CN101582259B - Methods, devices and systems for coding and decoding dimensional sound signal - Google Patents

Abstract

The invention discloses methods, devices and systems for coding and decoding a dimensional sound signal. The method for coding a dimensional sound signal comprises the following steps: acquiring a frequency spectrum of a sound channel of the dimensional sound signal in a modification discrete cosine transform (MDCT) field, dividing the sound channel of the dimensional sound signal in the MDCT field into at least two mutually spaced sub-frequency spectrums, and carrying out coding processing of the dimensional sound signal in the MDCT field to the sub-frequency spectrums of the sound channel ofthe divided dimensional sound signal. The method for decoding a dimensional sound signal comprises the following steps: decoding the frequency spectrum vector of a coded dimensional sound signal so a s to obtain the frequency spectrum of the dimensional sound signal in the MDCT field, dividing the frequency spectrum of the dimensional sound signal in the MDCT field into at least two mutually spaced sub-frequency spectrums, and carrying out decoding processing of the dimensional sound signal in the MDCT field on the sub-frequency spectrums of the divided dimensional sound signal. The technical scheme can avoid phenomena, such as envelope distortion, and the like to a certain degree and improves the user experience to the tone of dimensional sound.

Description

Stereophonic signal decoding method, device and coding/decoding system

Technical field

The present invention relates to the network communications technology field, be specifically related to a kind of stereophonic signal coding method, stereophonic signal coding/decoding method, code device, decoding device, coding/decoding system.

Background technology

At present, stereo coding just develops towards the direction of parameter coding technology.The parameter coding technology mainly comprises: Parametric Stereo (parameter stereo; PS) technology and Principle Component Analysis are (based on principal component analysis (PCA); PCA) or Karhunen-Loeve Transform (KL conversion, KLT) sterophonic technique.

The ultimate principle of above-mentioned two kinds of technology is: stereo information and audio-frequency information that stereophonic signal comprised are distinguished, stereo information is encoded with parameter mode, and utilize traditional voice or music encoding device that audio-frequency information is encoded.Parameter stereo technology and PCA/KLT sterophonic technique all are to encode as parameter with energy ratio that extracts and rotation angle etc.Because under normal conditions; Audio-frequency information occur with the form of mixing signal under the monophony and the code check of stereo information parameter coding far below the code check of a complete sound channel signal; Therefore, with respect to the conventional stereo vocoding technique, the code check of parameter coding technical scheme is low.

In the parameter stereo technology; Need divide subband to extract stereo information at frequency domain; Stereo information such as Inter-channel Phase Difference (phase differential; IPD), Inter-channel Level Difference (intensity difference, ILD) and Inter-channel Coherence (degree of correlation, IC) etc.Audio-frequency information shows as evenly mixing down of left and right acoustic channels or weighting mixes the monophonic signal that obtains down.The parameter stereo mode itself does not rely on down the used specific coding method in mixing sound road, and this method can combine existing monophony speech coder or transform coder, forms low code check stereophonic encoder.

In the PCA/KLT sterophonic technique; Can stereophonic signal be resolved into independently two paths of signals through the KL conversion; The road signal that wherein energy is big is called as Principle Component (main composition), and what energy was little a road is called as Ambient Component (atmosphere composition).The rotation angle parameter, energy that stereo information comprises the KL conversion is than parameter and the atmosphere composition energy ratio with main composition.The PCA/KLT sterophonic technique has and code check like the parameter stereo technology type.

In realizing process of the present invention; The inventor finds: the parameter coding technology relates to all that (revise discrete cosine transform, MDCT) each sound channel signal of territory stereo sound is handled, and in processing procedure, all need use the frequency spectrum of each sound channel at ModifiedDiscrete Cosine Transform; For example; Utilize the frequency spectrum of each sound channel to extract the energy ratio, more for example, utilize the frequency spectrum of each sound channel to be rotated conversion etc.Realize in the encoding-decoding process that at the frequency spectrum that utilizes each sound channel the energy that tends to occur envelope distortion, extraction is than phenomenon such as inaccurate.

Summary of the invention

Embodiment of the present invention provides stereophonic signal decoding method, device and coding/decoding system, can avoid envelope distortion, energy than phenomenon such as inaccurate to a certain extent, thus the experience that has improved user's stereo matter.

The stereophonic signal coding method that embodiment of the present invention provides comprises:

Obtain the frequency spectrum of the sound channel of the stereophonic signal of stereophonic signal in revising discrete cosine transform MDCT territory;

With the spectrum division of the sound channel of the said MDCT territory neutral body acoustical signal of obtaining is at least two sub-frequency spectrums of spectral line space;

The sub-frequency spectrum of the sound channel of the stereophonic signal after dividing is carried out MDCT territory stereophonic signal encoding process.

The stereophonic signal coding/decoding method that embodiment of the present invention provides comprises:

Stereophonic signal behind the coding is carried out the frequency spectrum vector decode to obtain the frequency spectrum of the stereophonic signal of stereophonic signal in the MDCT territory;

With the spectrum division of the said MDCT territory neutral body acoustical signal of obtaining is at least two sub-frequency spectrums of spectral line space;

The sub-frequency spectrum of the stereophonic signal after dividing is carried out MDCT territory stereophonic signal decoding processing.

The code device that embodiment of the present invention provides comprises:

First acquisition module is used for obtaining the frequency spectrum of stereophonic signal in the sound channel of the stereophonic signal of revising discrete cosine transform MDCT territory;

First divides module, and the spectrum division of the sound channel of the MDCT territory neutral body acoustical signal that is used for said first acquisition module is obtained is at least two sub-frequency spectrums of spectral line space;

The encoding process module is used for the sub-frequency spectrum of the said first division Module Division is carried out MDCT territory stereophonic signal encoding process.

The decoding device that embodiment of the present invention provides comprises:

Second acquisition module is used for the stereophonic signal behind the coding is carried out the frequency spectrum vector decode to obtain the frequency spectrum of stereophonic signal at the stereophonic signal of revising discrete cosine transform MDCT territory;

Second divides module, and the spectrum division of the MDCT territory neutral body acoustical signal that is used for said second acquisition module is obtained is at least two sub-frequency spectrums of spectral line space;

Decoding process module, be used for said second divide the stereophonic signal of Module Division sub-frequency spectrum carry out MDCT territory stereophonic signal decoding processing.

The coding/decoding system that embodiment of the present invention provides comprises above-mentioned code device and decoding device.

Description through technique scheme can be known; Method in the embodiment of the present invention, device and coding/decoding system are at least two sub-frequency spectrums of spectral line space through the spectrum division with MDCT territory neutral body acoustical signal; And utilize sub-frequency spectrum to realize encoding and decoding; Can avoid envelope distortion, energy than phenomenon such as inaccurate to a certain extent, thus the experience that has improved user's stereo matter.

Description of drawings

Fig. 1 is the odd even spectral decomposition synoptic diagram of embodiment of the present invention;

Fig. 2 is that the odd even frequency spectrum of embodiment of the present invention merges synoptic diagram;

Fig. 3 (a) is the original sound waveform synoptic diagram of the L channel of embodiment of the present invention;

Fig. 3 (b) is the L channel sound waveform synoptic diagram of not distinguishing the odd even frequency spectrum of prior art;

Fig. 3 (c) is the L channel sound waveform synoptic diagram of the employing odd even frequency spectrum of embodiment of the present invention;

Fig. 4 is the stereophonic signal coding synoptic diagram of embodiment of the present invention;

Fig. 5 is the stereophonic signal decoding synoptic diagram of embodiment of the present invention.

Embodiment

Embodiment of the present invention is the encoding and decoding that at least two sub-frequency spectrums of spectral line space and the sub-frequency spectrum that utilizes stereophonic signal are realized stereophonic signal through the spectrum division with MDCT territory neutral body acoustical signal, and the energy that can avoid envelope distortion, extraction to a certain extent is than phenomenon such as inaccurate.

The stereophonic signal coding method that at first embodiment of the present invention is provided below describes.

Steps A, embodiment of the present invention need be obtained the frequency spectrum of the sound channel of MDCT territory neutral body acoustical signal.Embodiment of the present invention can adopt existing several different methods to obtain the frequency spectrum of the sound channel of MDCT territory neutral body acoustical signal, for example, the stereophonic signal on the time domain transformed in the MDCT territory, thus the frequency spectrum of the sound channel of acquisition MDCT territory neutral body acoustical signal.Embodiment of the present invention can adopt existing multiple mapping mode that the stereophonic signal on the time domain is transformed in the MDCT territory; For example; Can each sound channel signal on the time domain all be divided into isometric frame; Can include overlapped signal between the consecutive frame of a sound channel; The quantity of overlapped signal can be confirmed according to the actual needs of encoding and decoding between the consecutive frame, then, each frame of each sound channel on the time domain is all transformed to the MDCT territory.Embodiment of the present invention does not limit the stereophonic signal on the time domain is transformed to the concrete implementation procedure in the MDCT territory, does not limit the concrete implementation procedure of the frequency spectrum of the sound channel of obtaining MDCT territory neutral body acoustical signal yet.

When stereo be about during two sound channels stereo, embodiment of the present invention can obtain MDCT territory neutral body sound about the frequency spectrum of two sound channels.When being more multichannel (like triple-track, the quadraphonic, five-sound channel etc.) stereo, embodiment of the present invention can be obtained the frequency spectrum of each sound channel of MDCT territory neutral body sound, also can obtain the frequency spectrum of a plurality of sound channels of part of MDCT territory neutral body sound when stereo.

Step B, behind the frequency spectrum of the sound channel that has obtained MDCT territory neutral body acoustical signal; Need the spectrum division of the sound channel of the stereophonic signal in the MDCT territory that obtains be at least two sub-frequency spectrums; To a sound channel, be exactly that spectrum division with this sound channel is at least two sub-frequency spectrums.If obtained the frequency spectrum of a plurality of sound channels of MDCT territory neutral body sound, then to each sound channel, all the spectrum division with this sound channel is at least two sub-frequency spectrums.

Spectral line between a plurality of sub-frequency spectrum of a sound channel is the space, and the spectral line between the promptly sub-frequency spectrum interleaves mutually.For example; In the spectrum division with a sound channel is under the situation of two sub-frequency spectrums; If the spectral line of this sound channel is sorted and the spectral line after the ordering is carried out the order label according to the size of frequency; Then one of them sub-frequency spectrum can comprise that label is 0,2, the even number spectral line of 4......, and another sub-frequency spectrum can comprise that label is 1,3, the odd number spectral line of 5.......In this case, two sub-frequency spectrums that mark off can be called strange frequency spectrum and even frequency spectrum, perhaps can be called strange sub-frequency spectrum and the sub-frequency spectrum of idol.Again for example; In the spectrum division with a sound channel is under the situation of three sub-frequency spectrums; If the spectral line of this sound channel is sorted and the spectral line after the ordering is carried out the order label according to the size of frequency; Then sub-frequency spectrum can comprise that label is 0,3, the spectral line of 6......, and sub-frequency spectrum can comprise that label is 1,4, the spectral line of 7......, and sub-frequency spectrum can comprise that label is 2,5, the spectral line of 8.......In the foregoing description, according to the size of frequency the spectral line of sound channel being sorted can be for sorting according to frequency order from small to large, also can be for sorting according to frequency order from big to small.In addition, in the description of embodiment of the present invention, with numeral spectral line being carried out label is that example describes, and embodiment of the present invention is not got rid of the employing alternate manner carries out label to spectral line possibility.

Step C, after having obtained the sub-frequency spectrum of sound channel, the next code that can utilize the sub-frequency spectrum of sound channel to carry out stereophonic signal is handled.The next code that embodiment of the present invention can utilize the sub-frequency spectrum of sound channel to adopt existing multiple mode to carry out stereophonic signal is handled, and for example, can adopt sub-frequency spectrum to extract the energy ratio of left and right acoustic channels; Again for example, can adopt sub-frequency spectrum to be rotated conversion etc.In existing stereo coding technical scheme; The use of the frequency spectrum of the sound channel of the stereophonic signal in MDCT territory is diversified; Promptly not only be used to extract energy ratio, rotational transform etc., every when needs use the frequency spectrum of sound channel of stereophonic signal in MDCT territory, all can adopt the sub-frequency spectrum in the embodiment of the present invention; That is to say, can each sub-frequency spectrum of a sound channel all be used as the frequency spectrum of this sound channel.Though be that to be example with the rotational transform be elaborated to the encoding-decoding process of embodiment of the present invention in the description of following embodiment,, embodiment of the present invention does not limit the concrete application mode of sub-frequency spectrum in encoding-decoding process of sound channel.In addition; Embodiment of the present invention can adopt existing several different methods to extract the energy ratio fully; Just energy than leaching process in, with the frequency spectrum of sub-frequency spectrum as the sound channel of stereophonic signal, embodiment of the present invention does not limit the concrete implementation procedure of extracting the energy ratio.

If adopt sub-frequency spectrum to be rotated conversion, then can carry out the quadrature rotational transform respectively to the sub-frequency spectrum of each sound channel.Can be unit with the subband in quadrature rotational transform process carries out the quadrature rotational transform to the sub-frequency spectrum of each sound channel.Promptly can be earlier be a plurality of subbands with the sub-spectrum division of each sound channel, then, the sub-frequency spectrum to each subband carries out the quadrature rotational transform again.Certainly, the partition process of subband also can be given the partition process of sub-frequency spectrum earlier, promptly can be earlier be a plurality of subbands with the spectrum division of the sound channel of stereophonic signal; Again to each subband; Spectrum division is at least two sub-frequency spectrums, then, carries out the quadrature rotational transform at sub-frequency spectrum to each subband.Embodiment of the present invention does not limit the sequencing of sub-band division and sub-spectrum division.

Embodiment of the present invention can adopt existing multiple sub-band division method, and for example, the length of each subband of division can identical, also can be inequality.Embodiment of the present invention does not limit the concrete implementation procedure of sub-band division.

Be rotated in the conversion process at the antithetical phrase frequency spectrum, can adopt the mode of greatly relevant rotation angle to come the antithetical phrase frequency spectrum to carry out the quadrature rotational transform, certainly, also can adopt existing other rotational transform mode to come the antithetical phrase frequency spectrum to be rotated conversion.After rotational transform; Sub-frequency spectrum after the rotational transform of the same timeslice of different sound channels is closed the road to be handled; For example, carry out mixing, the even frequency spectrum of the left and right acoustic channels after the rotational transform is carried out mixing according to the strange frequency spectrum of the label of the spectral line left and right acoustic channels after with rotational transform.Then, interleave according to the sub-frequency spectrum of spectral line label after and to merge into one road signal, merge into one road signal as the strange frequency spectrum after the mixing and the even frequency spectrum after the mixing are interleave with mixing.Between the antithetical phrase frequency spectrum carries out, plug together after the road handles, can be the MDCT frequency-region signal with the conversion of signals of closing behind the road, and carry out encoding process at the MDCT frequency domain.In the encoding process of MDCT frequency domain as quantizing entropy coding etc.Embodiment of the present invention does not limit the concrete implementation procedure of quadrature rotational transform, is not limited in the concrete implementation procedure that the MDCT frequency domain carries out encoding process yet.

Stereo, the sub-frequency spectrum that with stereo is left and right acoustic channels below is based on the example that is applied as in the quadrature rotational transform process of greatly relevant rotation angle, and the stereophonic signal coding method that embodiment of the present invention is provided describes.

Being utilized in has overlapping windowing on the time domain, make stereophonic signal be formed on last front and back of time signal frame in succession.The signal frame here comprises the signal frame of left and right acoustic channels, promptly to each sound channel in the left and right acoustic channels, signal frame in time before and after in succession.Then, will all transform to the MDCT territory corresponding to the signal frame of the left and right acoustic channels of same time slice.Afterwards, the spectral line of the frequency spectrum in the MDCT territory of each sound channel according to frequency series arrangement from low to high, and 0 being beginning label, is carried out label to the spectral line of the MDCT territory frequency spectrum of tactic each sound channel.

After the spectral line to frequency spectrum carries out label, the MDCT territory frequency spectrum of left and right acoustic channels is divided into two isometric sub-frequency spectrums respectively, promptly strange frequency spectrum and even frequency spectrum according to the odd even of each spectral line label.The frequency spectrum in MDCT territory is divided into strange frequency spectrum and even frequency spectrum shown in accompanying drawing 1.

Among Fig. 1, the frequency spectrum in the MDCT territory of a sound channel is divided into strange frequency spectrum and even frequency spectrum.

In being divided into odd even frequency spectrum process, can with the subband partition process that unit carries out the odd even frequency spectrum, promptly elder generation is a plurality of subbands with the spectrum division of the left and right acoustic channels in the MDCT territory, is the odd even frequency spectrum with each sub-band division again.Certainly, the process that is divided into subband also can be carried out after odd even spectrum division process.

The division of subband can reference man's ear non-linear sense of hearing resolution, and parameter such as sampling rate.For example, when 48kHz,, then can divide subband by the mode of table 1 if the dimension of MDCT spectral line is 1024 (being that the MDCT frequency spectrum has 1024 spectral lines):

Table 1

b	ib	b	ib
				0	0	13	74
1	4	14	84
				2	8	15	96
3	12	16	110
				4	16	17	126

5	20	18	145
				6	24	19	172
7	29	20	210
				8	35	21	256
9	41	22	318
				10	48	23	400
11	56	24	512
				12	64

In the table 1, b representes the sequence number of subband, i _bThe start point information of the MDCT frequency spectrum spectral line of expression corresponding subband, start point information is to represent with the label of spectral line, the spectral line that certain subband comprises arrives the spectral line of the starting point of next subband for the corresponding starting point of this subband.The MDCT frequency spectrum that comprises 1024 spectral lines in the table 1 is divided into 25 subbands, the label of subband from 0 to 24.It is 0 to 3 spectral line that the 0th subband comprises label, and it is 4 to 7 spectral line that the 1st subband comprises label ..., and the like.The quantity of the spectral line that each subband comprises in the table 1 is also incomplete same, and still, the quantity of the spectral line that each subband comprises also is can be identical.In addition, to different sampling rates, can adopt different sub-band division modes, and can adopt according to code check antithetical phrase band and merge or processing such as decomposition, above-mentioned table 1 only is a concrete example.

After carrying out sub-band division, if the subband spectrum of point from 2N to N in a sound channel MDCT territory can be expressed as according to frequency order from low to high: X ₀..., X _i..., X _N-1(wherein N is an even number), then even frequency spectrum can be expressed as: X ₀..., X _2i..., X _{2 (N/2-1)}, strange frequency spectrum can be expressed as: X ₁.., X _2i+1..., X _{2 (N/2-1)+1}Need to prove X ₀..., X _2i..., X _{2 (N/2-1)}Can be called even spectral line, X ₁..., X _2i+1..., X _{2 (N/2-1)+1}Can be called strange spectral line.

For b subband, if the spectral line label of the corresponding frequency spectrum of this subband is i _b..., i _B+1-1, then the even spectral line of left and right acoustic channels (being even frequency spectrum) is respectively: $X_e^b=(X_{{2i}_b},...,X_{2(i_{b+1}-1)}),$ And $Y_e^b=(Y_{{2i}_b},...,Y_{2(i_{b+1}-1)});$ The strange spectral line of left and right acoustic channels (being strange frequency spectrum) is respectively: $X_o^b=(X_{2i_b+1},...,X_{2(i_{b+1}-1)+1}),$ And

$Y_o^b=(Y_{{2i}_b+1},...,Y_{2(i_{b+1}-1)+1}).$

To the odd even frequency spectrum of each subband of left and right acoustic channels, calculate greatly relevant rotation angle respectively, promptly calculate the greatly relevant rotation angle and the strange very big relevant rotation angle of frequency spectrum of the even frequency spectrum of each subband.Afterwards, be unit with the subband, according to the greatly relevant rotation angle of even frequency spectrum the even frequency spectrum of left and right acoustic channels is carried out the quadrature rotational transform, according to the greatly relevant rotation angle of strange frequency spectrum the strange frequency spectrum of left and right acoustic channels carried out the quadrature rotational transform.

So-called greatly relevant rotation angle just is meant: regard the same subband signal X and the Y of left and right acoustic channels as arrange by frequency subordinate ordered array from low to high, the formula (1) below utilizing carries out orthogonal transformation:

$\left(\begin{array}{c}{X}^{\′}\\ Y^{\′}\end{array}\right)=\left(\begin{array}{cc}\cos \mathrm{\θ}& \sin \mathrm{\θ}\\ -\sin \mathrm{\θ}& \cos \mathrm{\θ}\end{array}\right)\left(\begin{array}{c}X\\ Y\end{array}\right),---\left(1\right)$

After orthogonal transformation, make X ' and Y ' that the angle θ of maximal correlation degree arranged.

In above-mentioned formula (1), angle θ is by the spatial relationship decision of X and Y, that is:

$\mathrm{\&theta;}=\left\{\begin{array}{cc}{\mathrm{\&theta;}}_0,& \&lang;X,Y\&rang;\&GreaterEqual;0\\ {\mathrm{\&theta;}}_0-\mathrm{\&pi;}/2,& \&lang;X,Y\&rang;<0,{\mathrm{\&theta;}}_0\&GreaterEqual;0\\ {\mathrm{\&theta;}}_0+\mathrm{\&pi;}/2,& \&lang;X,Y\&rang;<0,{\mathrm{\&theta;}}_0<0\end{array}\right.,---\left(2\right)$

Wherein, in the formula (2) ${\mathrm{\&theta;}}_0=\frac{1}{2}\mathrm{Arctan}\frac{\&lang;Y,Y\&rang;-\&lang;X,X\&rang;}{2\&lang;X,Y\&rang;}---\left(3\right)$

The very big relevant rotation angle of the even frequency spectrum that calculates through formula (2) and formula (3) can be expressed as θ _e ^bThe very big relevant rotation angle of the strange frequency spectrum that calculates through formula (2) and formula (3) can be expressed as θ _o ^bThe θ that calculates through formula (2) and formula (3) can make the maximal correlation degree that has between signal X ' and the Y ' under the quadrature rotation, and can make ‖ X ' ‖=‖ Y ' ‖.

If the even frequency spectrum of the subband of the same timeslice of left and right acoustic channels is expressed as respectively: X _e ^bAnd Y _e ^b, the greatly relevant rotation angle of the even frequency spectrum of this subband is expressed as: θ _e ^b, then according to formula (1) with θ _e ^bAfter carrying out the quadrature rotational transform, the subband signal that obtains can be expressed as X _e ^b' and Y _e ^b'.

If the strange frequency spectrum of the subband of the same timeslice of left and right acoustic channels is expressed as respectively: X _o ^bAnd Y _o ^b, the greatly relevant rotation angle of the strange frequency spectrum of this subband is expressed as: θ _o ^b, then according to formula (1) with θ _o ^bAfter carrying out the quadrature rotational transform, the subband signal that obtains can be expressed as X _o ^b' and Y _o ^b'.

The even frequency spectrum of the left and right acoustic channels after the rotational transform is carried out under the idol mixing spectrum handles, for example, can according to $M_e^b=({X_e^b}^{\&prime;}+{Y_e^b}^{\&prime;})/2$ Carry out mixing spectrum processing under the idol.

Carry out very descending the mixing spectrum to handle the strange frequency spectrum of the left and right acoustic channels after the rotational transform, for example, can according to $M_o^b=({X_o^b}^{\&prime;}+{Y_o^b}^{\&prime;})/2$ Carry out very descending the mixing spectrum to handle.

Afterwards, with mixing spectrum under the idol of left and right acoustic channels, very descend that frequency spectrum after the mixing spectrum is handled is alternate slottingly to be merged into one road signal, for example, according to frequency order from high to low with M _e ^bAnd M _o ^bInterleave and be merged into one road signal M.

Insert alternate with even frequency spectrum of strange frequency spectrum in MDCT territory is merged into one road signal shown in accompanying drawing 2.

Among Fig. 2, the strange frequency spectrum of left and right acoustic channels, evenly interleave according to frequency order from high to low with the even frequency spectrum of left and right acoustic channels and to merge into one road signal.

At last, the left and right acoustic channels two paths of signals is evenly down blended together one road MDCT frequency-region signal, and carry out the MDCT frequency domain and handle, quantize operations such as entropy coding processing, the coded data in mixing sound road under obtaining.The process of signal being carried out the processing of MDCT frequency domain, quantification entropy coding can be identical with the processing procedure in the existing transform coder; For example; The road MDCT frequency-region signal that evenly mixes is down carried out Advanced Audio Coding, and (Advanced Audio Coding AAC) is handled.AAC handles can comprise the processing of MDCT frequency domain, reach the quantification and the entropy coding that are undertaken by psychologic acoustics.

Coded data after the coding side encoding process can transfer to decoding end.In addition, need to prove that stereo information also needs coding transmission to decoding end.For example, the greatly relevant rotation angle of the odd even frequency spectrum of subband is carried out vector coding, and behind vector coding, transfer to decoding end.Embodiment of the present invention can adopt existing multiple mode stereo acoustic intelligence (like greatly relevant rotation angle) to encode, and embodiment of the present invention does not limit the concrete implementation procedure that greatly relevant rotation angle is encoded.

After the stereophonic signal coding method of having described embodiment of the present invention, the stereophonic signal coding/decoding method in the face of embodiment of the present invention describes down.

The decode procedure of stereophonic signal is the reciprocal process of cataloged procedure.

At first, decoding end is obtained the frequency spectrum of MDCT territory neutral body sound from the signal that coding side sends, and for example, decoding end is carried out the processing of spectral line vector decode to the received signal, thereby obtains the frequency spectrum of MDCT territory neutral body sound.Certainly, can adopt other existing method to obtain the frequency spectrum of MDCT territory neutral body sound, embodiment of the present invention does not limit the concrete implementation procedure of the frequency spectrum that obtains MDCT territory neutral body sound yet.

Then, decoding end is at least two sub-frequency spectrums of spectral line space with the spectrum division of the MDCT territory neutral body sound that obtains.For example, can the frequency spectrum of the MDCT territory neutral body sound of above-mentioned acquisition be arranged according to the frequency size order, and the order label, be divided into the sub-frequency spectrum that a plurality of spectral line labels interleave each other again.Decoding end in the embodiment of the present invention can be the odd even frequency spectrum, also can be divided into plural sub-frequency spectrum spectrum division, and the implementation procedure of the sub-frequency spectrum of division of concrete implementation procedure and above-mentioned coding side is basic identical, in this no longer repeat specification.

At last, the sub-frequency spectrum that will divide of decoding end carries out follow-up MDCT territory stereophonic signal decoding processing as this stereosonic frequency spectrum.Decoding end can utilize sub-frequency spectrum to carry out energy than processing such as extraction, rotational transforms equally.In being rotated conversion process, can earlier each the sub-frequency spectrum that marks off being gone up respectively to mix and handle, to obtain the sub-frequency spectrum of stereo each sound channel; Sub-spectrum division with each sound channel is a plurality of subbands then, and to be unit with the subband carry out the contrary rotational transform of quadrature respectively to the sub-frequency spectrum of each sound channel.The contrary rotational transform of the quadrature here can be the contrary rotational transform of the quadrature that utilizes greatly relevant rotation angle to carry out.The rotation angle that the contrary rotational transform of quadrature is adopted can be from the signal that coding side sends, to obtain, and for example, the rotation angle vector coding data that coding side sends is decoded, to obtain the greatly relevant rotation angle of each subband odd even frequency spectrum.Certainly, decoding end also can not carried out the contrary rotational transform of quadrature, but the antithetical phrase frequency spectrum directly quantizes.In addition, decoding end also can be carried out sub-band division before last mixed the processing.Description in the concrete implementation procedure of sub-band division etc. such as the above-mentioned coding embodiment.In this no longer repeat specification.

Decoding end can plug together the road between the sub-frequency spectrum with each sound channel after the contrary rotational transform of quadrature, and each sound channel signal that will close behind the road carries out contrary MDCT conversion, afterwards, contrary signal after the MDCT conversion is carried out the overlap-add processing.Plug together the description in road such as the above-mentioned coding embodiment between decoding end, in this no longer repeat specification.

The above-mentioned process of utilizing the odd even frequency spectrum to encode of giving an example that continues below describes decoding.

Decode procedure comprises the decoding to the decoding of the coded data of signal M and stereo acoustic intelligence, and the decoding of stereo acoustic intelligence is like the greatly relevant rotation angle θ to the odd even frequency spectrum of each subband _e ^bAnd θ _o ^bDecoding.To the decode procedure of the coded data of signal M can for:

Decoding end is decoded to the coded data of the signal M in the MDCT territory, to obtain the signal M in MDCT territory.Signal M is divided into a plurality of sub-frequency spectrums such as even frequency spectrum M in the MDCT territory _eWith strange frequency spectrum M _oThe odd even frequency spectrum is carried out sub-band division.After sub-band division, to subband b, even frequency spectrum is M _e ^b, strange frequency spectrum is M _o ^bEven frequency spectrum and strange frequency spectrum are blended together two groups of signals on respectively, and sound was logical about these two groups of signals were directed against respectively, was about to M _e ^bGo up to mix and be M _e ^b, M _e ^b' and M _e ^b, ${M_e^b}^{\&prime;}=M_e^b,$ With M _o ^bGo up to mix and be M _o ^b, M _o ^b' and M _o ^b, ${M_o^b}^{\&prime;}=M_o^b.$

Decoding end is according to the greatly relevant rotation angle θ of the even frequency spectrum of decoding acquisition _e ^bConfirm-θ _e ^b, the greatly relevant rotation angle θ of the strange frequency spectrum that obtains according to decoding _o ^bConfirm-θ _o ^bAccording to formula (1) and-θ _e ^bTo the signal M after last the mixing _e ^bAnd M _e ^b' carry out the quadrature rotational transform, obtain the even spectrum signal of subband b

With

According to formula (1) and-θ _o ^bTo the signal M after last the mixing _o ^bAnd M _o ^b' carry out the quadrature rotational transform, obtain the even spectrum signal of subband b

With

According to the size order of frequency, with

alternate with

insert be merged into one road signal

according to the size order of frequency, is alternate with

slottingly be merged into one road signal

at last;

carried out Inverse MDCT (contrary MDCT conversion, IMDCT) and operation such as overlap-add obtain the L channel decoded signal;

carried out operations such as IMDCT and overlap-add obtain the R channel decoded signal.

Need to prove, though the above-mentioned description that is directed against decoding only relates to a kind of concrete realization situation,, also there be the multiple variation corresponding with cataloged procedure in the implementation procedure of decoding.Because decoding is the inverse process of coding, therefore, no longer decode procedure is carried out detailed description one by one here to multiple different implementation in the coding.

Can know that from the description of above-mentioned encoding-decoding process the stereophonic signal coding in the embodiment of the present invention belongs to lossy coding, the odd even frequency spectrum rotation parameter θ of each subband _e ^bAnd θ _o ^bThe spatial information of sound source in the expression stereophonic signal, the effect of rotation parameter is similar to IPD, ILD and the IC in the parameter stereo technology.

Because MDCT territory frequency spectrum has following characteristic: strange label part and even label in the secondary lobe component of the MDCT territory frequency spectrum of simple signal partly are the distribution that overlaps, and the consistance of promptly strange frequency spectrum or even frequency spectrum itself is higher than the whole consistance of MDCT territory frequency spectrum; Therefore, under the situation that is not divided into the odd even frequency spectrum, the spectral line consistance is low, and under the situation that is divided into the odd even frequency spectrum, the spectral line consistance of odd even frequency spectrum is high.

Under the low situation of spectral line consistance, can make the quadrature rotary course for example utilize greatly relevant rotation angle to carry out the quadrature rotary course and cause bigger error.With single-frequency, pure tone class signal; The original sound waveform of L channel is shown in Fig. 3 (a); Under the situation of the spectrum division technical scheme that does not adopt embodiment of the present invention; The sound waveform of decoding end L channel is shown in Fig. 3 (b), and under the situation of the odd even spectrum division technical scheme that adopts embodiment of the present invention, the sound waveform of decoding end L channel is shown in Fig. 3 (c).Can observe from Fig. 3 (a) and (b), (c): have apparent in view envelope distortion the sound waveform of Fig. 3 (b), be reflected on the subjective auditory perception, feel that promptly sound is trembling.Do not observe envelope distortion in the sound waveform of Fig. 3 (c), and on subjective auditory perception with the original sound basically identical.Hence one can see that; Embodiment of the present invention is through splitting into the frequency spectrum in the MDCT territory for example odd even frequency spectrum of sub-frequency spectrum; Utilize sub-frequency spectrum to carry out the quadrature rotational transform again and carry out the quadrature rotational transform as utilizing greatly relevant rotation angle; Avoid the envelope distortion phenomenon, improved the experience of user's stereo matter.

Code device below in conjunction with 4 pairs of embodiment of the present invention of accompanying drawing describes.

Code device among Fig. 4 comprises: windowing module 300, MDCT module 310, odd even frequency spectrum separation module 320, the greatly relevant rotation angle module 330 of calculating, quadrature rotary module 340, down mixed module 350, odd even frequency spectrum merge module 360, rotation angle vector coding module 380 and spectral line vector coding module 370.

Windowing module 300 among Fig. 4 is formed first acquisition module 301 with MDCT module 310, i.e. the operation of the frequency spectrum of the sound channel of obtaining the stereophonic signal in the MDCT territory of first acquisition module, 301 execution can be carried out by windowing module 300 and MDCT module 310.

Odd even frequency spectrum separation module 320 among Fig. 4 promptly first is divided module 302; Promptly first divide the sound channel of the MDCT territory neutral body acoustical signal that module 302 obtains first acquisition module spectrum division when being two sub-frequency spectrums of odd even of spectral line space, first divides module 302 can be called odd even frequency spectrum separation module 320.First to divide module 302 can also be the plural sub-frequency spectrum of spectral line space with the spectrum division of the sound channel of MDCT territory neutral body acoustical signal.

Quadrature rotary module 340 among Fig. 4, down mixed module 350, odd even frequency spectrum merge module 360, spectral line vector coding module 370 is formed encoding process modules 303, promptly encoding process module 303 carry out with ready-portioned sub-frequency spectrum as the operation that the frequency spectrum of this sound channel carries out follow-up stereophonic signal encoding process can merge module 360 by quadrature rotary module 340, down mixed module 350, odd even frequency spectrum, spectral line vector coding module 370 formed the encoding process module and carry out.

Windowing module 300 mainly is that the signal with left and right acoustic channels is respectively that unit multiply by sinusoidal windows with the frame, and output.For example; Windowing module 300 is divided into the signal of each sound channel 2048 frames of equal length; And have the overlapping of 1024 points (being sampled point) between the consecutive frame, then, windowing module 300 all multiplies each other every frame signal with window function; Window function should satisfy the complete recondition of MDCT, typical window function such as sinusoidal windows function.

MDCT module 310 is carried out the MDCT conversion with the frame of windowing module 300 outputs, and signal frame is transformed to the MDCT territory, thereby obtains left and right acoustic channels two-way MDCT territory frequency spectrum, and output.

Odd even frequency spectrum separation module 320 is separated into strange frequency spectrum and even frequency spectrum respectively with left and right sides two-way MDCT frequency spectrum, and to quadrature rotary module 340, the greatly relevant rotation angle module 330 output odd even frequency spectrums of calculating.For example; MDCT module 310 can become 2048 time-domain windowed signal transformation 1024 spectral line, then, according to the size order of frequency spectral line is sorted and label; Label to each spectral line can be from 0 to 1023; At last, odd even frequency spectrum separation module 320 all is divided into strange frequency spectrum and even frequency spectrum according to the label of spectral line with 1024 spectral lines of each sound channel, and each strange frequency spectrum, even frequency spectrum all comprise 512 spectral lines.

Calculating greatly relevant rotation angle module 330 is that each odd even frequency spectrum calculates greatly relevant rotation angle respectively, and the greatly relevant rotation angle that will calculate exports rotation angle vector coding module 380 and quadrature rotary module 340 to.Quadrature calculates greatly relevant rotation angle module 330 can carry out sub-band division to each odd even frequency spectrum; And to each subband, utilize the very big related angle of the strange frequency spectrum that the formula (2) described in the above-mentioned embodiment and formula (3) calculate each subband left and right acoustic channels and the very big related angle of even frequency spectrum.

Quadrature rotary module 340 carries out the quadrature rotation transformation based on the greatly relevant anglec of rotation that receives respectively to each strange frequency spectrum and even frequency spectrum, and mixes the odd even frequency spectrum after the module 350 output orthogonal rotation transformations downwards.Quadrature rotary module 340 can carry out the quadrature rotational transform to the odd even frequency spectrum according to the formula of describing in the above-mentioned embodiment (1); And carry out the quadrature rotation according to the very big related angle of the odd even frequency spectrum that calculates, have the quadrature rotation subband signal that equates of maximal correlation degree and energy down thereby can obtain two groups of odd evens.

Down the strange frequency spectrum of the left and right acoustic channels of quadrature rotary module 340 outputs is mixed down is one road signal to mixed module 350, and it is one road signal that the even frequency spectrum of the left and right acoustic channels of quadrature rotary module 340 outputs is mixed down, the two paths of signals after output mixes down.Down mixed module 350 carry out the mode of mixing operation down can be with mix under the average of strange frequency spectrum, with mixing under the average of even frequency spectrum; Thereby obtain the odd even two paths of signals; The corresponding strange frequency spectrum of in this two paths of signals a road, the corresponding even frequency spectrum in another road, and every road signal can all comprise 512 spectral lines.

The odd even frequency spectrum merges module 360 will be descended to mix strange frequency spectrum and even frequency spectrum after following the mixing of module 350 outputs and interleave according to the size order of frequency and be merged into one road spectrum signal, and export MDCT territory, this road signals to spectral line vector coding module 370.If the two paths of signals of down mixed module 350 outputs all comprises 512 spectral lines, then the MDCT territory signal of odd even frequency spectrum merging module 360 outputs should comprise 1024 spectral lines.

Spectral line vector coding module 370 is carried out encoding process with the signal that receives and is generated the spectrum coding data, and spectral line vector coding module 370 can be stored the spectrum coding data, also can send this spectrum coding data to decoding end.For example; Spectral line vector coding module 370 can all resolve into subvector with 1024 spectral lines; And in corresponding vector coding code book, search the vector that matees most with subvector; After finding the vector that matees most, with code word or the spectrum coding data of the vector label that finds as this vector coding.

The greatly relevant rotation angle that rotation angle vector coding module 380 will receive is carried out encoding process; Greatly relevant rotation angle after rotation angle vector coding module 380 can memory encoding be handled, greatly relevant rotation angle that also can be after decoding end is sent this encoding process.For example; Rotation angle vector coding module 380 is formed 2 dimension or 4 dimensional vectors with the greatly relevant rotation angle of the odd even frequency spectrum that receives; And in the vector coding code book that calculates storage in advance, search the vector that matees most with this vector; After finding the vector of coupling, with the label of this vector that finds code word or parameter coding data as this vector coding.

Need to prove that embodiment of the present invention can be provided with windowing module 300, MDCT module 310, odd even frequency spectrum separation module 320 respectively for left and right acoustic channels, calculate very big relevant rotation angle module 330, quadrature rotary module 340 and following mixed module 350; Also can a windowing module 300, MDCT module 310, odd even frequency spectrum separation module 320 only be set, calculate very big relevant rotation angle module 330, quadrature rotary module 340 and following mixed module 350 for left and right acoustic channels.

Decoding device below in conjunction with 5 pairs of embodiment of the present invention of accompanying drawing describes.

Decoding device among Fig. 5 comprises: spectral line vector decode module 400, rotation angle vector decode module 410, odd even frequency spectrum separation module 320, upward mix module 420, the contrary rotary module 430 of quadrature, odd even frequency spectrum and merge module 440, IMDCT module 450 and overlap-add module 460.

Spectral line vector decode module 400 among Fig. 5 is second acquisition module 401; Promptly when second acquisition module 401 obtained the frequency spectrum of MDCT territory neutral body sound through the mode of the next signal of coding side transmission being carried out the frequency spectrum vector decode, second acquisition module 401 can be called spectral line vector decode module 400.

Odd even frequency spectrum separation module 320 among Fig. 5 promptly second is divided module 402; Promptly second divide the MDCT territory neutral body acoustical signal that module 402 obtains second acquisition module 401 spectrum division when being two sub-frequency spectrums of odd even of spectral line space, second divides module 402 can be called odd even frequency spectrum separation module 320.The second division module 402 can also be the plural sub-frequency spectrum of spectral line space with the spectrum division of MDCT territory neutral body acoustical signal.Concrete like the description in the above-mentioned method embodiment.

The contrary rotary module 430 of quadrature among Fig. 5, upward mix module 420, the odd even frequency spectrum merges module 440, IMDCT module 450 and overlap-add module 460 and forms decoding process modules 403, promptly decoding process module 403 carry out with ready-portioned sub-frequency spectrum as the frequency spectrum of stereophonic signal carry out the operation of follow-up stereophonic signal decoding processing can be by the contrary rotary module 430 of quadrature, upward mix module 420, the odd even frequency spectrum merges module 440, IMDCT module 450 and overlap-add module 460 and carries out.

Spectral line vector decode module 400 is transmitted mixing spectrum signal under the next signal reconstruction according to coding side, and sends the following mixing spectrum signal of rebuilding to odd even frequency spectrum separation module 320.For example, spectral line vector decode module 400 is searched corresponding vector according to spectrum coding data or code word label in the vector coding code book of storage in advance, and rebuilds the MDCT territory frequency spectrum that comprises 1024 spectral lines according to the vector of searching.

Rotation angle vector decode module 410 obtains the rotation angle of each subband like greatly relevant rotation angle according to the signal that receives, and to the contrary rotary module 430 output rotation angle of quadrature.For example, rotation angle vector decode module 410 is searched corresponding vector according to the parameter coding data or the code word label that receive in the vector coding code book of storage in advance, and obtains the greatly relevant rotation angle of odd even frequency spectrum according to the vector that finds.

The following mixing spectrum signal that odd even frequency spectrum separation module 320 will receive separates into strange frequency spectrum and even frequency spectrum two-way subsignal, and upwards mixes module 420 these two-way subsignals of output.If the MDCT territory frequency spectrum of spectral line vector decode module 400 outputs comprises 1024 spectral lines, then the two-way subsignal of odd even frequency spectrum separation module 320 outputs all comprises 512 spectral lines.

Go up mixed module 420 the odd even frequency spectrum that receives is copied into two groups of signals, all comprise the left and right acoustic channels two paths of signals in every group of signal, and to contrary rotary module 430 each the road signal of output of quadrature.For example, upward mix module 420 and receive the subsignal that two-way all comprises 512 spectral lines,, all be copied into the left and right acoustic channels two paths of signals and every road signal all comprises 512 spectral lines to each way signal.

The contrary rotary module of quadrature 430 is unit with the subband, according to the opposite number of the rotation angle that receives each the road odd even spectrum signal that receives is carried out the quadrature rotational transform; Obtaining the odd even frequency spectrum of decoded left and right acoustic channels, and merge the odd even frequency spectrum that module 440 outputs obtain to the odd even frequency spectrum.Need to prove that rotation angle vector decode module 410 can be with after the rotation angle negate that obtains, to contrary rotary module 430 outputs of quadrature, at this moment, the contrary rotary module 430 of quadrature can directly carry out the quadrature rotational transform according to the rotation angle that receives to the odd even frequency spectrum.In addition, the contrary rotary module 430 of quadrature can be carried out the sub-band division operation, and certainly, the sub-band division operation also can be carried out by other module, for example, is carried out by last mixed module 420 or odd even frequency spectrum separation module 320.

The strange frequency spectrum that odd even frequency spectrum merging module 440 will receive, even frequency spectrum obtain the decoding frequency spectrum of left and right acoustic channels according to the alternate merging of frequency size order, and export the decoding frequency spectrum of left and right acoustic channels to IMDCT module 450.

The decoding frequency spectrum that IMDCT module 450 will receive carries out spatial transform, obtains the time domain frame of left and right acoustic channels, and exports the time domain frame of left and right acoustic channels to overlap-add module 460.For example, IMDCT module 450 can be with 1024 line reversals of the decoding frequency spectrum that receives in time domain, thereby obtains 2048 time domain sampling points, and to 460 outputs of overlap-add module.

Overlap-add module 460 is carried out overlap-add with the frame that receives, and obtains left and right acoustic channels two-way time-domain audio signal, and output.For example; 2048 sampled points and the sinusoidal windows function of every frame that overlap-add module 460 will receive multiply each other, and exist between the consecutive frame 1024 overlapping, then; Carry out the overlap-add of interframe, the first half that is about to present frame with former frame back half take advantage of addition after the sinusoidal windows respectively; To obtain the time-domain sampling point of L channel and R channel.

Need to prove that embodiment of the present invention can be mixed module 420, the contrary rotary module 430 of quadrature, the merging of odd even frequency spectrum module 440, IMDCT module 450 and overlap-add module 460 for left and right acoustic channels is provided with respectively; Also can go up mixed module 420, the contrary rotary module 430 of quadrature, the merging of odd even frequency spectrum module 440, IMDCT module 450 and overlap-add module 460 for left and right acoustic channels only is provided with one.In addition, above-mentioned code device and decoding device can close to be located in the consolidated network equipment, also can distinguish independent setting.

Be provided with the code device and the decoding device of foregoing description in the coding/decoding system that embodiment of the present invention provides, the description of code device, decoding device such as above-mentioned embodiment is in this no longer repeat specification.

Code device in the embodiment of the present invention, decoding device and coding/decoding system are through being divided at least two sub-frequency spectrums of spectral line space respectively with the frequency spectrum of the sound channel of MDCT territory neutral body acoustical signal, the frequency spectrum of stereophonic signal; And utilize the sub-frequency spectrum of sound channel to realize encoding and decoding; Can avoid envelope distortion, energy than phenomenon such as inaccurate to a certain extent, thus the experience that has improved user's stereo matter.

Description through above embodiment; Those skilled in the art can be well understood to the present invention and can realize by the mode that software adds essential hardware platform; Can certainly all implement, but the former is better embodiment under a lot of situation through hardware.Based on such understanding; All or part of can the coming out that technical scheme of the present invention contributes to background technology with the embodied of software product; This computer software product can be stored in the storage medium, like ROM/RAM, magnetic disc, CD etc., comprises that some instructions are with so that a computer equipment (can be a personal computer; Server, the perhaps network equipment etc.) carry out the described method of some part of each embodiment of the present invention or embodiment.

Though described the present invention through embodiment, those of ordinary skills know, the present invention has many distortion and variation and do not break away from spirit of the present invention, and the claim of application documents of the present invention comprises these distortion and variation.

Claims (19)

1. a stereophonic signal coding method is characterized in that, comprising:

Obtain the frequency spectrum of the sound channel of the stereophonic signal of stereophonic signal in revising discrete cosine transform MDCT territory;

With the spectrum division of the sound channel of the said MDCT territory neutral body acoustical signal of obtaining is at least two sub-frequency spectrums of spectral line space;

The sub-frequency spectrum of the sound channel of the stereophonic signal after dividing is carried out MDCT territory stereophonic signal encoding process;

The spectrum division of said sound channel with the said MDCT territory neutral body acoustical signal of obtaining is that the step of at least two sub-frequency spectrums of spectral line space comprises:

The frequency spectrum of each sound channel in the MDCT territory is arranged based on the frequency size order, and the order label;

According to said label the frequency spectrum of said each sound channel is divided into the sub-frequency spectrum of odd even respectively.

2. the method for claim 1 is characterized in that, the step of the frequency spectrum of the said sound channel of obtaining the stereophonic signal of stereophonic signal in revising discrete cosine transform MDCT territory comprises:

Stereophonic signal on the time domain is transformed to the MDCT territory;

Obtain the frequency spectrum of each sound channel of the stereophonic signal in the MDCT territory.

3. method as claimed in claim 2 is characterized in that, the said step that stereophonic signal on the time domain is transformed to the MDCT territory comprises:

Each sound channel signal on the time domain all be divided between isometric frame and consecutive frame include overlapped signal, and with said each frame transform to the MDCT territory.

4. like the described method of arbitrary claim in the claim 1 to 3, it is characterized in that the step that the sub-frequency spectrum of the sound channel of the stereophonic signal after said will the division carries out MDCT territory stereophonic signal encoding process comprises:

With the sub-spectrum division of said each sound channel is a plurality of subbands, and to be unit with the subband carry out the quadrature rotational transform respectively to the sub-frequency spectrum of each sound channel;

The sub-frequency spectrum of the different sound channels after the quadrature rotational transform is mixed down, and plug together the road;

With said conversion of signals of closing behind the road is frequency-region signal, and carries out encoding process at frequency domain.

5. method as claimed in claim 4 is characterized in that, said sub-spectrum division with said each sound channel is a plurality of subbands, and the step that to be unit with the subband carry out the quadrature rotational transform respectively to the sub-frequency spectrum of each sound channel comprises:

Is the greatly relevant anglec of rotation that unit calculates each sub-frequency spectrum with the subband, and respectively the sub-frequency spectrum of each subband is carried out the quadrature rotation transformation based on the greatly relevant anglec of rotation of the sub-frequency spectrum of each subband that calculates.

6. method as claimed in claim 4 is characterized in that, said method also comprises after the quadrature rotational transform:

Rotation angle to the sub-frequency spectrum of each subband adopts in quadrature rotational transform process is carried out Frequency Domain Coding, and the rotary angle information after decoding end is sent said coding.

7. a stereophonic signal coding/decoding method is characterized in that, comprising:

Stereophonic signal behind the coding is carried out the frequency spectrum vector decode to obtain the frequency spectrum of the stereophonic signal of stereophonic signal in the MDCT territory;

With the spectrum division of the said MDCT territory neutral body acoustical signal of obtaining is at least two sub-frequency spectrums of spectral line space;

The sub-frequency spectrum of the stereophonic signal after dividing is carried out MDCT territory stereophonic signal decoding processing;

The step that said spectrum division with the said MDCT territory neutral body acoustical signal of obtaining is at least two sub-frequency spectrums of spectral line space comprises:

The frequency spectrum of the stereophonic signal in the MDCT territory is arranged according to the frequency size order, and the order label;

According to said label the frequency spectrum of said stereophonic signal is divided into the sub-frequency spectrum of odd even respectively.

8. method as claimed in claim 7 is characterized in that, the step that the sub-frequency spectrum of the stereophonic signal after said will the division carries out MDCT territory stereophonic signal decoding processing comprises:

Each sub-frequency spectrum gone up respectively mix, to obtain the sub-frequency spectrum of stereo each sound channel;

With the sub-spectrum division of said each sound channel is a plurality of subbands, and to be unit with the subband carry out the contrary rotational transform of quadrature respectively to the sub-frequency spectrum of each sound channel;

Plug together the road between sub-frequency spectrum with each sound channel after the contrary rotational transform of quadrature;

Said each sound channel signal that closes behind the road is carried out contrary MDCT conversion, and carry out overlap-add and handle.

9. method as claimed in claim 8 is characterized in that, the said step that to be unit with the subband carry out the contrary rotational transform of quadrature respectively to the sub-frequency spectrum of each sound channel comprises:

The greatly relevant anglec of rotation of each subband that obtains based on the frequency spectrum vector decode is carried out the contrary rotation transformation of quadrature to the sub-frequency spectrum of each subband respectively.

10. a code device is characterized in that, comprising:

First acquisition module is used for obtaining the frequency spectrum of stereophonic signal in the sound channel of the stereophonic signal of revising discrete cosine transform MDCT territory;

First divides module, and the spectrum division of the sound channel of the MDCT territory neutral body acoustical signal that is used for said first acquisition module is obtained is at least two sub-frequency spectrums of spectral line space;

The encoding process module is used for the sub-frequency spectrum of the said first division Module Division is carried out MDCT territory stereophonic signal encoding process;

Said first divides module comprises:

Odd even frequency spectrum separation module is used for the frequency spectrum of each sound channel of MDCT territory neutral body acoustical signal is arranged according to the frequency size order, and the order label, according to said label the frequency spectrum of said each sound channel is divided into the sub-frequency spectrum of odd even respectively.

11. device as claimed in claim 10 is characterized in that, said first acquisition module comprises:

The windowing module is used for each sound channel signal on the time domain all is divided between isometric frame and consecutive frame and includes overlapped signal;

The MDCT module is used for said each frame transform to the MDCT territory, and obtains the frequency spectrum of each sound channel of the stereophonic signal in the MDCT territory.

12., it is characterized in that said encoding process module comprises like claim 10 or 11 described devices:

The quadrature rotary module, the sub-spectrum division that is used for said each sound channel is a plurality of subbands, and to be unit with the subband carry out the quadrature rotational transform respectively to the sub-frequency spectrum of each sound channel;

Mixed module is used for the sub-frequency spectrum of the different sound channels after the quadrature rotational transform is mixed down down;

The odd even frequency spectrum merges module, is used for plugging together the road between the sub-frequency spectrum after mixing under said;

Spectral line vector coding module, being used for the conversion of signals of closing behind the road is frequency-region signal, and carries out encoding process at frequency domain.

13. device as claimed in claim 12 is characterized in that, said device also comprises following at least one module:

Calculate greatly relevant rotation angle module, being used for the subband is the greatly relevant rotation angle that unit calculates each sub-frequency spectrum, and to the greatly relevant rotation angle of said each that calculates of quadrature rotary module output;

Rotation angle vector coding module, the rotation angle that is used for the employing of pair of orthogonal rotary module is carried out Frequency Domain Coding, and the rotary angle information after decoding end is sent said coding.

14. a decoding device is characterized in that, comprising:

Second acquisition module is used for the stereophonic signal behind the coding is carried out the frequency spectrum vector decode to obtain the frequency spectrum of stereophonic signal at the stereophonic signal of revising discrete cosine transform MDCT territory;

Second divides module, and the spectrum division of the MDCT territory neutral body acoustical signal that is used for said second acquisition module is obtained is at least two sub-frequency spectrums of spectral line space;

Decoding process module, be used for said second divide the stereophonic signal of Module Division sub-frequency spectrum carry out MDCT territory stereophonic signal decoding processing;

Said second divides module comprises:

Odd even frequency spectrum separation module is used for the frequency spectrum of MDCT territory neutral body acoustical signal is arranged according to the frequency size order, and the order label, according to said label said frequency spectrum is divided into the sub-frequency spectrum of odd even respectively.

15. device as claimed in claim 14 is characterized in that, decoding process module comprises:

Go up to mix module, be used for each sub-frequency spectrum is gone up respectively mixed, to obtain the sub-frequency spectrum of stereo each sound channel;

The contrary rotary module of quadrature, the sub-spectrum division that is used for said each sound channel is a plurality of subbands, and to be unit with the subband carry out the contrary rotational transform of quadrature respectively to the sub-frequency spectrum of each sound channel;

The odd even frequency spectrum merges module, is used for plugging together the road between the sub-frequency spectrum with each sound channel after the contrary rotational transform of quadrature;

The IMDCT module is used for said each sound channel signal that closes behind the road is carried out contrary MDCT conversion;

The overlap-add module is used for that the signal after the contrary MDCT conversion is carried out overlap-add and handles.

16. device as claimed in claim 15 is characterized in that, said device also comprises:

Rotation angle vector decode module is used for the signal behind the coding is rotated the angle vector decode, obtaining the rotation angle of each subband, and sends said rotation angle to the contrary rotary module of quadrature.

17. a coding/decoding system is characterized in that, comprises code device and decoding device,

Said code device comprises:

First acquisition module is used for obtaining the frequency spectrum of stereophonic signal in the sound channel of the stereophonic signal of revising discrete cosine transform MDCT territory;

First divides module, and the spectrum division of the sound channel of the MDCT territory neutral body acoustical signal that is used for said first acquisition module is obtained is at least two sub-frequency spectrums of spectral line space;

The encoding process module is used for the sub-frequency spectrum of the said first division Module Division is carried out MDCT territory stereophonic signal encoding process;

Said first divides module comprises:

Odd even frequency spectrum separation module is used for the frequency spectrum of each sound channel of MDCT territory neutral body acoustical signal is arranged according to the frequency size order, and the order label, according to said label the frequency spectrum of said each sound channel is divided into the sub-frequency spectrum of odd even respectively;

Said decoding device comprises:

Second acquisition module is used for the stereophonic signal behind the coding is carried out the frequency spectrum vector decode to obtain the frequency spectrum of stereophonic signal at the stereophonic signal of revising discrete cosine transform MDCT territory;

Second divides module, and the spectrum division of the MDCT territory neutral body acoustical signal that is used for said second acquisition module is obtained is at least two sub-frequency spectrums of spectral line space;

Decoding process module, be used for said second divide the stereophonic signal of Module Division sub-frequency spectrum carry out MDCT territory stereophonic signal decoding processing;

Said second divides module comprises:

Odd even frequency spectrum separation module is used for the frequency spectrum of MDCT territory neutral body acoustical signal is arranged according to the frequency size order, and the order label, according to said label said frequency spectrum is divided into the sub-frequency spectrum of odd even respectively.

18. system as claimed in claim 17 is characterized in that, said encoding process module comprises:

The quadrature rotary module, the sub-spectrum division that is used for said each sound channel is a plurality of subbands, and to be unit with the subband carry out the quadrature rotational transform respectively to the sub-frequency spectrum of each sound channel;

Mixed module is used for the sub-frequency spectrum of the different sound channels after the quadrature rotational transform is mixed down down;

The odd even frequency spectrum merges module, is used for plugging together the road between the sub-frequency spectrum after mixing under said;

Spectral line vector coding module, being used for the conversion of signals of closing behind the road is frequency-region signal, and carries out encoding process at frequency domain.

19. system as claimed in claim 17 is characterized in that, decoding process module comprises:

Go up to mix module, be used for each sub-frequency spectrum is gone up respectively mixed, to obtain the sub-frequency spectrum of stereo each sound channel;

The contrary rotary module of quadrature, the sub-spectrum division that is used for said each sound channel is a plurality of subbands, and to be unit with the subband carry out the contrary rotational transform of quadrature respectively to the sub-frequency spectrum of each sound channel;

The odd even frequency spectrum merges module, is used for plugging together the road between the sub-frequency spectrum with each sound channel after the contrary rotational transform of quadrature;

The IMDCT module is used for said each sound channel signal that closes behind the road is carried out contrary MDCT conversion;

The overlap-add module is used for that the signal after the contrary MDCT conversion is carried out overlap-add and handles.

Images