CN102246229B - Predicting method and apparatus for frequency domain pulse decoding and decoder - Google Patents
Predicting method and apparatus for frequency domain pulse decoding and decoder Download PDFInfo
- Publication number
- CN102246229B CN102246229B CN2009801486921A CN200980148692A CN102246229B CN 102246229 B CN102246229 B CN 102246229B CN 2009801486921 A CN2009801486921 A CN 2009801486921A CN 200980148692 A CN200980148692 A CN 200980148692A CN 102246229 B CN102246229 B CN 102246229B
- Authority
- CN
- China
- Prior art keywords
- frame
- frequency spectrum
- present frame
- spectral coefficient
- spectrum blocks
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L19/00—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
- G10L19/04—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using predictive techniques
Abstract
A predicting method and apparatus for frequency domain pulse decoding and a decoder are provided. The method includes: dividing current frame and previous frame into spectral blocks according to spectral coefficient of the previous frame (11), judging whether the prediction between frames for the spectral block divided in the current frame is needed according to the correlation between the spectral blocks divided in the current frame and previous frame (12), for the spectral block of the current frame which is judged that the prediction between frames is needed, predicting the spectral coefficient which is not decoded of the spectral block of the current frame by using the decoded spectral coefficient in the spectral block corresponding to the previous frame and the decoded spectral coefficient in the current frame (13).
Description
Technical field
The present invention relates to the audio encoding and decoding technique field, particularly relate to a kind of Forecasting Methodology and prediction unit and demoder of frequency domain pulse decoding.
Background technology
In present frequency domain pulse decoding, when bit number is not enough, generally according to the spectrum energy size, perhaps according to auditory properties, select some relatively important spectral coefficients to encode.And the spectral coefficient to not being encoded when decoding end is recovered, is generally filled with 0.
Because between the different frame, frequency spectrum relative energy size or auditory properties in the frame have difference, so, under same bit rate, between continuous multiframe, can not guarantee that the spectral coefficient of same frequency range all is encoded.Like this, in the spectral coefficient that decoding end recovers, some frame of same frequency range can decode spectral coefficient, and some frame can only be to obtain by filling 0, therefore will cause the discontinuous of same frequency range spectral coefficient, and then cause acoustical quality not ideal enough, especially strong to harmonic wave signal.
In the prior art, according to the similarity of different frequency range spectral coefficient in the frequency domain, do prediction in the frame by a small amount of bit or without bit, so that the spectral coefficient prediction of the spectral coefficient that does not decode by other frequency range or frequency obtains.Bandwidth expansion (BWE such as present more employing, Bandwidth Extension) algorithm, use the spectral coefficient of low frequency to predict the spectral coefficient that obtains high band according to the correlativity between the low-and high-frequency, with the bandwidth of lifting output signal, and then the acoustical quality of lifting output signal.
In realizing process of the present invention, the inventor studies discovery: often deviation is larger in some frequency range for the spectrum energy that infra-frame prediction goes out, when particularly and high-frequency harmonic strong when low-frequency harmonics is also strong, the crest location that dopes often has relatively large deviation with real crest location, thereby so that sound signal has been introduced more noise, affected the acoustical quality of sound signal.
Summary of the invention
Given this, the embodiment of the invention provides a kind of Forecasting Methodology and prediction unit and demoder of frequency domain pulse decoding, can promote better the acoustical quality of audio output signal.
The Forecasting Methodology of the frequency domain pulse decoding that the embodiment of the invention provides comprises the steps:
According to the spectral coefficient of frame before to present frame and before frame carry out frequency spectrum blocks and divide;
According to present frame and the correlativity of the frame frequency spectrum blocks of dividing before, judge whether the frequency spectrum blocks of present frame division needs to do inter prediction;
To judging the present frame frequency spectrum blocks that need to do inter prediction, the spectral coefficient that uses the spectral coefficient that decodes in the corresponding frequency spectrum blocks of frame before described and present frame to decode, the spectral coefficient that does not decode in the prediction present frame frequency spectrum blocks.
The prediction unit of the frequency domain pulse decoding that the embodiment of the invention provides comprises:
The piece division unit, be used for according to before frame spectral coefficient to present frame and before frame carry out the frequency spectrum blocks division;
Judging unit is used for the present frame divided according to described division unit and the correlativity of frame frequency spectrum blocks before, judges whether the frequency spectrum blocks of present frame division needs to do inter prediction;
Predicting unit, be used for described judgment unit judges is gone out the present frame frequency spectrum blocks that need to do inter prediction, the spectral coefficient that uses the spectral coefficient that decodes in the corresponding frequency spectrum blocks of frame before described and present frame to decode, the spectral coefficient that does not decode in the prediction present frame frequency spectrum blocks.
The demoder that the embodiment of the invention provides, the prediction unit and the transducer that comprise above-mentioned frequency domain pulse decoding, the prediction unit of described frequency domain pulse decoding, be used for judging the present frame frequency spectrum blocks that to do inter prediction, the spectral coefficient that the spectral coefficient that decodes in the corresponding frequency spectrum blocks of frame before using and present frame have decoded, the spectral coefficient that does not decode in the prediction present frame frequency spectrum blocks; Described transducer is used for carrying out frequency domain to the conversion of time domain, the output time-domain sound signal according to the frame frequency spectral coefficient after the prediction unit prediction of described frequency domain pulse decoding.
The above technical scheme that provides according to the embodiment of the invention as can be known, at first according to before frame spectral coefficient to present frame and before frame carry out frequency spectrum blocks division, then according to present frame and the correlativity of the frame frequency spectrum blocks of dividing before, judge whether the frequency spectrum blocks that present frame is divided needs to do inter prediction, at last to judging the present frame frequency spectrum blocks that to do inter prediction, the spectral coefficient that the spectral coefficient that decodes in the corresponding frequency spectrum blocks of frame before using and present frame have decoded, the spectral coefficient that does not decode in the prediction present frame frequency spectrum blocks.Compare with the intra-frame prediction method of prior art, for the signal of strong harmonic wave, so that the frequency spectrum of present frame is more level and smooth, reduced discontinuous phenomenon, thereby more approaching with real frequency spectrum, promoted better the especially by force acoustical quality of the output signal of harmonic wave of audio output signal.
Description of drawings
In order to be illustrated more clearly in the embodiment of the invention or technical scheme of the prior art, the accompanying drawing of required use was done to introduce simply during the below will describe embodiment, apparently, accompanying drawing in the following describes only is some embodiments of the present invention, for those of ordinary skills, under the prerequisite of not paying creative work, can also obtain according to these accompanying drawings other accompanying drawing.
The Forecasting Methodology process flow diagram of a kind of frequency domain pulse decoding that Fig. 1 provides for the embodiment of the invention one;
The Forecasting Methodology process flow diagram of a kind of frequency domain pulse decoding that Fig. 2 provides for the embodiment of the invention two;
A kind of present frame that Fig. 3 provides for the embodiment of the invention and the spectrum information exemplary plot of former frame;
A kind of algorithm structure block diagram that Fig. 4 provides for the embodiment of the invention two;
The another kind of algorithm structure block diagram that Fig. 5 provides for the embodiment of the invention two;
A kind of algorithm structure block diagram that Fig. 6 provides for the embodiment of the invention three;
The prediction unit structural representation of a kind of frequency domain pulse decoding that Fig. 7 provides for the embodiment of the invention four;
Fig. 8 forms block diagram for a kind of demoder that the embodiment of the invention five provides.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the invention, the technical scheme in the embodiment of the invention is clearly and completely described, obviously, described embodiment only is the present invention's part embodiment, rather than whole embodiment.Based on the embodiment among the present invention, those of ordinary skills belong to the scope of protection of the invention not making the every other embodiment that obtains under the creative work prerequisite.
Embodiment one
Referring to Fig. 1, the Forecasting Methodology of the frequency domain pulse decoding that the embodiment of the invention provides comprises the steps:
Step 12 according to present frame and the correlativity of the frame frequency spectrum blocks of dividing before, judges whether the frequency spectrum blocks of present frame division needs to do inter prediction;
Need to do in finishing present frame after the prediction of each frequency spectrum blocks of inter prediction, the present frame after can the prediction of output can further carry out follow-up decoding to present frame afterwards and process, final output audio signal.
The Forecasting Methodology of the frequency domain pulse decoding that the embodiment of the invention provides, at first according to before frame spectral coefficient to present frame and before frame carry out frequency spectrum blocks division, then according to present frame and the correlativity of the frame frequency spectrum blocks of dividing before, judge whether the frequency spectrum blocks that present frame is divided needs to do inter prediction, at last to judging the present frame frequency spectrum blocks that to do inter prediction, the spectral coefficient that the spectral coefficient that decodes in the corresponding frequency spectrum blocks of frame before using and present frame have decoded, the spectral coefficient that does not decode in the prediction present frame frequency spectrum blocks.Compare with the intra-frame prediction method of prior art, for the signal of strong harmonic wave, so that the frequency spectrum of present frame is more level and smooth, reduced discontinuous phenomenon, thereby more approaching with true spectrum, promoted better the especially by force acoustical quality of the output signal of harmonic wave of audio output signal.
Embodiment two
Referring to Fig. 2, the Forecasting Methodology of the frequency domain pulse decoding that the embodiment of the invention provides comprises the steps:
Frequency spectrum blocks is comprised of spectral coefficient corresponding to a continuous L frequency, centered by frame decoding before the correspondence goes out each frequency of spectral coefficient, the scope of each N frequency marks off a frequency spectrum blocks in front and back, the number of the spectral coefficient that frame decoding went out before at this moment the number of the frequency spectrum blocks that marks off of present frame equaled, wherein L is the length of frequency spectrum blocks, represent the frequency spectrum number in each frequency spectrum blocks, L=2N+1.Referring to Fig. 3, suppose that the number of the total frequency in the pending frequency band of every frame is 64, in decoding end, present frame decodes 4 non-zero spectral coefficients, and corresponding frequency is respectively 4,15,40 and 50 (strictly according to the facts shown in the vertical lines); The former frame of present frame also decodes 4 non-zero spectral coefficients, and corresponding frequency is respectively 3,27,40 and 50 (strictly according to the facts shown in the vertical lines).If get N=2, i.e. frequency spectrum number L=5 in each frequency spectrum blocks, then corresponding before 4 non-zero spectral coefficient place frequencies going out of frame decoding, present frame and former frame are marked off respectively four pending frequency spectrum blocks (shown in the dotted line frame).Corresponding to Fig. 3, the number 4 of present frame and the spectral coefficient that frame decoding goes out before the number of the frequency spectrum blocks that marks off of frame equals before.
Need to prove, if frame decodes plural spectral coefficient in an adjacent N frequency point ranges before, then present frame is centered by arbitrary frequency at the place of described plural spectral coefficient, and the scope of each N frequency only marks off a frequency spectrum blocks in front and back.If frame all decodes non-zero spectral coefficient on the 3rd and 4 frequencies before, if get N=2, then corresponding to the 3rd and 4 frequency present frames of frame before and before frame only mark off a frequency spectrum blocks, this frequency spectrum blocks can be centered by the 3rd or the 4th frequency.At this moment between each frequency spectrum blocks that marks off without region overlapping, the number of the spectral coefficient that goes out of frame decoding is greater than present frame and the number of the frequency spectrum blocks that marks off of frame before before.
Step 22, judge present frame and before all spectral coefficients that decode of frame the spacing of corresponding frequency less than or equal to the number of N whether more than or equal to M; If more than or equal to M then judge that present frame need to do inter prediction, otherwise do not need; Wherein M is predefined numerical value.
Need to prove, the value of M is according to the number decision of the spectral coefficient that decodes, and the value of the more M of spectral coefficient that decode is larger.
Still referring to Fig. 3, get N=2, M=3, present frame and former frame relevant position differ less than or equal to N 3 positions (3/4,40/40,50/50) are arranged, more than or equal to M, judge that then present frame need to do inter prediction.Owing to having carried out the division of frequency spectrum blocks in the step 21, so, in the step 22, can judge in the following way also whether present frame needs to do inter prediction: if in each frequency spectrum blocks in the present frame, have the number of frequency spectrum blocks of the spectral coefficient that decodes more than or equal to predefined numerical value M, judge that then present frame need to do inter prediction; Otherwise present frame does not need to do inter prediction.
2 points that have that need explanation: the one, the processing of present frame, not necessarily only with reference to former frame, also can be with reference to the information of former frames.The 2nd, when before frame when the frame losing phenomenon is arranged, both the spectral coefficient of frame was set to 0 entirely before, so just can guarantee that the first frame after the frame losing is not done inter prediction, thereby avoid bad impact; The spectral coefficient of frame is constant before in the time of also can keeping frame losing, and the condition by whether present frame being done inter prediction (location gap that frame decoding goes out spectral coefficient before when present frame and frame losing less than or equal to the number of N more than or equal to M), guarantee the robustness of algorithm, can not cause bad effect.
Step 23, judge one by one whether the frequency spectrum blocks of dividing needs to do inter prediction, if the spectral coefficient that does not decode in the frequency spectrum blocks of present frame, and in the corresponding frequency spectrum blocks of frame the spectral coefficient that decodes is arranged before, judge that then the described frequency spectrum blocks of present frame need to be done inter prediction.
Still referring to Fig. 3,4 non-zero spectral coefficient place frequencies that decode in corresponding former frame, in present frame and frame marks off before four the pending frequency spectrum blocks, only at the frequency spectrum blocks place of frequency 27 correspondences, former frame has decoded spectral coefficient, and in the scope of the relevant position (27+/-2) of present frame, do not decode spectral coefficient, and other frequency spectrum blocks (frequency 4, frequency spectrum blocks corresponding to 40 and 50 difference) in, present frame has corresponding spectral coefficient to decode out, therefore, corresponding to Fig. 3, judge the frequency spectrum blocks of frequency 27 correspondences and need to do inter prediction.
Step 24, to judging the frequency spectrum blocks that to do inter prediction, the amplitude of spectral coefficient place frequency will be decoded in the corresponding frequency spectrum blocks of frame before, the δ of the minimum radius in all spectral coefficients that decoded with present frame doubly is weighted summation, with the result of weighted sum as the amplitude that does not decode the corresponding frequency of spectral coefficient in the present frame frequency spectrum blocks, the symbol of spectral coefficient is identical with the symbol of the spectral coefficient of the corresponding frequency of frame before, wherein 0≤δ≤1.
Current frequency spectrum blocks is done the concrete grammar of inter prediction suc as formula shown in (1):
cur_spec＝α*pre_spec+β*(δ*cur_spec
min)；α+β＝1，0≤δ≤1(1)
Wherein, cur_spec
MinMinimum radius in all spectral coefficients that the expression present frame has decoded, the amplitude of the spectral coefficient place frequency that decodes in the corresponding frequency spectrum blocks of frame before pre_spec represents.α, β, δ are weighting coefficients, can select according to actual conditions, for strong harmonic wave signal, preferably, α=β=0.5, δ=0.8
Still referring to accompanying drawing 3, frequency 27 is done the spectral coefficient prediction, it is 0.8 times weighted sum with the minimum radius in the amplitude of the spectral coefficient at former frame 27 frequency places and all spectral coefficients that present frame decodes, as the amplitude of the spectral coefficient at present frame frequency 27 places, the symbol at symbol and former frame 27 frequency places is consistent.For example, former frame 27 bin magnitudes are 30, and four bin magnitudes of present frame are 25,15,20,21 successively, α=β=0.5, δ=0.8 is 21 by the amplitude that formula (1) prediction obtains present frame 27 frequencies, and symbol is being for just, is consistent with the symbol at former frame 27 frequency places.
Need to prove, the embodiment of the invention is based on frequency spectrum blocks and judges when judging whether current frequency spectrum blocks needs to process, rather than judge based on frequency.Some frequency of frame has decoded spectral coefficient before, the same frequency of present frame does not decode spectral coefficient, but there is spectral coefficient to decode at a distance of very little position around it, do not do corresponding processing this moment yet, in accompanying drawing 3, former frame frequency 3 has decoded spectral coefficient, and present frame frequency 3 does not decode spectral coefficient, but present frame frequency 4 has decoded spectral coefficient, and be not need do inter prediction process to the frequency 3 of present frame this moment.
Also need to prove, if decode plural spectral coefficient in the same frequency spectrum blocks of frame before, and do not have spectral coefficient to decode out in the corresponding frequency spectrum blocks in the present frame, each frequency at the plural spectral coefficient place that then decodes in the corresponding before same frequency spectrum blocks of frame carries out respectively spectrum prediction to present frame.
Through above-mentioned steps 24, present frame dopes spectral coefficient at frequency 27, this moment, present frame had 5 non-zero spectral coefficients, corresponding frequency is respectively 4,15,27,40 and 50, spectral coefficient to the present frame after the prediction processing is exported, and preserves the spectral coefficient (corresponding frequency is respectively 4,15,40 and 50) of the present frame before the prediction processing as the frame information of frame before the next frame.
So far, finished the whole treatment scheme of the predicted current frame of frequency domain pulse decoding.
The algorithm structure block diagram of the embodiment of the invention according to present frame and the spectral coefficient that goes out of frame decoding before, is at first determined the number of pending frequency spectrum blocks as shown in Figure 4, and k represents the ordinal number of frequency spectrum blocks, and initialization is since 0.Corresponding to the frame information shown in the accompanying drawing 3, the number of pending frequency spectrum blocks equals the number 4 of the spectral coefficient that former frame decodes.Then judge present frame and before frame decoding go out location gap in all corresponding blocks of spectral coefficient less than or equal to the number of N whether more than or equal to M, if satisfy condition, then present frame is done inter prediction, and preserve the front spectral coefficient of predicted current frame as frame information before the next frame; Otherwise, present frame is not processed, just preserve the front spectral coefficient of predicted current frame as frame information before the next frame.
Corresponding to accompanying drawing 3, desirable N=2, M=3.Decoded spectral coefficient at frame frequency point 27 before, and the frequency spectrum blocks of dividing at present frame (27+/-N) do not decode spectral coefficient in the scope; Simultaneously, present frame and before the frame relevant position differ have 3 (3/4 less than or equal to N, 40/40,50/50), more than or equal to M=3, at this moment, the δ of minimum radius in the amplitude of the amplitude of the spectral coefficient of frame frequency point 27 before and all spectral coefficients that decode of present frame=0.8 times is asked weighted sum, with the amplitude of weighted sum result as the spectral coefficient of present frame frequency 27, symbol and before the symbol of frame frequency point 27 be consistent.
Need to prove, when most of frames can both satisfy the inter prediction treatment conditions, when judging whether present frame does inter prediction, can after handling all frequency spectrum blocks, judge again, rather than judge average complexity that like this can saving-algorithm carrying out the frequency spectrum blocks inter prediction before processing.
As shown in Figure 5, the present embodiment can also adopt another kind of algorithm structure block diagram, judges that whether present frame does inter prediction is just to learn after handling all frequency spectrum blocks, and concrete Processing Algorithm is as follows:
At first the initialization present frame and before the frame decoding location gap that goes out spectral coefficient and determine maximum pending frequency spectrum blocks number less than or equal to the ordinal number k=0 of the number i=0 of N and pending frequency spectrum blocks.No matter whether present frame satisfies the condition of doing inter prediction, all with present frame temporarily by algorithm of the present invention: for each pending frequency spectrum blocks, if decoded spectral coefficient in the spectral coefficient piece of frame before, and the relevant block of present frame does not decode spectral coefficient, frequency spectrum blocks such as frequency 27 places in the accompanying drawing 3 namely satisfies this condition, thereby can be with the weighted sum of the amplitude of the spectral coefficient of former frame frequency 27 and the δ of the minimum radius in all spectral coefficients that decode of present frame=0.8 times, as the amplitude of the spectral coefficient of present frame frequency 27, the symbol of symbol and former frame frequency 27 is consistent; If the present frame frequency spectrum blocks has spectral coefficient to decode with the corresponding frequency spectrum blocks of frame before, judge that whether present frame position and former frame location gap are less than or equal to N=2 in the frequency spectrum blocks, if so, i adds one, otherwise i is constant, until handle all frequency spectrum blocks.After all frequency spectrum blocks are handled, whether judge i more than or equal to M, if so, present frame satisfies treatment conditions, the present frame spectral coefficient after the prediction of output is processed, and preserve the front present frame spectral coefficient of prediction processing as the former frame spectral coefficient of next frame; Otherwise present frame does not satisfy treatment conditions, the spectral coefficient that the present frame spectral coefficient is reduced to before the prediction processing is exported, and preserved spectral coefficient before predicted current frame is processed as frame frequency spectral coefficient before the next frame.
Embodiment three
The embodiment of the invention provides the method for the inter prediction of another kind of frequency domain pulse decoding, the present embodiment and embodiment two are not both, the present embodiment judges that whether present frame does the processing of inter prediction is to judge according to the correlativity of the energy of the spectral coefficient that decodes in present frame and the corresponding frequency spectrum blocks of frame before, and embodiment two be according to present frame and before frame decoding go out the location gap of spectral coefficient and judge less than or equal to the number (being the correlativity of the position of spectral coefficient) of N.
Particularly, when judging whether present frame needs to do inter prediction, the present embodiment be according to present frame and before in each frequency spectrum blocks of frame the corresponding energy information of all spectral coefficients that decode judge, if the number of described energy equivalence frequency spectrum blocks, is then judged present frame more than or equal to predefined numerical value and need to be done inter prediction.The frequency spectrum blocks energy can with the root mean square of the quadratic sum of spectral coefficient amplitude in the frequency spectrum blocks or quadratic sum or spectral coefficient amplitude and etc. represent, a kind of based on the spectral coefficient amplitude square and the computing formula of frequency spectrum blocks energy be:
Wherein, k is the ordinal number of frequency spectrum blocks, and L is the number of spectral coefficient in the frequency spectrum blocks.
So-called energy equivalence, the energy Ratios before referring between frame and the present frame frequency spectrum blocks are in [1/E, E] scope, and desirable 0.8 such as E, in general the E value can be near 1, to guarantee the accuracy of prediction.
Whether the present embodiment does the judgement flow process of inter prediction to present frame, can adopt the algorithm structure block diagram of similar Fig. 4 or Fig. 5.When most of frames can both satisfy the inter prediction treatment conditions, when judging whether present frame does inter prediction, can after handling all frequency spectrum blocks, judge again, with the average complexity of saving-algorithm.Referring to Fig. 6, concrete Processing Algorithm is as follows:
At first the initialization present frame and before frame decoding go out the number i=0 of the energy equivalence of spectral coefficient in the frequency spectrum blocks and the ordinal number k=0 of pending spectral coefficient piece, and determine maximum pending frequency spectrum blocks number.Though then present frame whether satisfy do motion estimation algorithm condition all with present frame temporarily by algorithm of the present invention: for each frequency spectrum blocks, if frame decoding has gone out the spectral coefficient piece before, and present frame does not decode, frequency spectrum blocks such as frequency 27 places among Fig. 3, weighted sum with the amplitude of the spectral coefficient of frame frequency point 27 before and the δ of the minimum radius in all spectral coefficients that decode of present frame=0.8 times, as the amplitude of the spectral coefficient of present frame frequency 27, the symbol of symbol and former frame frequency 27 is consistent; If the present frame frequency spectrum blocks and before the frame frequency spectrum blocks have spectral coefficient to decode, judge present frame in this frequency spectrum blocks and whether the energy of the spectral coefficient that goes out of frame decoding suitable before, if so, i adds one, otherwise i remains unchanged, until handle all frequency spectrum blocks.After all frequency spectrum blocks are handled, whether judge i more than or equal to M, M is predefined numerical value.If so, present frame satisfies treatment conditions, the present frame spectral coefficient after the prediction of output is processed, and preserve the front present frame spectral coefficient of prediction processing as frame frequency spectral coefficient before the next frame; Otherwise present frame does not satisfy treatment conditions, the present frame spectral coefficient that the present frame spectral coefficient is reduced to before the prediction processing is exported, and preserved present frame spectral coefficient before the prediction processing as frame frequency spectral coefficient before the next frame.
Need to prove, frequency spectrum number L in the frequency spectrum blocks of dividing in the present embodiment is preferably less than the frequency spectrum number L in the frequency spectrum blocks among the embodiment two, because when carrying out the energy equivalence counting, when frequency spectrum blocks is larger, even if the energy equivalence of spectral coefficient in the frequency spectrum blocks between the two continuous frames, also the correlativity between the possible two continuous frames is but very low.Therefore, if the length L of embodiment two intermediate frequency spectrum pieces=5, then the length L of frequency spectrum blocks is preferably got less than or equal to 3 in the present embodiment, i.e. L≤3.
Be understandable that, when judging present frame according to frame-to-frame correlation needs done inter prediction, also can with based on present frame and before frame decoding go out the judgement of the correlativity of the energy of spectral coefficient in the frequency spectrum blocks and combine based on the judgement of the number of the frequency spectrum blocks that decodes spectral coefficient in the present frame to carry out whether present frame is done the judgement of inter prediction.The method flow of concrete inter prediction can referring to the explanation of above inventive embodiments, not repeat them here.
Above embodiment of the method provided by the invention, at first according to before frame spectral coefficient to present frame and before frame carry out frequency spectrum blocks division, and according to present frame and the frame frequency spectrum blocks of dividing before, judge whether present frame needs to do inter prediction, then to judging the present frame that to do inter prediction, judge whether the frequency spectrum blocks of dividing needs to do inter prediction, if the spectral coefficient that does not decode in the frequency spectrum blocks of present frame, and in the corresponding frequency spectrum blocks of frame the spectral coefficient that decodes is arranged before, then judge the described frequency spectrum blocks of present frame, at last needs are done the frequency spectrum blocks of inter prediction, amplitude with the spectral coefficient place frequency that decodes in the corresponding frequency spectrum blocks of frame before this, (0≤δ≤1 times is weighted summation to the δ of the minimum radius in all spectral coefficients that decoded with present frame, with the result of weighted sum as the amplitude that does not decode the corresponding frequency of spectral coefficient in the described present frame frequency spectrum blocks, the symbol of spectral coefficient is identical with the symbol of the described before corresponding frequency of frame, wherein according to frame-to-frame correlation judge present frame whether needs do inter prediction can according to present frame and before all spectral coefficients that decode of frame the pitch information of corresponding frequency judge, and/or, according to present frame and before all spectral coefficients that decode of frame the energy information of corresponding frequency spectrum blocks judge.Compare with the intra-frame prediction method of prior art, for the signal of strong harmonic wave, so that the frequency spectrum of present frame is more level and smooth, reduced discontinuous phenomenon, thereby more approaching with true spectrum, promoted better the especially by force acoustical quality of the audio output signal of harmonic wave of audio output signal.
Embodiment four
As shown in Figure 7, corresponding to said method embodiment, the prediction unit of a kind of frequency domain pulse decoding that the embodiment of the invention provides comprises:
Piece division unit 71, be used for according to before frame spectral coefficient to present frame and before frame carry out the frequency spectrum blocks division;
Judging unit 70 is used for the present frame divided according to described division unit 71 and frame frequency spectrum blocks before, judges whether the frequency spectrum blocks of present frame division needs to do inter prediction;
Predicting unit 74, be used for described judging unit 70 is judged the present frame frequency spectrum blocks that need to do inter prediction, the spectral coefficient that uses the spectral coefficient that decodes in the corresponding frequency spectrum blocks of frame before described and present frame to decode, the spectral coefficient that does not decode in the prediction present frame frequency spectrum blocks.
Wherein, described division unit 71 comprises:
First divides module 711, be used for corresponding before the spectral coefficient place frequency that goes out of frame decoding, present frame and before the scope of each N frequency marks off a frequency spectrum blocks in front and back centered by this frequency respectively in the frame, wherein, N 〉=1.The number of the spectral coefficient that frame decoding went out before at this moment the number of the frequency spectrum blocks that marks off of present frame equaled.And/or,
Second divides module 712, if be used for before frame decodes plural spectral coefficient in an adjacent N frequency point ranges, then from the frequency of described plural spectral coefficient place, select arbitrary frequency, present frame and before the scope of each N frequency marks off a frequency spectrum blocks in front and back centered by the frequency of this selection respectively in the frame, wherein, N 〉=1.At this moment between each frequency spectrum blocks that marks off without region overlapping, the number of the frequency spectrum blocks that marks off greater than present frame of the number of the spectral coefficient that goes out of frame decoding before.
Described judging unit 70 comprises:
Frame judgment sub-unit 72 is used for the present frame divided according to described division unit 71 and frame frequency spectrum blocks before, judges whether present frame needs to do inter prediction;
Piece judgment sub-unit 73 is used for described frame judging unit 72 is judged the present frame that need to do inter prediction, judges whether the frequency spectrum blocks that present frame is divided needs to do inter prediction.
Described frame judgment sub-unit 72 comprises:
Position judging module 721, the number that is used for the frequency spectrum blocks of the spectral coefficient that decodes according to present frame is judged, if the number of the frequency spectrum blocks of the spectral coefficient that decodes in the present frame, is then judged present frame more than or equal to predefined numerical value and need to be done inter prediction; And/or,
Energy judge module 722, be used for the frequency spectrum blocks of the spectral coefficient that decodes according to present frame and before in the frame energy information of corresponding frequency spectrum blocks judge, if the number of corresponding frequency spectrum blocks energy equivalence, is then judged present frame more than or equal to predefined numerical value and need to be done inter prediction.
Described judgment sub-unit 73 can specifically be used for, according to present frame with have or not the spectral coefficient that decodes to judge in the corresponding frequency spectrum blocks of frame before, if the spectral coefficient that does not decode in the frequency spectrum blocks of present frame, and in the corresponding frequency spectrum blocks of frame the spectral coefficient that decodes is arranged before, judge that then the described frequency spectrum blocks of present frame need to be done inter prediction.
Described predicting unit 74 can specifically be used for, amplitude with the spectral coefficient place frequency that decodes in the corresponding frequency spectrum blocks of frame before described, the δ of the minimum radius in all spectral coefficients that decoded with present frame doubly is weighted summation, with the result of weighted sum as the amplitude that does not decode the corresponding frequency of spectral coefficient in the described present frame frequency spectrum blocks, the symbol of spectral coefficient is identical with the symbol of the described before corresponding frequency of frame, wherein 0≤δ≤1.
Be understandable that, the device that the embodiment of the invention provides also can comprise:
The prediction unit of the frequency domain pulse decoding that the embodiment of the invention provides, at first by the frame judging unit whether present frame being needed to do inter prediction judges, then to judging the present frame that to do inter prediction, by the piece judging unit whether the frequency spectrum blocks of dividing being needed to do inter prediction judges, at last to judging the frequency spectrum blocks that to do inter prediction, the spectral coefficient that the spectral coefficient that decodes in the corresponding frequency spectrum blocks of frame before using by predicting unit and present frame have decoded, the spectral coefficient that does not decode in the prediction present frame frequency spectrum blocks.Compare with the intra-frame prediction method of prior art, for the signal of strong harmonic wave, so that the frequency spectrum of present frame is more level and smooth, reduced discontinuous phenomenon, thereby more approaching with true spectrum, promoted better the especially by force acoustical quality of the output signal of harmonic wave of audio output signal.
Embodiment five
As shown in Figure 8, the embodiment of the invention also provides a kind of demoder, comprises prediction unit 81 and the transducer 82 of frequency domain pulse decoding;
The prediction unit 81 of described frequency domain pulse decoding, be used for judging the present frame frequency spectrum blocks that to do inter prediction, the spectral coefficient that the spectral coefficient that decodes in the corresponding frequency spectrum blocks of frame before using and present frame have decoded, the spectral coefficient that does not decode in the prediction present frame frequency spectrum blocks;
Described transducer 82 is used for carrying out frequency domain to the conversion of time domain, the output time-domain sound signal according to the frame frequency spectral coefficient after prediction unit 81 predictions of described frequency domain pulse decoding.
The prediction unit 81 of described frequency domain pulse decoding more specifically can referring to the prediction unit of the frequency domain pulse decoding among above-mentioned embodiment of the method and the device embodiment four, not repeat them here.
The professional can also recognize, unit and the algorithm steps of each example of describing in conjunction with embodiment disclosed herein, can realize with electronic hardware, computer software or the combination of the two, for the interchangeability of hardware and software clearly is described, composition and the step of each example described in general manner according to function in the above description.These functions are carried out with hardware or software mode actually, depend on application-specific and the design constraint of technical scheme.The professional and technical personnel can specifically should be used for realizing described function with distinct methods to each, but this realization should not thought and exceeds scope of the present invention.
The method of describing in conjunction with embodiment disclosed herein or the step of algorithm can use the software module of hardware, processor execution, and perhaps the combination of the two is implemented.Software module can place random access memory (RAM), internal memory, ROM (read-only memory) (ROM), electrically programmable ROM, electrically erasable ROM, register, hard disk, moveable magnetic disc, CD-ROM or any storage medium of other form.
The above only is preferred embodiment of the present invention, not in order to limiting the present invention, all any modifications of doing within the spirit and principles in the present invention, is equal to and replaces and improvement etc., all should be included within protection scope of the present invention.
Claims (15)
1. the Forecasting Methodology of a frequency domain pulse decoding is characterized in that, comprising:
According to the spectral coefficient of frame before to present frame and before frame carry out frequency spectrum blocks and divide;
According to present frame and the correlativity of the frame frequency spectrum blocks of dividing before, judge whether the frequency spectrum blocks of present frame division needs to do inter prediction;
To judging the present frame frequency spectrum blocks that need to do inter prediction, the spectral coefficient that uses the spectral coefficient that decodes in the corresponding frequency spectrum blocks of frame before described and present frame to decode, the spectral coefficient that does not decode in the prediction present frame frequency spectrum blocks.
2. method according to claim 1 is characterized in that, before the described basis spectral coefficient of frame to present frame and before the frame step of carrying out the frequency spectrum blocks division comprise:
The spectral coefficient place frequency that frame decoding goes out before corresponding, present frame and before the scope of each N frequency marks off a frequency spectrum blocks in front and back centered by this frequency respectively in the frame; And/or,
If frame decodes plural spectral coefficient in an adjacent N frequency point ranges before, then from the frequency of described plural spectral coefficient place, select arbitrary frequency, present frame and before the scope of each N frequency marks off a frequency spectrum blocks in front and back centered by the frequency of this selection respectively in the frame;
Wherein, N 〉=1.
3. method according to claim 1 and 2 is characterized in that, and is described according to present frame and the correlativity of the frame frequency spectrum blocks of dividing before, judges that the step whether frequency spectrum blocks of present frame division needs to do inter prediction comprises:
According to present frame and the correlativity of the frame frequency spectrum blocks of dividing before, judge whether present frame needs to do inter prediction;
To judging the present frame that to do inter prediction, judge one by one whether the frequency spectrum blocks that present frame is divided needs to do inter prediction.
4. method according to claim 3 is characterized in that, and is described according to present frame and the frame frequency spectrum blocks correlativity of dividing before, judges that the step whether frequency spectrum blocks of present frame division needs to do inter prediction comprises:
If the number of the frequency spectrum blocks of the spectral coefficient that decodes in the present frame, is then judged present frame more than or equal to predefined numerical value and need to be done inter prediction; And/or,
If the frequency spectrum blocks of the spectral coefficient that decodes in the present frame and before in the frame number of the energy equivalence of corresponding frequency spectrum blocks judge then that more than or equal to predefined numerical value present frame need to do inter prediction.
5. method according to claim 3 is characterized in that, the step whether described frequency spectrum blocks of judging that present frame is divided needs to do inter prediction comprises:
If the spectral coefficient that does not decode in the frequency spectrum blocks of present frame, and in the corresponding frequency spectrum blocks of frame the spectral coefficient that decodes is arranged before, judge that then this frequency spectrum blocks of present frame need to be done inter prediction.
6. method according to claim 5, it is characterized in that, described to judging the present frame frequency spectrum blocks that to do inter prediction, the spectral coefficient that uses the spectral coefficient that decodes in the corresponding frequency spectrum blocks of frame before described and present frame to decode, predict that the step of the spectral coefficient that does not decode in this current frame frequency spectrum piece comprises:
Amplitude with the spectral coefficient place frequency that decodes in the corresponding frequency spectrum blocks of frame before described, the δ of the minimum radius in all spectral coefficients that decoded with present frame doubly is weighted summation, with the result of weighted sum as the amplitude that does not decode the corresponding frequency of spectral coefficient in the described present frame frequency spectrum blocks, the symbol of spectral coefficient is identical with the symbol of the described before corresponding frequency of frame, wherein 0≤δ≤1.
7. method according to claim 1 is characterized in that, described method also comprises:
The spectral coefficient of the present frame after the prediction of output is processed, and preserve the spectral coefficient of the present frame before the prediction processing as the frame information of frame before the next frame.
8. the prediction unit of a frequency domain pulse decoding is characterized in that, comprising:
The piece division unit, be used for according to before frame spectral coefficient to present frame and before frame carry out the frequency spectrum blocks division;
Judging unit is used for the present frame divided according to described division unit and the correlativity of frame frequency spectrum blocks before, judges whether the frequency spectrum blocks of present frame division needs to do inter prediction;
Predicting unit, be used for described judgment unit judges is gone out the present frame frequency spectrum blocks that need to do inter prediction, the spectral coefficient that uses the spectral coefficient that decodes in the corresponding frequency spectrum blocks of frame before described and present frame to decode, the spectral coefficient that does not decode in the prediction present frame frequency spectrum blocks.
9. device according to claim 8 is characterized in that, described division unit comprises:
First divides module, be used for corresponding before the spectral coefficient place frequency that goes out of frame decoding, present frame and before the scope of each N frequency marks off a frequency spectrum blocks in front and back centered by this frequency respectively in the frame, wherein, N 〉=1; And/or,
Second divides module, if be used for before frame decodes plural spectral coefficient in an adjacent N frequency point ranges, then from the frequency of described plural spectral coefficient place, select arbitrary frequency, present frame and before the scope of each N frequency marks off a frequency spectrum blocks in front and back centered by the frequency of this selection respectively in the frame, wherein, N 〉=1.
10. according to claim 8 or 9 described devices, it is characterized in that, described judging unit comprises:
The frame judgment sub-unit, the present frame of dividing according to described division unit and the correlativity of frame frequency spectrum blocks before judge whether present frame needs to do inter prediction;
The piece judgment sub-unit is used for described frame judge module is judged the present frame that need to do inter prediction, judges one by one whether the frequency spectrum blocks that present frame is divided needs to do inter prediction.
11. device according to claim 10 is characterized in that, described frame judgment sub-unit comprises:
Position judging module, the number that is used for the frequency spectrum blocks of the spectral coefficient that decodes according to present frame is judged, if the number of the frequency spectrum blocks of the spectral coefficient that decodes in the present frame, is then judged present frame more than or equal to predefined numerical value and need to be done inter prediction; And/or,
The energy judge module, be used for the frequency spectrum blocks of the spectral coefficient that decodes according to present frame and before in the frame energy information of corresponding frequency spectrum blocks judge, if the number of corresponding frequency spectrum blocks energy equivalence, is then judged present frame more than or equal to predefined numerical value and need to be done inter prediction.
12. device according to claim 11, it is characterized in that, described judging unit specifically is used for, according to present frame with have or not the spectral coefficient that decodes to judge in the corresponding frequency spectrum blocks of frame before, if the spectral coefficient that does not decode in the frequency spectrum blocks of present frame, and in the corresponding frequency spectrum blocks of frame the spectral coefficient that decodes is arranged before, judge that then this frequency spectrum blocks of present frame need to be done inter prediction.
13. device according to claim 12, it is characterized in that, described predicting unit specifically is used for, amplitude with the spectral coefficient place frequency that decodes in the corresponding frequency spectrum blocks of frame before described, the δ of the minimum radius in all spectral coefficients that decoded with present frame doubly is weighted summation, with the result of weighted sum as the amplitude that does not decode the corresponding frequency of spectral coefficient in the described present frame frequency spectrum blocks, the symbol of spectral coefficient is identical with the symbol of the described before corresponding frequency of frame, wherein 0≤δ≤1.
14. device according to claim 8 is characterized in that, described device also comprises:
The output storage unit is used for the spectral coefficient of the present frame after the prediction of output is processed, and preserves the spectral coefficient of the present frame before the prediction processing as the frame information of frame before the next frame.
15. a demoder is characterized in that, comprises prediction unit and the transducer of each described frequency domain pulse decoding of claim 8 to 14,
The prediction unit of described frequency domain pulse decoding, be used for judging the present frame frequency spectrum blocks that to do inter prediction, the spectral coefficient that the spectral coefficient that decodes in the corresponding frequency spectrum blocks of frame before using and present frame have decoded, the spectral coefficient that does not decode in the prediction present frame frequency spectrum blocks;
Described transducer is used for carrying out frequency domain to the conversion of time domain, the output time-domain sound signal according to the frame frequency spectral coefficient after the prediction unit prediction of described frequency domain pulse decoding.
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/CN2009/071161 WO2010111841A1 (en) | 2009-04-03 | 2009-04-03 | Predicting method and apparatus for frequency domain pulse decoding and decoder |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102246229A CN102246229A (en) | 2011-11-16 |
| CN102246229B true CN102246229B (en) | 2013-03-27 |
Family
ID=42827473
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2009801486921A Active CN102246229B (en) | 2009-04-03 | 2009-04-03 | Predicting method and apparatus for frequency domain pulse decoding and decoder |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN102246229B (en) |
| WO (1) | WO2010111841A1 (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1145512A (en) * | 1995-06-20 | 1997-03-19 | 索尼公司 | Method and apparatus for reproducing speech signals and method for transmitting same |
| CN1504993A (en) * | 2002-11-29 | 2004-06-16 | ���ǵ�����ʽ���� | Audio decoding method and apparatus for reconstructing high frequency components with less computation |
| US7003448B1 (en) * | 1999-05-07 | 2006-02-21 | Fraunhofer-Gesellschaft Zur Foerderung Der Angewandten Forschung E.V. | Method and device for error concealment in an encoded audio-signal and method and device for decoding an encoded audio signal |
| JP3798681B2 (en) * | 2001-11-14 | 2006-07-19 | 日本電信電話株式会社 | Speech spectrum estimation method, apparatus thereof, program thereof, and recording medium thereof |
| CN1813286A (en) * | 2004-01-23 | 2006-08-02 | 微软公司 | Efficient coding of digital media spectral data using wide-sense perceptual similarity |
| US7684981B2 (en) * | 2005-07-15 | 2010-03-23 | Microsoft Corporation | Prediction of spectral coefficients in waveform coding and decoding |
-
2009
- 2009-04-03 WO PCT/CN2009/071161 patent/WO2010111841A1/en active Application Filing
- 2009-04-03 CN CN2009801486921A patent/CN102246229B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1145512A (en) * | 1995-06-20 | 1997-03-19 | 索尼公司 | Method and apparatus for reproducing speech signals and method for transmitting same |
| US7003448B1 (en) * | 1999-05-07 | 2006-02-21 | Fraunhofer-Gesellschaft Zur Foerderung Der Angewandten Forschung E.V. | Method and device for error concealment in an encoded audio-signal and method and device for decoding an encoded audio signal |
| JP3798681B2 (en) * | 2001-11-14 | 2006-07-19 | 日本電信電話株式会社 | Speech spectrum estimation method, apparatus thereof, program thereof, and recording medium thereof |
| CN1504993A (en) * | 2002-11-29 | 2004-06-16 | ���ǵ�����ʽ���� | Audio decoding method and apparatus for reconstructing high frequency components with less computation |
| CN1813286A (en) * | 2004-01-23 | 2006-08-02 | 微软公司 | Efficient coding of digital media spectral data using wide-sense perceptual similarity |
| US7684981B2 (en) * | 2005-07-15 | 2010-03-23 | Microsoft Corporation | Prediction of spectral coefficients in waveform coding and decoding |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2010111841A1 (en) | 2010-10-07 |
| CN102246229A (en) | 2011-11-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP2047457B1 (en) | Systems, methods, and apparatus for signal change detection | |
| CN102436820B (en) | High frequency band signal coding and decoding methods and devices | |
| CN103325377B (en) | audio coding method | |
| US9779749B2 (en) | Audio signal coding method and apparatus | |
| RU2009107043A (en) | SYSTEMS, METHODS AND DEVICE FOR WIDEBAND CODING AND DECODING OF INACTIVE FRAMES | |
| CN104966517A (en) | Voice frequency signal enhancement method and device | |
| CN100578618C (en) | Decoding method and device | |
| US10170128B2 (en) | Method and apparatus for processing temporal envelope of audio signal, and encoder | |
| CN104217727A (en) | Signal encoding method and device | |
| CN105225670A (en) | A kind of audio coding method and device | |
| CN104299614A (en) | Decoding method and decoding device | |
| RU2651184C1 (en) | Method of processing a speech/audio signal and apparatus | |
| CN103456307B (en) | In audio decoder, the spectrum of frame error concealment replaces method and system | |
| CN101770778B (en) | Pre-emphasis filter, perception weighted filtering method and system | |
| CN101308651B (en) | Detection method of audio transient signal | |
| CN102246229B (en) | Predicting method and apparatus for frequency domain pulse decoding and decoder | |
| CN101853664A (en) | Signal denoising method and device and audio decoding system | |
| JP2002366195A (en) | Method and device for encoding voice and parameter | |
| US20220208201A1 (en) | Apparatus and method for comfort noise generation mode selection | |
| CN101609677A (en) | A kind of preprocess method, device and encoding device | |
| WO2009025441A1 (en) | Method and apparatus for encoding continuation sinusoid signal information of audio signal and method and apparatus for decoding same | |
| KR20020068585A (en) | Method of reducing a mount of calculation needed for pitch search in vocoder | |
| JPH11119798A (en) | Method of encoding speech and device therefor, and method of decoding speech and device therefor |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20210508 Address after: Unit 3401, unit a, building 6, Shenye Zhongcheng, No. 8089, Hongli West Road, Donghai community, Xiangmihu street, Futian District, Shenzhen, Guangdong 518040 Patentee after: Honor Device Co.,Ltd. Address before: 518129 Bantian HUAWEI headquarters office building, Longgang District, Guangdong, Shenzhen Patentee before: HUAWEI TECHNOLOGIES Co.,Ltd. |
|
| TR01 | Transfer of patent right |