is transferred to Audio recovery parts 500 to 800.

Audio recovery parts 500 to 800 comprise that multichannel Knock-Down Component 500, core codec parts 600, spatial information decoding parts 700 and multichannel generate parts 800.

Multichannel Knock-Down Component 500 resolves into audio-frequency unit and spatial information part with the bit stream multichannel that receives.In this case, audio-frequency unit is the sound signal of compression, and spatial information partly is the spatial information of compression.

Core codec parts 600 receive compressed sound signal from multichannel Knock-Down Component 500.Core codec parts 600 are by decoding to generate down-mix audio signal to compressed sound signal.

Spatial information decoding parts 700 receive compressed spatial information from multichannel Knock-Down Component 500.Spatial information decoding parts 700 are by decoding to generate spatial information to compressed spatial information.

After doing like this, from the bit stream that receives, extract indication and be included in various marshalling information in the data structure shown in Figure 23 and the identification information of decoding information.From at least one or a plurality of decoding scheme, select concrete decoding scheme according to identification information.And, generate spatial information according to selected decoding scheme by the decoding spatial information.In this case, the decoding scheme of spatial information decoding parts 700 can be determined according to what data decoding scheme of spatial information decoding unit 300 usefulness.And, after a while with reference to the detailed version space information decoding of Figure 26 parts 700.

Multichannel generates parts 800 and receives the output of core decoding unit 600 and the output of spatial information decoding parts 700.Multichannel generates parts 800 and generates the multichannel Y that N sound channel arranged by two outputs that receive ₁, Y ₂..., Y _NSound signal.

Simultaneously, audio compression parts 105～400 provide indication spatial information decoding unit 300 to use the identifier of what decoding scheme to Audio recovery parts 500～800.Get ready for the situation to above explanation, Audio recovery parts 500～800 comprise the device for the grammatical analysis identification information.

So 700 references of spatial information decoding parts are determined decoding scheme by the identification information that audio compression parts 105～400 provide.Preferably, being used for grammatical analysis indicates the device of the identification information of decoding scheme to be provided for spatial information decoding unit 700.

Figure 25 is the detailed diagram of spatial information addressable part according to an embodiment of the invention, wherein spatial information is called spatial parameter.

With reference to Figure 25, decoding unit according to an embodiment of the invention comprises PCM decoding unit 310, DIFF (differential decoding) parts 320 and Huffman decoding unit 330.Huffman decoding unit 330 is corresponding to an embodiment who carries out above-mentioned entropy decoding.

PCM decoding unit 310 comprises marshalling PCM decoding unit 311 and PCB parts 312.311 pairs of spatial parameters of marshalling PCM decoding unit carry out PCM decoding.In some cases, marshalling PCM decoding unit 311 can carry out PCM decoding to spatial parameter by group part (group part).And 312 pairs of spatial parameters of PBC parts are carried out above-mentioned PBC.

320 pairs of spatial parameters of DIFF parts are carried out above-mentioned DIFF.

Particularly, in the present invention, a decoding that is used for selectively spatial parameter in marshalling PCM decoding unit 311, PBC parts 312 and the DIFF parts 320.And, its control device is not shown separately in the accompanying drawings.

In above description, explained in detail the PBC that is carried out by PBC parts 312, so will omit its explanation in the following description.

As another example of PBC, PBC carries out once at spatial parameter.And PBC also can carry out N time (N＞1) in the result of a PBC.Particularly, implementing at least one times PBC as the result's who carries out a PBC guiding value and difference value.In some cases, preferably from the 2nd PBC only on difference value and do not implement PBC in the guiding value.

DIFF parts 320 comprise to be carried out the DIFF_FREQ decoding unit 321 of DIFF_FREQ and spatial parameter is carried out the DIFF_TIME decoding unit 322 and 323 of DIFF_TIME spatial parameter.

In DIFF parts 320, an enforcement of selecting from the group that the DIFF_TIME decoding unit 322 and 323 by DIFF_FREQ decoding unit 321 and DIFF_TIME forms is to the processing of input space parameter.

In this case, the DIFF_TIME decoding unit can be divided into spatial parameter is carried out the DIFF_TIME_FORWARD parts 322 of DIFF_TIME_FORWARD and spatial parameter carried out the DIFF_TIME_BACKWARD parts 323 of DIFF_TIME_BACKWARD.

In DIFF_TIME decoding unit 322 and 323, the spatial parameter implementation data decoding to input of selecting in DIFF_TIME_FORWARD parts 322 and the DIFF_TIME_BACKWARD parts 323 is processed.In addition, in above description, described the DIFF decoding by each execution in the inner member 321,322 and 323 of DIFF parts 320 in detail, so will omit its explanation in the following description.

In the output of 330 pairs of PBC parts 312 of Huffman decoding unit and the output of DIF parts 320 at least one carried out Huffman decoding.

Huffman decoding unit 330 comprises 1 dimension Huffman decoding unit (hereinafter referred to as the HUFF_1D parts) 331 of the data of processing seriatim to be decoded and transmission and is 2 dimension HUFF decoding units (hereinafter referred to as the HUFF_2D parts) 332 and 333 of the data of the to be decoded and transmission of processed in units by two data splittings.

Huffman decoding is carried out in input processed for one that selects in HUFF_1D parts 331 in Huffman decoding unit 330 and HUFF_2D parts 332 and 333.

In this case, HUFF_2D parts 332 and 333 be divided into data that the data that bundle based on frequency are bundled to 2-dimension Huffman decoding unit (hereinafter referred to as the HUFF_2D_FREQ_PAIR parts) 332 with to time-based the frequency of carrying out Huffman decoding to time of carrying out Huffman decoding to 2-dimension Huffman decoding unit (hereinafter referred to as the HUFF_2D_TIME_PAIR parts) 333.

In HUFF_2D parts 332 and 333, Huffman decoding is carried out in input processed for one that from HUFF_2D_FREQ_PAIR parts 332 and HUFF_2D_TIME_PAIR parts 333, selects.

To explain in detail in the following description by each the performed Huffman decoding in the inner member 331,332 and 333 of Huffman decoding unit 330.

Afterwards, the output of Huffman decoding unit 330 is multiplexing so that transmission with the output multi-channel of marshalling PCM decoding unit 311.

In spatial information decoding unit according to the present invention, the various identification information that generates from the decoding of data decoding and entropy is inserted into the transmission bit stream.And transmission bit stream is transferred to spatial information shown in Figure 26 decoding parts.

With reference to Figure 26, spatial information decoding parts receive the transmission bit stream that comprises spatial information, then by the transmission bit stream that receives is decoded to generate spatial information.

Spatial information decoding parts 700 comprise that identifier extracts parts (sign grammatical analysis parts) 710, PCM decoding parts 720, Huffman decoding parts 730 and differential decoding parts 740.

The identifier grammatical analysis parts 710 of spatial information decoding parts extract various identifiers from transmission bit stream, then the identifier that extracts of grammatical analysis.This means the various information of having extracted described in the above description of Figure 23.

Spatial information decoding parts can use the output of identifier grammatical analysis parts 710 and know which kind of decoding scheme spatial parameter has been used, and then determine the decoding scheme corresponding to the decoding scheme that identifies.In addition, the execution of identifier grammatical analysis parts 710 also can be finished by above-mentioned multichannel Knock-Down Component 500.

PCM decoding parts 720 comprise marshalling PCM decoding parts 721 and based on the decoding parts 722 of guiding.

Marshalling PCM decoding parts 721 decode to generate spatial parameter by carrying out PCM in transmission bit stream.In some cases, marshalling PCM decoding parts 721 generate the spatial parameter of a group part by the decoding transmission bit stream.

Decoding parts 722 based on guiding generate spatial parameter value by the decoding that the output of Huffman decoding parts 730 is carried out based on guiding.This is included in the situation in the output of Huffman decoding parts 730 corresponding to the guiding value.As special case, can comprise that based on the decoding parts 722 of guiding a guiding extraction parts (not shown) is directly to extract the guiding value from transmission bit stream.So, use the difference value span parameter value that extracts guiding value that parts extract and consisted of by the output of Huffman decoding parts 730 by guiding.

730 pairs of transmission bit stream of Huffman decoding parts are carried out the Huffman decoding.Huffman decoding parts 730 comprise by carry out one dimension Huffman decode seriatim the one dimension Huffman decoding parts of output data value (hereinafter referred to as HUFF_1D decoding parts) 731 and the two-dimentional Huffman decoding parts (hereinafter referred to as HUFF_2D decoding parts) 732 and 733 by carry out a pair of data value of the each output of two-dimentional Huffman decoding in transmission bit stream in transmission bit stream.

Identifier grammatical analysis parts 710 extract indication Huffman decoding scheme from transmission bit stream be (for example to indicate HUFF_1D or the identifier of indication HUFF_2D, " bsCodingScheme "), the identifier that then extracts by grammatical analysis identifies employed Huffman decoding scheme.So, be confirmed as the Huffman decoding scheme corresponding to HUFF_1D or the HUFF_2D decoding of each situation.

HUFF_1D decoding parts 731 are carried out the HUFF_1D decoding, and

HUFF_2D decoding parts

732 and 733 are carried out the HUFF_2D decoding separately.

Huffman decoding scheme in transmission bit stream is in the situation of HUFF_2D, identifier grammatical analysis parts 710 also extract indication HUFF_2D scheme be HUFF_2D_FREQ_PAIR or the identifier of HUFF_2D_TIME_PAIR (for example, " bsParsing "), then the identifier that extracts is done grammatical analysis.So identifier grammatical analysis parts 710 can be identified and consist of that two right data are based on frequency or the time bundles.And, will corresponding to the frequency of corresponding situation two-dimentional Huffman is decoded (hereinafter referred to as the HUFF_2D_FREQ_PAIR decoding) and the time to two-dimentional Huffman decode one of (decoding referred to as HUFF_2D_TIME_PAIR hereinafter) be defined as the Huffman decoding scheme.

In

HUFF_2D decoding parts

732 and 733, HUFF_2D_FREQ_PAIR parts 732 are carried out the HUFF_2D_FREQ_PAIR decoding, and HUFF_2D_TIME_PAIR parts 733 are carried out the HUFF_2D_FREQ_TIME decoding.

The output of Huffman decoding parts 730 is transferred to decoding parts 722 or the differential decoding parts 740 based on guiding based on the output of identifier grammatical analysis parts 710.

Differential decoding parts 740 generate spatial parameter value by differential decoding is carried out in the output of Huffman decoding parts 730.

Identifier grammatical analysis parts 710 from transmission bit stream, extract indication DIFF scheme be DIF_FREQ or the identifier of DIF_TIME (for example, " bsDiffType "), then by being carried out grammatical analysis, the identifier that extracts identifies employed DIFF scheme.So one in will decoding corresponding to the DIFF_FREQ of corresponding situation decoding and DIFF_TIME is defined as the differential decoding scheme.

DIFF_FREQ decoding parts 741 are carried out the DIFF_FREQ decoding, and each the execution DIF_TIME decoding in the

DIFF_TIME decoding parts

742 and 743.

In the situation of DIFF_TIME in the DIFF scheme, identifier grammatical analysis parts 710 also from transmission bit stream, extract indication DIFF_TIME be DIFF_TIME_FORWARD or the identifier of DIFF_TIME_BACKWARD (for example, " bsDiffTimeDirection "), then the identifier that extracts is carried out grammatical analysis.

So the output that can identify Huffman decoding parts 730 is difference value between current data and the last data or the difference value between current data and next data.Will be corresponding to the DIFF_TIME_FORWARD decoding of corresponding situation and a kind of DIFF_TIME scheme that is defined as in the DIFF_TIME_BACKWARD decoding.

In

DIFF_TIME decoding parts

742 and 743, DIFF_TIME_FORWARD parts 742 are carried out the DIFF_TIME_FORWARD decoding, and DIFF_TIME_BACKWARD parts 743 are carried out the DIFF_TIME_BACKWARD decoding.

Below explain the process of determining Huffman decoding scheme and data decode scheme based on the output of the identifier grammatical analysis parts 710 in the spatial information decoding parts.

What for example, identifier grammatical analysis parts 710 read that indication uses in the spatial data in decoding is which the first identifier (for example, " bsPCMCoding ") among PCM and the DIFF.

If the first identifier is corresponding to the value of indication PCM, then among indication PCM and the PBC which identifier grammatical analysis parts 710 also read and be used to second identifier (for example, " bsPilotCoding ") of the decoding of spatial parameter.

If the second identifier is corresponding to the value of indication PBC, then spatial information decoding parts are carried out the decoding corresponding to PBC.

If the second identifier is corresponding to the value of indication PCM, then spatial information decoding parts are carried out the decoding corresponding to PCM.

On the other hand, if the first identifier corresponding to the value of indication DIFF, then spatial information decoding parts are carried out corresponding to the decoding of DIFF and are processed.

The pattern that is used for invention

Therefore, the embodiment in conjunction with above-mentioned optimal mode has explained various embodiment of the present invention.

Industrial applicibility

Therefore, marshalling of the present invention, data decoding and entropy decoding are applicable to various fields and product.

And the present invention has been applicable to store the medium of having used the data of at least one feature of the present invention thereon.

Can make various modifications and variations and not break away from the spirit or scope of the present invention the present invention although described with reference to preferred embodiment of the present invention and illustrated that the present invention, those skilled in that art are appreciated that.Therefore, the present invention is intended to contain interior all such modifications of the present invention and the variation of scope of appended claims and equivalent technique scheme thereof.

Claims

1. the method for an audio signal may further comprise the steps:

Extract first frequency marshalling information, described first frequency marshalling information is identified the parameter band number in the frame;

Extract second frequency marshalling information, described second frequency marshalling information indication bundlees the span of described parameter band according to a frequency ranges of data, wherein provide parameter value by frequency ranges of data, and the frequency ranges of data that generates by the second frequency marshalling is divided into low-frequency data band and high-frequency data band;

Extract sound channel marshalling information, described sound channel marshalling information identification is applied to the parameter band number of low-frequency channel LFE pattern;

Extract the guiding reference value and the guiding difference value that are associated with the LFE pattern, wherein said guiding reference value is that low-frequency band generates, and described guiding difference value is by poor calculating the between described guiding reference value and the described parameter value; And

Utilize described guiding reference value and described guiding difference value to obtain described parameter value according to following formula:

x[n]＝d[n]+P，

X[n wherein] be described parameter value, d[n] be described guiding difference value, P is the guiding reference value.

2. the method for claim 1 is characterized in that, described first frequency marshalling information and described sound channel marshalling information are to extract from the head zone of described signal.

3. the method for claim 1 characterized by further comprising:

Reception is corresponding to the compressing audio signal of monophonic signal or stereophonic signal;

To described compressing audio signal decoding, to generate the multi-channel signal;

Generate multi-channel signal by described parameter value being put on described multi-channel signal,

Wherein, described parameter value comprises at least one in levels of channels poor (CLD), inter-channel coherence (ICC), sound channel predictive coefficient (CPC) and the arbitrariness multi-channel audio gain (ADG).

4. device for the treatment of sound signal comprises:

Spatial information decoding parts, it is configured to:

Extract first frequency marshalling information, first frequency marshalling information is identified the parameter band number in the frame;

Extract second frequency marshalling information, the indication of second frequency marshalling information bundlees the span of described parameter band according to a frequency ranges of data, wherein provide parameter value by frequency ranges of data, and the frequency ranges of data that generates by the second frequency marshalling is divided into low-frequency data band and high-frequency data band;

Extract the guiding reference value and the guiding difference value that are associated with described LFE pattern, wherein said guiding reference value is that low-frequency band generates, and described guiding difference value is by poor calculating the between described guiding reference value and the described parameter value; And

Utilize described guiding reference value and described guiding difference value according to the following formula value of getting parms:

x[n]＝d[n]+P，

5. device as claimed in claim 4 is characterized in that, described first frequency marshalling information and described sound channel marshalling information are to extract from the head zone of described signal.

6. device as claimed in claim 4 is characterized in that, described device also comprises:

The core codec parts decode to generate the multi-channel signal to compressing audio signal; And

Multichannel generates parts, generates multi-channel signal by the multi-channel signal that described parameter value is put on described decoding,