CN107895580A - The method for reconstructing and device of a kind of audio signal - Google Patents
The method for reconstructing and device of a kind of audio signal Download PDFInfo
- Publication number
- CN107895580A CN107895580A CN201610879165.XA CN201610879165A CN107895580A CN 107895580 A CN107895580 A CN 107895580A CN 201610879165 A CN201610879165 A CN 201610879165A CN 107895580 A CN107895580 A CN 107895580A
- Authority
- CN
- China
- Prior art keywords
- compressed data
- audio signal
- passage
- frequency coefficient
- signal
- 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.)
- Granted
Links
- 230000005236 sound signal Effects 0.000 title claims abstract description 402
- 238000000034 method Methods 0.000 title claims abstract description 78
- 230000006835 compression Effects 0.000 claims abstract description 43
- 238000007906 compression Methods 0.000 claims abstract description 43
- 238000005070 sampling Methods 0.000 claims abstract description 41
- 239000011159 matrix material Substances 0.000 claims description 115
- 238000004364 calculation method Methods 0.000 claims description 96
- 238000006243 chemical reaction Methods 0.000 claims description 18
- 238000005259 measurement Methods 0.000 claims description 18
- 238000013139 quantization Methods 0.000 claims description 7
- 238000004891 communication Methods 0.000 abstract description 21
- 238000004422 calculation algorithm Methods 0.000 description 38
- 230000008569 process Effects 0.000 description 28
- 238000012545 processing Methods 0.000 description 24
- 238000013461 design Methods 0.000 description 21
- 230000006870 function Effects 0.000 description 20
- 230000009466 transformation Effects 0.000 description 10
- 230000008859 change Effects 0.000 description 7
- 230000005540 biological transmission Effects 0.000 description 5
- 238000000205 computational method Methods 0.000 description 3
- 238000000802 evaporation-induced self-assembly Methods 0.000 description 3
- 230000002093 peripheral effect Effects 0.000 description 3
- 241000209140 Triticum Species 0.000 description 2
- 235000021307 Triticum Nutrition 0.000 description 2
- 230000009471 action Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000012562 intraclass correlation Methods 0.000 description 2
- 238000001228 spectrum Methods 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000012634 fragment Substances 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000009527 percussion Methods 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000004549 pulsed laser deposition Methods 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000011282 treatment Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; 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
- G10L19/008—Multichannel audio signal coding or decoding using interchannel correlation to reduce redundancy, e.g. joint-stereo, intensity-coding or matrixing
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; 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
- G10L19/02—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 spectral analysis, e.g. transform vocoders or subband vocoders
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/03—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/03—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters
- G10L25/06—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters the extracted parameters being correlation coefficients
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Signal Processing (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Mathematical Physics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
The embodiment of the present invention provides a kind of method for reconstructing and terminal of audio signal, is related to the communications field, can solve the problem that signal reconstruction low precision and slow-footed problem.Its method is:When being compressed sampling at least two audio signals, at least two audio signals are grouped according to the correlation between at least two audio signals, and grouping information is passed into remote terminal, remote terminal can the compressed data according to corresponding to grouping information by least two audio signals be grouped, in signal reconstruction using concurrent reconstruction between packet, inline occlusal reconstruction is grouped.The embodiment of the present invention is used for compression sampling and the reconstruction of audio signal.
Description
Technical field
The present invention relates to the method for reconstructing and device of the communications field, more particularly to a kind of audio signal.
Background technology
Audio signal compression sample with rebuild during, what algorithm for reconstructing needed to meet simultaneously precision and speed will
Ask, precision is excessively poor or speed can not all meet actual application demand slowly excessively.The compression sampling of signal is by by primary signal
Be multiplied realization with a calculation matrix, and calculation matrix needs to pass to signal reconstruction end to realize the recovery of compressed signal.With biography
The audio coding decoding scheme of system is similar, and the compression sampling of audio signal is also to be carried out with " frame " for unit.
For the special circumstances of multi-microphone array signal transacting, the computing speed of the mass data of multichannel to algorithm for reconstructing
Degree proposes higher requirement.Meanwhile due to strong correlation be present between the reception signal of each passage, it is also smart to reconstruction is improved
Degree brings more possibilities.For the voice data of multichannel compression sampling, can be improved by way of parallel processing
Arithmetic speed.But between the parallel unit of Parallel processing demands independently of each other, it means that between passage and be about to cause passage
Between correlation can not be utilized so that reconstruction precision is limited.Conversely, carrying out joint reconstruction to multiple passages will lead
Cause to intercouple between passage, parallel acceleration can not be realized.
The content of the invention
The embodiment of the present invention provides a kind of method for reconstructing and terminal of audio signal, can solve the problem that signal reconstruction low precision and
The problem of interchannel can not accelerate parallel.
On the one hand, there is provided a kind of method for reconstructing of audio signal, including:Obtain at least two audios of at least two passages
Compressed data corresponding to signal, at least two passages and at least two audio signals correspond;Obtain at least two audios letter
The grouping information of group where passage corresponding to number;According to grouping information, by compressed data corresponding at least two audio signals
It is grouped, so as to obtain compressed data group;Calculation matrix is obtained, according to the compressed data in compressed data group and measurement square
Battle array, frequency coefficient corresponding to the compressed data that joint is rebuild in compressed data group;The change of frequency domain to time domain is carried out to frequency coefficient
Change, so as to obtain audio signal corresponding to the compressed data in compressed data group.Then, the reconstruction end for audio signal is come
Say, can be according at least two audios after compressed data corresponding at least two audio signals of at least two passages is received
Compressed data corresponding at least two audio signals is grouped by the grouping information of group where passage corresponding to signal, so may be used
Joint reconstruction is carried out to the compressed data in obtained compressed data group, a precision for the inline occlusal reconstruction of group can be lifted, entered between each group
Row joint, which is rebuild, can lift the speed that joint is rebuild.
In a kind of possible design, methods described also includes:Obtain the language musical sound label letter of at least two audio signals
Breath, language musical sound label information are used to indicate that at least two audio signals are voice signal or note signal;It is logical to obtain at least two
Compressed data corresponding at least two audio signals in road includes:According to language musical sound label information, language musical sound label information is obtained
Corresponding frame length;According to frame length, measurement data corresponding at least two audio signals is extracted;Inverse quantization is carried out to measurement data,
So as to obtain compressed data corresponding at least two audio signals.Then, it is relatively slow for this time-varying characteristics of note signal
Signal, on the one hand, the precision that can be rebuild by increasing signal frame length come promotion signal, on the other hand, for equal length
Signal, the increase of signal frame length reduces the quantity for needing signal frame to be processed, also reduce further signal processing algorithm
Run time.
In a kind of possible design, according to the compressed data and calculation matrix in compressed data group, joint rebuilds compression
Frequency coefficient corresponding to compressed data in data group includes:According to compressed data pair corresponding to a passage in compressed data group
Compressed data and calculation matrix corresponding to another passage, calculate compressed data group in the frequency coefficient answered, compressed data group
Frequency coefficient corresponding to compressed data corresponding to another interior passage.Its computational methods can be approximate information transmission
(Approximate Message Passing, AMP) algorithm, or other algorithms, the application do not limit, can be with
Obtain frequency coefficient corresponding to the higher compressed data of precision.
In a kind of possible design, the frequency domain system corresponding to compressed data according to corresponding to a passage in compressed data group
Compressed data and calculation matrix corresponding to another passage in number, compressed data group, calculate interior another of compressed data group and lead to
Frequency coefficient includes corresponding to compressed data corresponding to road:According to compressed data pair corresponding to i-th of passage in compressed data group
Compressed data and calculation matrix corresponding to i+1 passage in the frequency coefficient answered, compressed data group, it is logical to calculate i+1
Frequency coefficient corresponding to compressed data corresponding to road, until compression number corresponding to k-th of passage is calculated in compressed data group
According to corresponding frequency coefficient, i is the positive integer less than k, and k is the total number of channels in compressed data group.It is for same compression
Compressed data in data group, joint reconstruction can be carried out to the compressed data of each passage in packet, lift frequency coefficient
Precision.
In a kind of possible design, method also includes:According to compressed data corresponding to j-th of passage in compressed data group
Compressed data and calculation matrix corresponding to -1 passage of jth in corresponding frequency coefficient, compressed data group, calculate jth -1
Frequency coefficient corresponding to compressed data corresponding to passage, compressed until being calculated in compressed data group corresponding to the 1st passage
Frequency coefficient corresponding to data, j are the positive integer less than or equal to k, and more than 1.That is, for same compression number
For the compressed data in group, it can be calculated since i-th of passage up to obtaining compressed data pair corresponding to k-th of passage
The frequency coefficient answered, then calculated since k-th of passage up to obtaining frequency domain system corresponding to compressed data corresponding to j-th of passage
Number, the process are properly termed as an algorithm iteration, until frequency coefficient corresponding to obtained compressed data reaches preset requirement, i.e.,
Joint reconstruction is carried out by the compressed data in group and obtains the frequency coefficient of the packet.
In a kind of possible design, method also includes:It is corresponding according to default initialization frequency coefficient, the 1st passage
Compressed data and calculation matrix, calculate the 2nd frequency coefficient corresponding to compressed data corresponding to passage.That is compressed data group
The 1st interior frequency coefficient corresponding to passage is preset.
In a kind of possible design, the frequency domain system corresponding to compressed data according to corresponding to a passage in compressed data group
Compressed data and calculation matrix corresponding to another passage in number, compressed data group, calculate interior another of compressed data group and lead to
Frequency coefficient includes corresponding to compressed data corresponding to road:The frequency coefficient corresponding to compressed data according to corresponding to a passage,
Determine priori frequency coefficient corresponding to compressed data corresponding to another passage;Using priori frequency coefficient as another passage pair
The priori of frequency coefficient corresponding to the compressed data answered, and compressed data and calculation matrix according to corresponding to another passage,
Calculate frequency coefficient corresponding to compressed data corresponding to another passage in compressed data group.That is, a upper passage is obtained
Frequency coefficient corresponding to the compressed data arrived is as priori frequency coefficient corresponding to compressed data corresponding to next passage, with elder generation
It is priori to test frequency coefficient, calculates frequency coefficient corresponding to compressed data corresponding to next passage, to obtain the compressed data
The higher frequency coefficient of group precision.
On the other hand, there is provided a kind of compressive sampling method of audio signal, including:Obtain at least the two of at least two passages
Individual audio signal, at least two passages and at least two audio signals correspond;Calculate between at least two audio signals
Correlation, at least two audio signals are grouped according to correlation, so as to obtain the place of the passage at least two passages
Group grouping information;At least two audio signals are carried out with conversion of the time domain to frequency domain, so as to obtain at least two groups of frequency domain systems
Number, at least two groups of frequency coefficients and at least two audio signals correspond;Calculation matrix is obtained, according to calculation matrix at least
Two groups of frequency coefficients are sampled, so as to obtain compressed data corresponding at least two audio signals.So after compression sampling
Compressed data when transmitting to the reconstructing device of audio signal, reconstructing device can be according to the grouping information that compressed data carries to extremely
The compressed data of few two audio signals is grouped, in order to which the compressed data of the passage high to correlation carries out joint weight
Build, concurrent reconstruction between packet, obtain the audio signal of at least two passages, the accuracy and speed rebuild with promotion signal.
In a kind of possible design, before calculation matrix is obtained, method also includes:Determine at least two audio signals
Language musical sound label information, language musical sound label information be used for indicate at least two audio signals be voice signal or note signal;
According to language musical sound label information, the frame length of at least two audio signals is determined.Then, for this time-varying characteristics phase of note signal
To slow signal, on the one hand, the precision that can be rebuild by increasing signal frame length come promotion signal, on the other hand, for phase
With the signal of length, the increase of signal frame length reduces the quantity for needing signal frame to be processed, also reduce further at signal
The run time of adjustment method.
In a kind of possible design, obtaining calculation matrix includes:According to frame length, calculation matrix corresponding to frame length is obtained.
That is, for voice signal and note signal, corresponding calculation matrix can be generated according to different frame lengths.Such as believe for musical sound
Number, musical sound structuring calculation matrix can be generated, for voice signal, voice signal structuring calculation matrix can be generated.
In a kind of possible design, the correlation between at least two audio signals is calculated, according to correlation at least
Two audio signals, which carry out packet, to be included:The first audio signal at least two audio signals is obtained, acquisition removes the first audio
In remaining outer audio signal of signal with m audio signal before the first audio signal correlation highest, and by the first audio signal
With as one group of audio signal, m is more than or equal to 1 just with m audio signal before the first audio signal correlation highest
Integer;From except the first audio signal and with remaining audio signal in addition to m audio signal before the first audio signal correlation highest
In continue choose the second audio signal and obtain except the first audio signal, the second audio signal and with the first audio signal correlation
Before highest in the outer rest channels of m audio signal with m passage before the second audio signal correlation highest, and by the second sound
Frequency signal and with the and m audio signal is as another group of audio signal before two audio signal correlation highests, until at least
The packet of two audios is completed.Wherein, the correlation calculated between two audio signals can be obtained by Euclidean distance algorithm,
It can also obtain in other way, the application does not limit.
In a kind of possible design, correlation between at least two audio signals include at least two audio signals it
Between distance.That is, the correlation between two audio signals can be understood as the correlation of audio signal spatially.
Another further aspect, there is provided a kind of reconstructing device of audio signal, including:Acquiring unit, it is logical for obtaining at least two
The compressed data of at least two audio signals in road, at least two passages and at least two audio signals correspond;Obtain single
Member, it is additionally operable to obtain the grouping information of the group corresponding at least two audio signals where passage;Grouped element, for basis point
Group information, the compressed data of at least two audio signals is grouped, so as to obtain compressed data group;Reconstruction unit, it is used for
Calculation matrix is obtained, the compression rebuild according to the compressed data and calculation matrix in compressed data group, joint in compressed data group
Frequency coefficient corresponding to data;Converter unit, for carrying out the conversion of frequency domain to time domain to frequency coefficient, so as to obtain compression number
According to audio signal corresponding to the compressed data in group.
In a kind of possible design, acquiring unit is additionally operable to:Obtain the language musical sound label letter of at least two audio signals
Breath, language musical sound label information are used to indicate that at least two audio signals are voice signal or note signal;Acquiring unit, it is used for:
According to language musical sound label information, frame length corresponding to language musical sound label information is obtained;According to frame length, at least two audio signals are extracted
Corresponding measurement data;Inverse quantization is carried out to measurement data, so as to obtain compressed data corresponding at least two audio signals.
In a kind of possible design, reconstruction unit is used for:Number is compressed according to corresponding to a passage in compressed data group
According to compressed data and calculation matrix corresponding to another passage in corresponding frequency coefficient, compressed data group, compression number is calculated
According to frequency coefficient corresponding to compressed data corresponding to another passage in group.
In a kind of possible design, reconstruction unit is used for:Number is compressed according to corresponding to i-th of passage in compressed data group
According to compressed data and calculation matrix corresponding to i+1 passage in corresponding frequency coefficient, compressed data group, i+1 is calculated
Frequency coefficient corresponding to compressed data corresponding to individual passage, pressed until being calculated in compressed data group corresponding to k-th of passage
Frequency coefficient corresponding to contracting data, i are the positive integer less than k, and k is the total number of channels in compressed data group.
In a kind of possible design, reconstruction unit is additionally operable to:The compression according to corresponding to j-th of passage in compressed data group
Frequency coefficient corresponding to data, compressed data and calculation matrix corresponding to -1 passage of jth in compressed data group, calculate jth -
Frequency coefficient corresponding to compressed data corresponding to 1 passage, pressed until being calculated in compressed data group corresponding to the 1st passage
Frequency coefficient corresponding to contracting data, j are the positive integer less than or equal to k, and more than 1.
In a kind of possible design, reconstruction unit is additionally operable to:According to default initialization frequency coefficient, the 1st passage
Corresponding compressed data and calculation matrix, calculate the 2nd frequency coefficient corresponding to compressed data corresponding to passage.
In a kind of possible design, reconstruction unit is used for:The frequency domain corresponding to compressed data according to corresponding to a passage
Coefficient, determine priori frequency coefficient corresponding to compressed data corresponding to another passage;It regard priori frequency coefficient as another
The priori of frequency coefficient corresponding to compressed data corresponding to passage, and compressed data and measurement according to corresponding to another passage
Matrix, calculate frequency coefficient corresponding to compressed data corresponding to another passage in compressed data group.
Another aspect, there is provided a kind of compression sampling device of audio signal, including:Acquiring unit, for obtaining at least two
At least two audio signals of individual passage, at least two passages and at least two audio signals correspond;Grouped element, it is used for
The correlation between at least two audio signals is calculated, at least two audio signals are grouped according to correlation, so as to
To the grouping information of the group where the passage at least two passages;Converter unit, for being carried out at least two audio signals
Time domain to frequency domain conversion, so as to obtain at least two groups of frequency coefficients, at least two groups of frequency coefficients and at least two audio signals
Correspond;Acquiring unit, it is additionally operable to obtain calculation matrix;Sampling unit, for according to calculation matrix at least two groups of frequency domains
Coefficient is sampled, so as to obtain compressed data corresponding at least two audio signals.
In a kind of possible design, in addition to determining unit, it is used for:Determine the language musical sound mark of at least two audio signals
Information is signed, language musical sound label information is used to indicate that at least two audio signals are voice signal or note signal;According to language musical sound
Label information, determine the frame length of at least two audio signals.
In a kind of possible design, acquiring unit is used for:According to frame length, calculation matrix corresponding to frame length is obtained.
In a kind of possible design, grouped element is used for:The first audio signal at least two audio signals is obtained,
Obtain in addition to the first audio signal in remaining audio signal with m audio signal before the first audio signal correlation highest, and
As one group of audio signal, m it is big using the first audio signal and with m audio signal before the first audio signal correlation highest
In or equal to 1 positive integer;From except the first audio signal and with addition to m audio signal before the first audio signal correlation highest
Continue in remaining audio signal choose the second audio signal and obtain except the first audio signal, the second audio signal and with the first sound
It is logical with m before the second audio signal correlation highest in the outer rest channels of m audio signal before frequency signal correlation highest
Road, and using and audio signal and with and m audio signal is believed as another group of audio before audio signal correlation highest
Number, until the packet of at least two audios is completed.
In a kind of possible design, correlation between at least two audio signals include at least two audio signals it
Between distance.
Thus one, in embodiments of the present invention, the compression sampling device of audio signal is according at least two audio signals
Between correlation at least two audio signals are grouped, so as to obtain point of group where the passage at least two passages
Group information, then at least two audio signals are carried out with conversion of the time domain to frequency domain, so as to obtain at least two groups of frequency coefficients, extremely
Few two groups of frequency coefficients and at least two audio signals correspond, and calculation matrix are obtained, according to calculation matrix at least two groups
Frequency coefficient is sampled, so as to obtain compressed data corresponding at least two audio signals, so by least two passages
Compressed data corresponding at least two audio signals can carry at least two audios letter when transmitting to the reconstructing device of audio signal
The grouping information of group where passage corresponding to number, so that reconstructing device can be according to grouping information by least two audio signals
Compressed data is grouped, so as to obtain compressed data group, so as to according to the compressed data and calculation matrix in compressed data group,
Frequency coefficient corresponding to the compressed data that joint is rebuild in compressed data group, the change of frequency domain to time domain is then carried out to frequency coefficient
Change, so as to obtain audio signal corresponding to the compressed data in compressed data group, that is to say, that reconstructing device is carrying out joint weight
Building is, is to carry out joint reconstruction to the compressed data in compressed data group, because compression sampling device is carried out to audio signal
It is that at least two audio signals are grouped according to the correlation between at least two audio signals during packet, that is to say, that
The strong audio signal of correlation is divided into one group, so, what reconstructing device can be strong according to intra class correlation in reconstructed audio signals
Compressed data carries out joint reconstruction, can lift the precision of audio signal reconstruction, and multiple packets can carry out joint reconstruction parallel,
So as to lift the speed that joint is rebuild.
Brief description of the drawings
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below will be in embodiment or description of the prior art
The required accompanying drawing used is briefly described, it should be apparent that, drawings in the following description are only some realities of the present invention
Example is applied, for those of ordinary skill in the art, on the premise of not paying creative work, can also be according to these accompanying drawings
Obtain other accompanying drawings.
Fig. 1 is a kind of schematic diagram of remote phone conference system provided in an embodiment of the present invention;
Fig. 2 is a kind of schematic flow sheet of the compressive sampling method of audio signal provided in an embodiment of the present invention;
Fig. 3 is a kind of schematic flow sheet of the method for reconstructing of audio signal provided in an embodiment of the present invention;
Fig. 4 is the schematic flow sheet of a kind of audio signal compression provided in an embodiment of the present invention and method for reconstructing;
Fig. 5 is a kind of structural representation of terminal provided in an embodiment of the present invention;
Fig. 6 is a kind of structural representation of terminal provided in an embodiment of the present invention;
Fig. 7 is a kind of structural representation of terminal provided in an embodiment of the present invention;
Fig. 8 is a kind of structural representation of computing device provided in an embodiment of the present invention;
Fig. 9 is a kind of structural representation of computing device provided in an embodiment of the present invention;
Figure 10 is a kind of structural representation of computing device provided in an embodiment of the present invention.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Site preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, rather than whole embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art are obtained every other under the premise of creative work is not made
Embodiment, belong to the scope of protection of the invention.
In embodiments of the present invention, audio signal compression sampling can be applied to plurality of application scenes with reconstruction, such as remotely
TeleConference Bridge, as shown in figure 1, the system may include the computing device and remote terminal for including microphone array, this includes
The computing device of microphone array can be compressed sampling to audio signal corresponding to speaker, and pass through wired or wireless side
Formula is transmitted to remote terminal, remote terminal and the data that received can be rebuild, and original audio signal is obtained, so as to wheat
Gram user of wind array side and the user of remote terminal side carry out videoconference in real time.
In embodiments of the present invention, microphone array can be one group of microphone for arranging placement at a certain distance, pass through
Sound wave arrives at the interaction of the small time difference between each microphone in array, and microphone array can be obtained than single wheat
Gram wind preferably directive property.Computing device can include at least two microphones, sound source processing module and voice data output mould
Block.Sound source processing module is used to be compressed the audio that microphone collects sampling, and voice data output module is used for pressure
Data after contracting sampling are transmitted to remote terminal after being quantified.Remote terminal with computing device communication can be PC
(Personal Computer, PC), smart mobile phone, multimedia terminal etc..
During the joint reconstruction of the invention for Solving Multichannel the problem of signal reconstruction low precision, it is proposed that a kind of audio signal
Compressive sampling method, multiple audio signals are entered by the correlation between at least two audio signals of at least two passages
Row packet, so as to obtain the grouping information organized where the passage at least two passages, in order to carry out signal in remote terminal
It will carry out joint reconstruction during reconstruction between the high audio signal of intra class correlation, the precision that promotion signal is rebuild, correspondingly, this hair
Bright embodiment also provides a kind of method for reconstructing of audio signal, and terminal is believed at least two audios for receiving at least two passages
Number compressed data when, the compressed data of at least two audio signals can be grouped according to grouping information, with to each group
Interior compressed data carries out joint reconstruction, because correlation is high between audio signal in group, can effective promotion signal reconstruction precision,
It is multiple packet channel parallels rebuild, can promotion signal reconstruction speed.
The embodiment of the present invention provides a kind of compressive sampling method of audio signal, as shown in Fig. 2 including:
201st, computing device obtains at least two audio signals of at least two passages, at least two passages and at least two
Audio signal corresponds.
Microphone array in computing device can collect at least two audios letter of at least two passages when people speaks
Number, passage corresponds with audio signal.
202nd, computing device calculates the correlation between at least two audio signals, according to correlation at least two audios
Signal is grouped, so as to obtain the grouping information of the group where the passage at least two passages.
Such as the relative time delay between the passage of multi-microphone array is different so that between the audio signal of different passages
Degree of correlation has difference, also causes the audio signal joint reconstruction precision of all passages to be affected.Therefore, the present invention can be from
The correlation between a frame data Acquisition channel is respectively taken in the audio signal of each passage, and then can be by the big sound of strength of correlation
Frequency signal is divided into one group, it can be understood as multi-microphone array is divided into multiple subarrays, so as to combine rebuild when in son
The inline occlusal reconstruction of array, the parallel essence accelerated the audio signal joint reconstruction between subarray internal channel between subarray can be lifted
Degree, while multiple subarrays are rebuild, the speed that joint is rebuild can be lifted.
Grouping information can distinguish different packets with different marks, i.e., each passage corresponds to point of the group belonging to itself
Group mark.
203rd, computing device carries out conversion of the time domain to frequency domain at least two audio signals, so as to obtain at least two groups frequencies
Domain coefficient, at least two groups of frequency coefficients and at least two audio signals correspond.
Computing device to a frame data of audio signal corresponding to each passage can transform from the time domain to the conversion of frequency domain,
Frequency coefficient corresponding to a frame data is obtained, a frame data corresponding to the difference of such at least two audio signal convert from time domain
To frequency domain, at least two groups of frequency coefficients are can obtain.This is due to the shape that time domain can intuitively observe signal, but can not
Signal is accurately described with limited parameter, and sophisticated signal can be decomposed into the folded of simple signal by frequency-domain analysis
Add, can more accurately understand " construction " of signal.Modified Discrete Cosine Transform (Modified can specifically be passed through
Discrete Cosine Transform, MDCT) frame data of audio signal transform from the time domain to frequency domain by algorithm, also may be used
To use other algorithms, the application does not limit.
204th, computing device obtains calculation matrix, and at least two groups of frequency coefficients are sampled according to calculation matrix, so as to
Obtain compressed data corresponding at least two audio signals.
Compressed data can be understood as the data after compression sampling.
Computing device can determine the columns of calculation matrix to be generated according to default frame length, such as frame length is 4096,
The columns of calculation matrix so to be generated is 4096, and the row of calculation matrix to be generated is would know that according to default compression ratio
Number, such as default compression ratio are 1/3, then the line number of calculation matrix to be generated is multiplied by 1/3 for columns 4096 and rounds to obtain, and
Afterwards, the type further according to default calculation matrix and the line number and columns that are obtained generate calculation matrix.Then, computing device can
Sampling is compressed so that calculation matrix to be multiplied with least two groups of frequency coefficients, obtains the compression number of at least two audio signals
According to.
Then, compressed data corresponding at least two audio signals that are obtained after sampling can be quantified, the amount of obtaining
Value after change.Wherein quantify to be with the range value of the approximate consecutive variations originally of limited individual range value, the continuous width of analog signal
Degree is changed into the centrifugal pump for having certain intervals of limited quantity, so as to encode the value after quantization, obtains being used to transmit
Signal, transmit to remote terminal.
Therefore, the compressive sampling method of audio signal provided in an embodiment of the present invention, can be in compression sampling according to sound
At least two audio signals are grouped by the correlation between frequency signal, group where obtaining the passage at least two passages
Grouping information, it can cause audio signal in reconstructing device according to concurrent reconstruction between grouping information implementation group, due in group
Audio signal correlation is high, and the audio signal in group can combine reconstruction, so as to lift the speed and precision of audio signal reconstruction.
On the basis of the sampling of above-mentioned audio signal compression, the embodiment of the present invention provides a kind of reconstruction side of audio signal
Method, as shown in figure 3, after above-mentioned steps 204, this method also includes:
205th, terminal obtains compressed data corresponding at least two audio signals of at least two passages, at least two passages
Corresponded with least two audio signals.
When collecting sound with the microphone array in the computing device of terminal wireless or wired connection, terminal can receive
To computing device send at least two passages at least two audio signals corresponding to compressed data, at least two passages with extremely
Few two audio signals correspond.Compressed data can be understood as the data after compression sampling, and terminal is in the audio received
, it is necessary to carry out inverse quantization to the measurement data of audio signal during the data of signal, the sampling of at least two audio signal compressions is obtained
Data afterwards, i.e. compressed data.
206th, terminal obtains the grouping information of the group where passage corresponding at least two audio signals.
The parameter for reconstructed audio signals can be carried at least audio signal of at least two passages, parameter can wrap
Include the grouping information of the signal frame length of audio signal, calculation matrix, sparse base and passage.
Wherein signal frame length is frame length corresponding to a frame data of each passage, and calculation matrix is that audio signal is being compressed
The matrix for compression sampling generated in sampling process, sparse base be carried out from time domain to frequency domain during compression sampling it is sparse
Algorithm used in conversion, the grouping information of passage are determined according to the correlation between audio signal during compression sampling
The packet situation of audio signal, the mark of the group corresponding at least two audio signals where passage can be included in grouping information
Know.
207th, the compressed data of at least two audio signals is grouped, so as to be pressed by terminal according to grouping information
Contracting data group.
Terminal can according to corresponding at least two audio signals the mark of group where passage at least two audio signals
Compressed data be grouped, i.e., be divided into one group to the mark identical audio signal of group.
208th, terminal obtains calculation matrix, and according to the compressed data and calculation matrix in compressed data group, joint rebuilds pressure
Frequency coefficient corresponding to compressed data in contracting data group.
Terminal can according to corresponding to a passage in compressed data group frequency coefficient, compressed data corresponding to compressed data
Compressed data and calculation matrix corresponding to another passage, calculate in compressed data group and are compressed corresponding to another passage in group
Frequency coefficient corresponding to data.Computational methods can use approximate information to transmit AMP algorithms, can also use other algorithms, this
Application does not limit, meanwhile, joint reconstruction can be carried out between each compressed data group parallel, i.e., using organizing inline occlusal reconstruction,
The strategy of parallel processing between different groups.
209th, terminal-pair frequency coefficient carries out the conversion of frequency domain to time domain, so as to obtain the compressed data in compressed data group
Corresponding audio signal.
Terminal, can be to the frequency coefficient of the compressed data after the frequency coefficient of the compressed data of obtained each passage
Carry out the conversion of frequency domain to time domain.Such as during compression sampling, the conversion from time domain to frequency domain is whole using MDCT algorithms
Hold the configured information that the sparse base in obtained parameter is MDCT algorithms, then terminal enters line frequency in the frequency coefficient to data
Domain to time domain inverse transformation when then use inverse MDCT algorithms, obtain the audio signal that computing device collects.
Therefore, in the method for reconstructing of audio signal provided in an embodiment of the present invention, terminal can be according to grouping information at least
The compressed data of two audio signals is grouped, to carry out group according to the compressed data in calculation matrix and compressed data group
Between concurrent reconstruction and organize inline occlusal reconstruction, due to concurrent reconstruction between group, and the compressed data in compressed data group has strong phase
Guan Xing, the speed and precision that can be rebuild with promotion signal.
Embodiments of the invention are further described below, the embodiment of the present invention provides a kind of compression of audio signal
Sampling and method for reconstructing, by taking the audio signal of k=32 passages as an example, as shown in figure 4, this method includes:
401st, computing device obtains at least two audio signals of at least two passages, at least two passages and at least two
Audio signal corresponds.
Microphone array in computing device can collect at least two audios letter of at least two passages when people speaks
Number, passage corresponds with audio signal.
402nd, computing device determines the language musical sound label information of at least two audio signals, and language musical sound label information is used to refer to
It is voice signal or note signal to show at least two audio signals.
Note signal includes the signal that wind instrument, stringed musical instrument and percussion instrument etc. are sent.
In the embodiment of the present invention, computing device can from 32 passages an optional passage, it is logical from this using frame length as 4096
Road takes a frame audio signal, detects in the frame audio signal whether include phonetic element, if comprising, it is determined that language musical sound label
Information indicates that at least two audio signals are voice signal, if do not included, it is determined that language musical sound label information instruction at least two
Individual audio signal is note signal.
403rd, computing device determines the frame length of at least two audio signals according to language musical sound label information.
Because note signal change is gentle, voice signal change is fast, and it is longer to preset the frame length of note signal, voice letter
Number frame length it is shorter.Then, for the relatively slow signal of this time-varying characteristics of note signal, on the one hand, increase can be passed through
Signal frame length carrys out the precision of promotion signal reconstruction, and on the other hand, for the signal of equal length, the increase of signal frame length reduces
The quantity of signal frame to be processed is needed, also reduce further the run time of signal processing algorithm.
With the frame length MuLen=4096 of note signal, exemplified by the frame length SpLen=1024 of voice signal, if it is determined that language
Musical sound label information indicates that at least two audio signals are note signal, it is determined that the frame length of audio signal is 4096, if really
Attribute musical sound label information indicates that at least two audio signals are voice signal, it is determined that the frame length of audio signal is 1024.
404th, computing device calculates the correlation between at least two audio signals, according to correlation at least two audios
Signal is grouped, so as to obtain the grouping information of the group where the passage at least two passages.
If computing device determines that frame length is 4096, the frame audio that frame length is 4096 is taken to believe respectively from 32 passages
Number, and the correlation between at least two audio signals is calculated according to the every frame audio signal taken.
Exemplary, computing device obtains the first audio signal at least two audio signals, and acquisition removes the first audio
In remaining outer audio signal of signal with m audio signal before the first audio signal correlation highest, and by the first audio signal
With as one group of audio signal, m is more than or equal to 1 just with m audio signal before the first audio signal correlation highest
Integer;From except the first audio signal and with remaining audio signal in addition to m audio signal before the first audio signal correlation highest
In continue choose the second audio signal and obtain except the first audio signal, the second audio signal and with the first audio signal correlation
Before highest in the outer rest channels of m audio signal with m passage before the second audio signal correlation highest, and the sound by the
Frequency signal and with the and m audio signal is as another group of audio signal before audio signal correlation highest, until at least two
The packet of individual audio is completed.
Wherein, the correlation between at least two audio signals includes the distance between at least two audio signals, i.e. sound
Frequency signal is spatially related, and the distance can be Euclidean distance, therefore in the strength of correlation of two audio signals of calculating,
It can be calculated using Euclidean distance formula.
Exemplary, if two audio signals are the first audio signal and the second audio signal, the one of the first audio signal
Frame data x=(X1, X2..., Xn), a frame data of the second audio signal are y=(y1, y2..., yn), then the first audio signal
It is expressed as with the correlation of the second audio signal:
R (x, y) represents the correlation of the first audio signal and the second audio signal, x1, x2... xnRepresent corresponding to the first audio signal
Audio intensity, y corresponding to each time point in one frame data1, y2... ynRepresent each in a frame data corresponding to the second audio signal
Audio intensity corresponding to time point, n represent frame length.
When R (x, y) value is bigger, the strength of correlation of two audio signals of expression is smaller, conversely, representing two audio signals
Strength of correlation it is bigger.
Exemplary, microphone array shows 32 microphones, then corresponds to 32 passages, appoints take one the 1st passage first
One frame audio signal, it is assumed that audio signal is note signal, then the frame length of a frame audio signal is the 4096, the 1st passage
One frame audio signal x=(X1, X2 ..., Xn), n represent frame length, also correspond to the audio intensity at the n time point collected, xn tables
Show the audio intensity at n-th time point of the audio signal of the 1st passage in time domain, similarly, a frame audio of each passage
The frame length of signal is identical, and as frame audio signal y=(y1, y2 ..., a yn) of the 2nd passage, yn represents the 2nd passage
The audio intensity at n-th time point of the audio signal in time domain, the audio of the 1st passage is being obtained according to Euclidean distance formula
Signal and the audio signal and the 3rd passage for after the audio signal correlation of the 2nd passage, continuing to obtain the 1st passage
The correlation of audio signal, until obtaining the audio signal and the audio signal correlation of the 32nd passage of the 1st passage, so
The audio signal of the 1st passage of selection preceding 3 passage most strong with the strength of correlation of the audio signal of other 31 passages afterwards
Audio signal, the 1st passage and preceding 3 passages are divided into one group, then again from the audio signal of remaining 28 passages
In an optional passage audio signal, calculate phase in the audio signal of this passage and the audio signal of remaining 27 passages
The audio signal of closing property most strong preceding 3 passages of intensity, by that analogy, by 32 passages, every 4 are divided into one group, are divided into 8 groups.
It is therefore, each logical when taking because the correlation of audio signal is the degree of correlation between audio signal in the spatial domain
The audio signal of a frame in road and after passage is grouped, the packet of passage determine constant.It is determined that point of passage
During group situation, passage can be indicated group character, to obtain the grouping information of the group where the passage at least two passages.
405th, computing device carries out conversion of the time domain to frequency domain at least two audio signals, so as to obtain at least two groups frequencies
Domain coefficient, at least two groups of frequency coefficients and at least two audio signals correspond.
Computing device can add window letter after the packet situation of passage is got to audio signal corresponding to each passage
Number, that is, adding Hann windows, this is due to be transmitted after audio signal needs to transform from the time domain to frequency domain in transmission to remote terminal, when
When realizing measurement signal in engineering processing, it is impossible to the signal of endless is measured in time domain and computing, but take it to have
The time slice of limit is analyzed, and then carries out periodic extension processing with the signal time fragment of interception, and it is virtual unlimited to obtain
Long signal, then can is to Mathematical treatments such as signal progress correlation analyses.But after the signal of endless is truncated, its
Frequency spectrum is distorted, and can be produced spectrum energy leakage, signal can be blocked using different cutted functions, truncation funcation
Referred to as window function, referred to as window.
Because computing device is that next frame audio letter is reprocessed to a frame Audio Signal Processing and after transmitting to remote terminal
Number, therefore, windowed function here is to carry out adding window to a frame audio signal of each passage., will be each after windowed function
Audio signal after one frame adding window of passage implements the sparse transformation from time domain to frequency domain, obtains dilute corresponding to a frame audio signal
Dredge conversion coefficient vector, i.e. frequency coefficient.The corresponding one group of frequency coefficient of each frame audio signal, i.e. at least two frequency coefficients with
At least two audio signals correspond.Here in addition to it can use the frequency coefficient that MDCT algorithms obtain as MDCT coefficients,
Other algorithms, such as wavelet transform (Discrete Wavelet Transform, DWT) etc., this Shen can also be used
Do not limit please.
406th, computing device obtains calculation matrix, and at least two groups of frequency coefficients are sampled according to calculation matrix, so as to
Obtain compressed data corresponding at least two audio signals.
Computing device can determine the row of calculation matrix to be generated, such as audio signal for pleasure according to frame length and compression ratio
Sound signal, frame length 4096, then the columns of calculation matrix to be generated is 4096., then can be with such as compression ratio is preset as 1/3
Line number is determined according to the columns of compression ratio and the calculation matrix determined, line number is that 4096* (1/3) is rounded, as 1365.It is determined that
After the line number and columns of calculation matrix, and then calculation matrix is generated according to the type of line number, columns and default calculation matrix.
In the embodiment of the present invention, calculation matrix uses structuring calculation matrix, such as the type of structuring calculation matrix can be part
Any of Fourier's matrix, some discrete cosine transform DTC matrixes or part Bernoulli Jacob's random matrix.
So that the type of calculation matrix is partial Fourier matrix as an example, generating the implementation of calculation matrix can be:It is first
First to unit matrix I4096×4096Do Fourier transformation, i.e. to unit matrix I4096×4096Each row do Fourier transformation and obtain
Fourier's matrix Φ4096×4096, then randomly select Fourier's matrix Φ4096×40961365 rows obtain partial Fourier matrix
Φ1365×4096。
For one group of frequency coefficient corresponding to a frame audio signal of any passage, by the frequency coefficient and calculation matrix phase
Multiply, obtain the value after frame audio signal compression sampling, and then the value after sampling is quantified, obtain being sent to remote terminal
Signal, and transmit to remote terminal.
407th, terminal receive computing device send reconstructed audio signals parameter, parameter include calculation matrix, sparse base,
The grouping information of passage, the language musical sound label information of audio signal and signal frame length corresponding to language musical sound label information, language pleasure
Sound label information is used to indicate that at least two audio signals are voice signal or note signal.
Terminal carries the parameter of audio signal, for making when receiving the signal of computing device transmission in the signal
Terminal carries out signal reconstruction according to the parameter.Calculation matrix is used for terminal and used in the algorithm that interchannel joint is rebuild;It is sparse
Basis representation computing device carries out time domain to the sparse transformation algorithm of frequency domain, such as can be MDCT algorithms, DWT algorithms etc., terminal
Algorithm when terminal carries out the inverse transformation of frequency domain to time domain can be determined according to the sparse base, can be correspondingly that inverse MDCT is calculated
Method, inverse DWT algorithms etc.;The grouping information of passage can include group mark corresponding to each passage;The language musical sound mark of audio signal
Label information can indicate the type of audio signal, can include voice signal and note signal, the parameter also includes audio signal
Type corresponding to signal frame length, such as the signal frame a length of 1024 of voice signal, the signal frame a length of 4096 of note signal.
408th, terminal obtains compressed data corresponding at least two audio signals of at least two passages.
Terminal frame length according to corresponding to language musical sound label information obtains language musical sound label information.If it is determined that be voice signal,
Then terminal determines that frame length is frame length corresponding to voice signal;If it is determined that being note signal, then terminal determines that frame length is note signal
Corresponding frame length.Such as when it is voice signal that terminal, which determines audio signal, its corresponding frame length is corresponding to voice signal
1024, when it is note signal that terminal, which determines audio signal, its corresponding frame length is 4096 corresponding to note signal.
Then, according to frame length, measurement data corresponding at least two audio signals is extracted.Such as multichannel audio signal is musical sound
During signal, it is determined that frame length be 4096, terminal each channel reception to signal in respectively take a frame length be 4096 measurement
Data, to get measurement data corresponding at least two audio signals corresponding at least two passages, at least two passages with
At least two audio signals correspond, and then carry out inverse quantization to measurement data corresponding to each passage, obtain at least two
Data after the compressed data of audio signal, the i.e. compression sampling of at least two audio signals.
409th, terminal obtains the grouping information of the group where passage corresponding at least two audio signals.
Terminal can be according to corresponding to the grouping information that is carried in parameter obtains at least two audio signals where passage group
Grouping information, grouping information may indicate that the corresponding group mark of each passage.Such as have 32 passages, it is divided into 8 groups, every group
4 passages, grouping information indicate that every 4 passages carry identical group mark.
410th, the compressed data of at least two audio signals is grouped, so as to be pressed by terminal according to grouping information
Contracting data group.
Terminal can organize mark according to corresponding to each passage, by least two audio signals corresponding at least two passages
Compressed data be grouped, so as to obtain compressed data group.Such as the compressed data of corresponding 32 audio signals of 32 passages,
One group of mark is corresponded to according to each passage, totally 4 groups marks, by the compression of the audio signal of the passage under 8 identical group of marks
Data are divided into one group, obtain 4 compressed data groups.
411st, terminal obtains calculation matrix, and according to the compressed data and calculation matrix in compressed data group, joint rebuilds pressure
Frequency coefficient corresponding to compressed data in contracting data group.
Terminal obtains calculation matrix from the parameter received, compresses number according to corresponding to a passage in compressed data group
According to compressed data and calculation matrix corresponding to another passage in corresponding frequency coefficient, compressed data group, compression number is calculated
According to frequency coefficient corresponding to compressed data corresponding to another passage in group.
Specifically, terminal can frequency coefficient corresponding to the compressed data according to corresponding to i-th of passage in compressed data group,
Compressed data and calculation matrix corresponding to i+1 passage in compressed data group, calculate compression number corresponding to i+1 passage
According to corresponding frequency coefficient, until frequency domain system corresponding to compressed data corresponding to k-th of passage is calculated in compressed data group
Number, i are the positive integer less than k, and k is the total number of channels in compressed data group.Its computational methods can use AMP algorithms, work as end
End is from the 1st path computation in each compressed data group to k-th of passage, to obtain compressed data corresponding to k-th of passage
Corresponding frequency coefficient, it is properly termed as a forward direction AMP algorithm iteration process.And then can be according to j-th in compressed data group
Frequency coefficient corresponding to compressed data corresponding to passage, compressed data and survey corresponding to -1 passage of jth in compressed data group
Moment matrix, frequency coefficient corresponding to compressed data corresponding to -1 passage of jth is calculated, until the is calculated in compressed data group
Frequency coefficient corresponding to compressed data corresponding to 1 passage, j are the positive integer less than or equal to k, and more than 1.So when
From k-th of path computation in each compressed data group to the 1st passage, to obtain the 1st compressed data pair corresponding to passage
The frequency coefficient answered, it is properly termed as once backward AMP algorithm iterations process.
Specifically, when realizing frequency coefficient corresponding to compressed data, pressure according to corresponding to a passage in compressed data group
Compressed data and calculation matrix corresponding to another passage in contracting data group, it is corresponding to calculate another passage in compressed data group
Compressed data corresponding to frequency coefficient when, according to AMP algorithms, its specific algorithm can be:According to corresponding to a passage
Frequency coefficient corresponding to compressed data, determine priori frequency coefficient corresponding to compressed data corresponding to another passage;By priori
Priori of the frequency coefficient as frequency coefficient corresponding to compressed data corresponding to another passage, and it is corresponding according to another passage
Compressed data and calculation matrix, calculate frequency domain system corresponding to compressed data corresponding to another passage in compressed data group
Number.It is to be understood that when getting frequency coefficient corresponding to compressed data corresponding to any passage, while the passage can be obtained
The posterior marginal probability of frequency coefficient corresponding to corresponding compressed data, by frequency domain corresponding to compressed data corresponding to a upper passage
Coefficient leads to as priori frequency coefficient corresponding to compressed data corresponding to next passage, i.e. the priori frequency coefficient to be next
The priori of frequency coefficient corresponding to compressed data corresponding to road, if compressed data and measurement square according to corresponding to next passage
Gust, in calculating compressed data group corresponding to compressed data corresponding to next passage during frequency coefficient, while also obtain next
The posterior marginal probability of frequency coefficient corresponding to compressed data corresponding to passage, if by it is preceding to AMP algorithm iterations process and after
To during AMP algorithm iterations, calculate to the posterior marginal probability of frequency coefficient corresponding to compressed data corresponding to any passage
Reach preset value, it is determined that frequency coefficient is the most accurate corresponding to compressed data corresponding to current channel, by corresponding to the passage
Frequency coefficient corresponding to compressed data is as frequency coefficient corresponding to the compressed data corresponding to the passage in compressed data group.
It should be noted that the present invention uses the calculation matrix of structuring in audio signal compression sampling process, such as
Partial Fourier matrix, part DCT matrixes and part Bernoulli Jacob's random matrix etc..It is for example random for non-structured calculation matrix
Gaussian matrix etc., the time complexity of matrix multiplication is ab (a, b are respectively the row and column of matrix), and uses the measurement of structuring
Matrix, algorithm complex can be significantly decreased on the premise of reconstruction precision is ensured, the time of matrix multiplication can be caused to answer
Miscellaneous degree is reduced to nlog (n).Further, since unstructured multiplication of matrices operation time and signal frame length are not linear relationship, and
Quadratic relationship, can cause signal frame length selection be restricted, frame length is longer, and time complexity is bigger, for note signal this
The relatively gentle signal of kind time-varying characteristics, it is impossible to by increasing signal frame length come the reconstruction precision of promotion signal.And the application adopts
With the calculation matrix of structuring, it can make it that the selection of signal frame length is more flexible, so that can while operation time is reduced
To lift reconstruction precision.
412nd, terminal-pair frequency coefficient carries out the conversion of frequency domain to time domain, so as to obtain the compressed data in compressed data group
Corresponding audio signal.
After frequency coefficient corresponding to the compressed data in each compressed data group is obtained, determined according to sparse base from frequency domain
To the algorithm of the inverse transformation of time domain, such as sparse base is MDCT algorithms, then inverse transformation just uses inverse MDCT algorithms, i.e., using inverse
MDCT algorithms carry out inverse transformation to the frequency coefficient of the compressed data of each compressed data group quilt, and the signal obtained after inverse transformation is just
For time-domain signal, i.e. audio signal corresponding to the compressed data after reconstruction in compressed data group.
Therefore, for the reconstruction terminal for audio signal, believe at least two audios for receiving at least two passages
Number compressed data after, can according to corresponding at least two audio signals group where passage grouping information by least two audios
The compressed data of signal is grouped, and so can be carried out joint reconstruction to the compressed data in obtained compressed data group, can be carried
The precision of the inline occlusal reconstruction of group is risen, joint reconstruction is carried out between each group can lift the speed that joint is rebuild..
It is above-mentioned that mainly scheme provided in an embodiment of the present invention is described from the angle of interaction between each network element.Can
With understanding, each network element, such as computing device, terminal etc., in order to realize above-mentioned function, it comprises perform each function
Corresponding hardware configuration and/or software module.Those skilled in the art should be readily appreciated that, with reference to disclosed herein
The unit and algorithm steps of each example of embodiment description, the present invention can be with the combination shape of hardware or hardware and computer software
Formula is realized.Some functions is performed in a manner of hardware or computer software driving hardware actually, depending on technical scheme
Application-specific and design constraint.Professional and technical personnel can be realized to each specific application using distinct methods
Described function, but this realization is it is not considered that beyond the scope of this invention.
The embodiment of the present invention can carry out the division of functional module, example according to above method example to computing device, terminal
Such as, each function can be corresponded to and divide each functional module, two or more functions can also be integrated at one
Manage in module.Above-mentioned integrated module can both be realized in the form of hardware, can also use the form of software function module
Realize.It should be noted that the division in the embodiment of the present invention to module is schematical, only a kind of logic function is drawn
Point, there can be other dividing mode when actually realizing.
In the case where dividing each functional module using corresponding each function, Fig. 5 shows involved in above-described embodiment
And terminal a kind of possible structural representation, terminal includes:Acquiring unit 501, grouped element 502, reconstruction unit 503,
Converter unit 504.Acquiring unit 501 is used to support terminal to perform the process 408 in the process 205,206,208, Fig. 4 in Fig. 3,
409,411, grouped element 502 is used to support that terminal performs the process 207 in Fig. 3, the process 410 in Fig. 4, reconstruction unit 503
For supporting terminal to perform the process 208 in Fig. 3, the process 411 in Fig. 4, converter unit 504 is used to support terminal to perform Fig. 3
In process 209, the process 412 in Fig. 4.Wherein, all related contents for each step that above method embodiment is related to
To quote the description of the function of corresponding function module, will not be repeated here.
In the case of using integrated unit, Fig. 6 shows that one kind of terminal involved in above-described embodiment may
Structural representation.Terminal includes:Processing module 602 and communication module 603.The action that processing module 602 is used for terminal is entered
Row control management, for example, processing module 602 is used to support terminal to perform the process 205,206,207,208,209, Fig. 4 in Fig. 3
In process 408,409,410,411,412, communication module 603 is used to support that terminal performs the process 407 in Fig. 4, and/or uses
In other processes of techniques described herein.Communication module 603 is used to support terminal and the communication of other network entities, such as
With the communication of the computing device shown in Fig. 1, Fig. 2 or 4.Terminal can also include memory module 601, for storing terminal
Program code and data.
Wherein, processing module 602 can be processor or controller, such as can be central processing unit (Central
Processing Unit, CPU), general processor, digital signal processor (Digital Signal Processor, DSP),
Application specific integrated circuit (Application-Specific Integrated Circuit, ASIC), field programmable gate array
It is (Field Programmable Gate Array, FPGA) or other PLDs, transistor logic, hard
Part part or its any combination.What it can realize or perform with reference to described by the disclosure of invention various exemplary patrols
Collect square frame, module and circuit.Processor can also be the combination for realizing computing function, such as include one or more microprocessors
Combination, combination of DSP and microprocessor etc..Communication module 603 can be transceiver, transmission circuit or communication interface etc..Deposit
It can be memory to store up module 601.
When processing module 602 is processor, communication module 603 is transceiver, when memory module 601 is memory, this hair
Terminal involved by bright embodiment can be the terminal shown in Fig. 7.
As shown in fig.7, the terminal includes:Processor 712, transceiver 713, memory 711 and bus 714.Wherein,
Transceiver 713, processor 712 and memory 711 are connected with each other by bus 714;Bus 714 can be Peripheral Component Interconnect
Standard (Peripheral Component Interconnect, PCI) bus or EISA (Extended
Industry Standard Architecture, EISA) bus etc..Bus can be divided into address bus, data/address bus, control
Bus etc..For ease of representing, only represented in Fig. 7 with a thick line, it is not intended that an only bus or a type of total
Line.
In the case where dividing each functional module using corresponding each function, Fig. 8 shows involved in above-described embodiment
And computing device a kind of possible structural representation, computing device includes:Acquiring unit 801, grouped element 802, conversion
Unit 803, sampling unit 804, determining unit 805.Acquiring unit 801 is used to support the process in computing device Fig. 2
Process 401 in 201,204, Fig. 4, grouped element 802 are used to supporting process 202 in computing device Fig. 2, in Fig. 4
Process 404, converter unit 803 are used to support the process 405 in 203, Fig. 4 in computing device Fig. 2, sampling unit 804
For supporting the process 406 in 204, Fig. 4 in computing device Fig. 2, determining unit 805 is used to support computing device
402,403 in Fig. 4.Wherein, all related contents for each step that above method embodiment is related to can be quoted corresponding
The function description of functional module, will not be repeated here.
In the case of using integrated unit, Fig. 9 shows one kind of computing device involved in above-described embodiment
Possible structural representation.Computing device includes:Processing module 902 and communication module 903.Processing module 902 is used for calculating
The action of equipment is controlled management, for example, processing module 902 be used to supporting process 202 in computing device Fig. 2,
203rd, the process 401 in 204, Fig. 4,402,403,404,405,406, communication module 903 is used to support computing device Fig. 2
In process 201, and/or other processes for techniques described herein.Communication module 903 be used for support computing device with
The communication of other network entities, such as the communication of the terminal with being shown in Fig. 1, Fig. 3 or 4.Computing device can also include storage
Module 901, for storing the program code and data of computing device.
Wherein, processing module 902 can be processor or controller, such as can be central processor CPU, general procedure
Device, digital signal processor DSP, application-specific integrated circuit ASIC, on-site programmable gate array FPGA or other FPGAs
Device, transistor logic, hardware component or its any combination.It can realize or perform with reference to the disclosure of invention
Described various exemplary logic blocks, module and circuit.Processor can also be the combination for realizing computing function, such as
Combined comprising one or more microprocessors, combination of DSP and microprocessor etc..Communication module 903 can be transceiver, receive
Power Generation Road or communication interface etc..Memory module 901 can be memory.
When processing module 902 is processor, communication module 903 is transceiver, when memory module 901 is memory, this hair
Computing device involved by bright embodiment can be the computing device shown in Figure 10.
As shown in fig.10, the computing device includes:Array microphone 101, sound source processing module 102 and voice data are defeated
Go out module 103, array microphone, sound source processing module and voice data output module are connected with each other by bus 104;Bus
104 can be Peripheral Component Interconnect standard PCI bus or EISA eisa bus etc..Bus can be divided into address
Bus, data/address bus, controlling bus etc..For ease of representing, only represented in Figure 10 with a thick line, it is not intended that only one
Bus or a type of bus.
The step of method or algorithm with reference to described by the disclosure of invention, can be realized in a manner of hardware, also may be used
By be by computing device software instruction in a manner of realize.Software instruction can be made up of corresponding software module, software mould
Block can be stored on random access memory (Random Access Memory, RAM), flash memory, read-only storage (Read
Only Memory, ROM), Erasable Programmable Read Only Memory EPROM (Erasable Programmable ROM, EPROM), electricity can
EPROM (Electrically EPROM, EEPROM), register, hard disk, mobile hard disk, read-only optical disc
(CD-ROM) or in the storage medium of any other form well known in the art.A kind of exemplary storage medium is coupled to place
Device is managed, so as to enable a processor to from the read information, and information can be write to the storage medium.Certainly, store
Medium can also be the part of processor.Processor and storage medium can be located in ASIC.In addition, the ASIC can position
In core network interface equipment.Certainly, processor and storage medium can also be present in core network interface as discrete assembly and set
In standby.
Those skilled in the art are it will be appreciated that in said one or multiple examples, work(described in the invention
It is able to can be realized with hardware, software, firmware or their any combination.When implemented in software, can be by these functions
It is stored in computer-readable medium or is transmitted as one or more instructions on computer-readable medium or code.
Computer-readable medium includes computer-readable storage medium and communication media, and wherein communication media includes being easy to from a place to another
Any medium of one place transmission computer program.It is any that storage medium can be that universal or special computer can access
Usable medium.
Above-described embodiment, the purpose of the present invention, technical scheme and beneficial effect are carried out further
Describe in detail, should be understood that the embodiment that the foregoing is only the present invention, be not intended to limit the present invention
Protection domain, all any modification, equivalent substitution and improvements on the basis of technical scheme, done etc., all should
It is included within protection scope of the present invention.
Claims (24)
- A kind of 1. method for reconstructing of audio signal, it is characterised in that including:Obtain compressed data corresponding at least two audio signals of at least two passages, at least two passage with it is described extremely Few two audio signals correspond;Obtain the grouping information of the group corresponding at least two audio signal where passage;According to the grouping information, compressed data corresponding at least two audio signal is grouped, so as to be pressed Contracting data group;Calculation matrix is obtained, according to the compressed data in the compressed data group and the calculation matrix, joint rebuilds the pressure Frequency coefficient corresponding to compressed data in contracting data group;The conversion of frequency domain to time domain is carried out to the frequency coefficient, it is corresponding so as to obtain the compressed data in the compressed data group Audio signal.
- 2. according to the method for claim 1, it is characterised in that methods described also includes:Obtain the language musical sound label information of at least two audio signal, institute's predicate musical sound label information be used to indicating it is described extremely Few two audio signals are voice signal or note signal;Compressed data corresponding at least two audio signals for obtaining at least two passages includes:According to institute's predicate musical sound mark Information is signed, obtains frame length corresponding to institute's predicate musical sound label information;According to the frame length, measurement data corresponding at least two audio signal is extracted;Inverse quantization is carried out to the measurement data, so as to obtain compressed data corresponding at least two audio signal.
- 3. method according to claim 1 or 2, it is characterised in that the compression number according in the compressed data group Include according to frequency coefficient corresponding to the compressed data rebuild with the calculation matrix, joint in the compressed data group:The frequency coefficient corresponding to compressed data, another in the compressed data group according to corresponding to a passage in compressed data group Compressed data corresponding to individual passage and the calculation matrix, calculate in the compressed data group corresponding to another described passage Frequency coefficient corresponding to compressed data.
- 4. according to the method for claim 3, it is characterised in that compress number according to corresponding to a passage in compressed data group According to compressed data corresponding to another passage and the calculation matrix in corresponding frequency coefficient, the compressed data group, meter Calculating frequency coefficient corresponding to compressed data corresponding to another described passage in the compressed data group includes:According in frequency coefficient corresponding to compressed data corresponding to i-th of passage in compressed data group, the compressed data group i-th Compressed data corresponding to+1 passage and the calculation matrix, it is corresponding to calculate compressed data corresponding to the i+1 passage Frequency coefficient, until frequency coefficient corresponding to compressed data corresponding to k-th of passage in the compressed data group is calculated, I is the positive integer less than k, and k is the total number of channels in the compressed data group.
- 5. according to the method for claim 4, it is characterised in that methods described also includes:According in frequency coefficient corresponding to compressed data corresponding to j-th of passage in compressed data group, the compressed data group Compressed data corresponding to j-1 passage and the calculation matrix, it is corresponding to calculate compressed data corresponding to described -1 passage of jth Frequency coefficient, until the 1st frequency coefficient corresponding to compressed data corresponding to passage in the compressed data group is calculated, J is the positive integer less than or equal to k, and more than 1.
- 6. the method according to claim 4 or 5, it is characterised in that methods described also includes:According to compressed data and the calculation matrix corresponding to default initialization frequency coefficient, the 1st passage, calculate 2nd frequency coefficient corresponding to compressed data corresponding to passage.
- 7. according to the method described in any one of claim 3 to 6, it is characterised in that according to a passage pair in compressed data group Compressed data and the survey corresponding to another passage in frequency coefficient, the compressed data group corresponding to the compressed data answered Moment matrix, calculating frequency coefficient corresponding to compressed data corresponding to another described passage in the compressed data group includes:According to frequency coefficient corresponding to compressed data corresponding to one passage, it is determined that being compressed corresponding to another described passage Priori frequency coefficient corresponding to data;Priori using the priori frequency coefficient as frequency coefficient corresponding to compressed data corresponding to another described passage, and According to compressed data and the calculation matrix corresponding to another described passage, calculate described another in the compressed data group Frequency coefficient corresponding to compressed data corresponding to individual passage.
- A kind of 8. compressive sampling method of audio signal, it is characterised in that including:Obtain at least two audio signals of at least two passages, at least two passage and at least two audio signal Correspond;The correlation between at least two audio signal is calculated, according to the correlation at least two audio signal It is grouped, so as to obtain the grouping information of the group where the passage at least two passage;Conversion of the time domain to frequency domain is carried out at least two audio signal, it is described so as to obtain at least two groups of frequency coefficients At least two groups of frequency coefficients correspond with least two audio signal;Calculation matrix is obtained, at least two groups of frequency coefficients are sampled according to the calculation matrix, it is described so as to obtain Compressed data corresponding at least two audio signals.
- 9. according to the method for claim 8, it is characterised in that before calculation matrix is obtained, methods described also includes:Determine the language musical sound label information of at least two audio signal, institute's predicate musical sound label information be used to indicating it is described extremely Few two audio signals are voice signal or note signal;According to institute's predicate musical sound label information, the frame length of at least two audio signal is determined.
- 10. according to the method for claim 9, it is characterised in that the acquisition calculation matrix includes:According to the frame length, the calculation matrix corresponding to the frame length is obtained.
- 11. according to the method described in any one of claim 8 to 10, it is characterised in that described to calculate at least two audio Correlation between signal, carrying out packet at least two audio signal according to the correlation includes:The first audio signal at least two audio signal is obtained, obtains remaining audio in addition to first audio signal In signal with m audio signal before the first audio signal correlation highest, and by first audio signal and with institute For m audio signal as one group of audio signal, m is just whole more than or equal to 1 before stating the first audio signal correlation highest Number;From except first audio signal and with its lingering sound in addition to m audio signal before the first audio signal correlation highest Continue to choose the second audio signal in frequency signal and obtain except first audio signal, the second audio signal and with described first Before audio signal correlation highest in the outer rest channels of m audio signal with before the second audio signal correlation highest M passage, and using described and audio signal and with described and before audio signal correlation highest m audio signal as Another group of audio signal, until the packet of at least two audio is completed.
- 12. according to the method described in any one of claim 8 to 11, it is characterised in that between at least two audio signal Correlation include the distance between described at least two audio signal.
- A kind of 13. reconstructing device of audio signal, it is characterised in that including:Acquiring unit, for obtain at least two passages at least two audio signals corresponding to compressed data, described at least two Individual passage corresponds with least two audio signal;The acquiring unit, it is additionally operable to obtain the grouping information of the group corresponding at least two audio signal where passage;Grouped element, for according to the grouping information, compressed data corresponding at least two audio signal to be divided Group, so as to obtain compressed data group;Reconstruction unit, for obtaining calculation matrix, according to the compressed data in the compressed data group and the calculation matrix, connection Frequency coefficient corresponding to compressed data described in occlusal reconstruction in compressed data group;Converter unit, for carrying out the conversion of frequency domain to time domain to the frequency coefficient, so as to obtain in the compressed data group Compressed data corresponding to audio signal.
- 14. device according to claim 13, it is characterised in that the acquiring unit is additionally operable to:Obtain the language musical sound label information of at least two audio signal, institute's predicate musical sound label information be used to indicating it is described extremely Few two audio signals are voice signal or note signal;The acquiring unit, is used for:According to institute's predicate musical sound label information, frame length corresponding to institute's predicate musical sound label information is obtained;According to the frame length, measurement data corresponding at least two audio signal is extracted;Inverse quantization is carried out to the measurement data, so as to obtain the compressed data of at least two audio signal.
- 15. the device according to claim 13 or 14, it is characterised in that the reconstruction unit is used for:The frequency coefficient corresponding to compressed data, another in the compressed data group according to corresponding to a passage in compressed data group Compressed data corresponding to individual passage and the calculation matrix, calculate in the compressed data group corresponding to another described passage Frequency coefficient corresponding to compressed data.
- 16. device according to claim 15, it is characterised in that the reconstruction unit is used for:According in frequency coefficient corresponding to compressed data corresponding to i-th of passage in compressed data group, the compressed data group i-th Compressed data corresponding to+1 passage and the calculation matrix, it is corresponding to calculate compressed data corresponding to the i+1 passage Frequency coefficient, until frequency coefficient corresponding to compressed data corresponding to k-th of passage in the compressed data group is calculated, I is the positive integer less than k, and k is the total number of channels in the compressed data group.
- 17. device according to claim 16, the reconstruction unit is additionally operable to:According in frequency coefficient corresponding to compressed data corresponding to j-th of passage in compressed data group, the compressed data group Compressed data corresponding to j-1 passage and the calculation matrix, it is corresponding to calculate compressed data corresponding to described -1 passage of jth Frequency coefficient, until the 1st frequency coefficient corresponding to compressed data corresponding to passage in the compressed data group is calculated, J is the positive integer less than or equal to k, and more than 1.
- 18. the device according to claim 16 or 17, it is characterised in that the reconstruction unit is additionally operable to:According to compressed data and the calculation matrix corresponding to default initialization frequency coefficient, the 1st passage, calculate 2nd frequency coefficient corresponding to compressed data corresponding to passage.
- 19. according to the device described in claim any one of 15-18, it is characterised in that the reconstruction unit is used for:According to frequency coefficient corresponding to compressed data corresponding to one passage, it is determined that being compressed corresponding to another described passage Priori frequency coefficient corresponding to data;Priori using the priori frequency coefficient as frequency coefficient corresponding to compressed data corresponding to another described passage, and According to compressed data and the calculation matrix corresponding to another described passage, calculate described another in the compressed data group Frequency coefficient corresponding to compressed data corresponding to individual passage.
- A kind of 20. compression sampling device of audio signal, it is characterised in that including:Acquiring unit, for obtain at least two passages at least two audio signals, at least two passage with it is described extremely Few two audio signals correspond;Grouped element, for calculating the correlation between at least two audio signal, according to the correlation to it is described extremely Few two audio signals are grouped, so as to obtain the grouping information of the group where the passage at least two passage;Converter unit, for carrying out conversion of the time domain to frequency domain at least two audio signal, so as to obtain at least two groups Frequency coefficient, at least two groups of frequency coefficients correspond with least two audio signal;The acquiring unit, it is additionally operable to obtain calculation matrix;Sampling unit, it is described so as to obtain for being sampled according to the calculation matrix at least two groups of frequency coefficients Compressed data corresponding at least two audio signals.
- 21. device according to claim 20, it is characterised in that also including determining unit, be used for:Determine the language musical sound label information of at least two audio signal, institute's predicate musical sound label information be used to indicating it is described extremely Few two audio signals are voice signal or note signal;According to institute's predicate musical sound label information, the frame length of at least two audio signal is determined.
- 22. device according to claim 21, it is characterised in that the acquiring unit is used for:According to the frame length, the calculation matrix corresponding to the frame length is obtained.
- 23. according to the device described in any one of claim 20 to 22, it is characterised in that the grouped element is used for:The first audio signal at least two audio signal is obtained, obtains remaining audio in addition to first audio signal In signal with m audio signal before the first audio signal correlation highest, and by first audio signal and with institute For m audio signal as one group of audio signal, m is just whole more than or equal to 1 before stating the first audio signal correlation highest Number;From except first audio signal and with its lingering sound in addition to m audio signal before the first audio signal correlation highest Continue to choose the second audio signal in frequency signal and obtain except first audio signal, the second audio signal and with described first Before audio signal correlation highest in the outer rest channels of m audio signal with before the second audio signal correlation highest M passage, and using described and audio signal and with described and before audio signal correlation highest m audio signal as Another group of audio signal, until the packet of at least two audio is completed.
- 24. according to the device described in claim any one of 20-23, it is characterised in that between at least two audio signal Correlation include the distance between described at least two audio signal.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610879165.XA CN107895580B (en) | 2016-09-30 | 2016-09-30 | Audio signal reconstruction method and device |
PCT/CN2017/086390 WO2018058989A1 (en) | 2016-09-30 | 2017-05-27 | Audio signal reconstruction method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610879165.XA CN107895580B (en) | 2016-09-30 | 2016-09-30 | Audio signal reconstruction method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107895580A true CN107895580A (en) | 2018-04-10 |
CN107895580B CN107895580B (en) | 2021-06-01 |
Family
ID=61763093
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610879165.XA Active CN107895580B (en) | 2016-09-30 | 2016-09-30 | Audio signal reconstruction method and device |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN107895580B (en) |
WO (1) | WO2018058989A1 (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110874626A (en) * | 2018-09-03 | 2020-03-10 | 华为技术有限公司 | Quantization method and device |
CN111128230A (en) * | 2019-12-31 | 2020-05-08 | 广州市百果园信息技术有限公司 | Voice signal reconstruction method, device, equipment and storage medium |
CN111345047A (en) * | 2019-04-17 | 2020-06-26 | 深圳市大疆创新科技有限公司 | Audio signal processing method, apparatus and storage medium |
CN111699700A (en) * | 2019-04-17 | 2020-09-22 | 深圳市大疆创新科技有限公司 | Audio signal processing method, apparatus and storage medium |
Citations (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6345246B1 (en) * | 1997-02-05 | 2002-02-05 | Nippon Telegraph And Telephone Corporation | Apparatus and method for efficiently coding plural channels of an acoustic signal at low bit rates |
US20030115041A1 (en) * | 2001-12-14 | 2003-06-19 | Microsoft Corporation | Quality improvement techniques in an audio encoder |
US20030169886A1 (en) * | 1995-01-10 | 2003-09-11 | Boyce Roger W. | Method and apparatus for encoding mixed surround sound into a single stereo pair |
WO2007011083A1 (en) * | 2005-07-18 | 2007-01-25 | Lg Electronics Inc. | Apparatus and method of encoding and decoding audio signal |
US20070172071A1 (en) * | 2006-01-20 | 2007-07-26 | Microsoft Corporation | Complex transforms for multi-channel audio |
CN101055721A (en) * | 2004-09-17 | 2007-10-17 | 广州广晟数码技术有限公司 | Multi-sound channel digital audio encoding device and its method |
CN101281749A (en) * | 2008-05-22 | 2008-10-08 | 上海交通大学 | Apparatus for encoding and decoding hierarchical voice and musical sound together |
CN101447190A (en) * | 2008-06-25 | 2009-06-03 | 北京大学深圳研究生院 | Voice enhancement method employing combination of nesting-subarray-based post filtering and spectrum-subtraction |
CN102047564A (en) * | 2008-05-30 | 2011-05-04 | 微软公司 | Factorization of overlapping transforms into two block transforms |
CN102089803A (en) * | 2008-07-11 | 2011-06-08 | 弗劳恩霍夫应用研究促进协会 | Method and discriminator for classifying different segments of a signal |
US20120087504A1 (en) * | 2002-09-04 | 2012-04-12 | Microsoft Corporation | Multi-channel audio encoding and decoding |
JP4983845B2 (en) * | 2009-04-17 | 2012-07-25 | 株式会社Jvcケンウッド | Audio signal transmission device, audio signal reception device, and audio signal transmission system |
CN102982805A (en) * | 2012-12-27 | 2013-03-20 | 北京理工大学 | Multi-channel audio signal compressing method based on tensor decomposition |
CN103745724A (en) * | 2014-01-13 | 2014-04-23 | 电子科技大学 | Time-frequency mixed down-mixing method applied to multi-channel audio decoding |
CN104240712A (en) * | 2014-09-30 | 2014-12-24 | 武汉大学深圳研究院 | Three-dimensional audio multichannel grouping and clustering coding method and three-dimensional audio multichannel grouping and clustering coding system |
CN104934032A (en) * | 2014-03-17 | 2015-09-23 | 华为技术有限公司 | Method and device for voice signal processing according to frequency domain energy |
US20150380000A1 (en) * | 2013-02-14 | 2015-12-31 | Dolby Laboratories Licensing Corporation | Signal Decorrelation in an Audio Processing System |
CN105518775A (en) * | 2013-07-22 | 2016-04-20 | 弗朗霍夫应用科学研究促进协会 | In reduction of comb filter artifacts in multi-channel downmix with adaptive phase alignment |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN100538821C (en) * | 2006-03-23 | 2009-09-09 | 北京东方利优科技发展有限公司 | The decoding method of fast audio-variable signal |
JP5483813B2 (en) * | 2007-12-21 | 2014-05-07 | 株式会社Nttドコモ | Multi-channel speech / acoustic signal encoding apparatus and method, and multi-channel speech / acoustic signal decoding apparatus and method |
US20090210222A1 (en) * | 2008-02-15 | 2009-08-20 | Microsoft Corporation | Multi-Channel Hole-Filling For Audio Compression |
CN102081926B (en) * | 2009-11-27 | 2013-06-05 | 中兴通讯股份有限公司 | Method and system for encoding and decoding lattice vector quantization audio |
CN103714825A (en) * | 2014-01-16 | 2014-04-09 | 中国科学院声学研究所 | Multi-channel speech enhancing method based on auditory perception model |
-
2016
- 2016-09-30 CN CN201610879165.XA patent/CN107895580B/en active Active
-
2017
- 2017-05-27 WO PCT/CN2017/086390 patent/WO2018058989A1/en active Application Filing
Patent Citations (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20030169886A1 (en) * | 1995-01-10 | 2003-09-11 | Boyce Roger W. | Method and apparatus for encoding mixed surround sound into a single stereo pair |
US6345246B1 (en) * | 1997-02-05 | 2002-02-05 | Nippon Telegraph And Telephone Corporation | Apparatus and method for efficiently coding plural channels of an acoustic signal at low bit rates |
US20030115041A1 (en) * | 2001-12-14 | 2003-06-19 | Microsoft Corporation | Quality improvement techniques in an audio encoder |
US20120087504A1 (en) * | 2002-09-04 | 2012-04-12 | Microsoft Corporation | Multi-channel audio encoding and decoding |
CN101055721A (en) * | 2004-09-17 | 2007-10-17 | 广州广晟数码技术有限公司 | Multi-sound channel digital audio encoding device and its method |
WO2007011083A1 (en) * | 2005-07-18 | 2007-01-25 | Lg Electronics Inc. | Apparatus and method of encoding and decoding audio signal |
US8190425B2 (en) * | 2006-01-20 | 2012-05-29 | Microsoft Corporation | Complex cross-correlation parameters for multi-channel audio |
US20070172071A1 (en) * | 2006-01-20 | 2007-07-26 | Microsoft Corporation | Complex transforms for multi-channel audio |
CN101281749A (en) * | 2008-05-22 | 2008-10-08 | 上海交通大学 | Apparatus for encoding and decoding hierarchical voice and musical sound together |
CN102047564A (en) * | 2008-05-30 | 2011-05-04 | 微软公司 | Factorization of overlapping transforms into two block transforms |
CN101447190A (en) * | 2008-06-25 | 2009-06-03 | 北京大学深圳研究生院 | Voice enhancement method employing combination of nesting-subarray-based post filtering and spectrum-subtraction |
CN102089803A (en) * | 2008-07-11 | 2011-06-08 | 弗劳恩霍夫应用研究促进协会 | Method and discriminator for classifying different segments of a signal |
JP4983845B2 (en) * | 2009-04-17 | 2012-07-25 | 株式会社Jvcケンウッド | Audio signal transmission device, audio signal reception device, and audio signal transmission system |
CN102982805A (en) * | 2012-12-27 | 2013-03-20 | 北京理工大学 | Multi-channel audio signal compressing method based on tensor decomposition |
US20150380000A1 (en) * | 2013-02-14 | 2015-12-31 | Dolby Laboratories Licensing Corporation | Signal Decorrelation in an Audio Processing System |
CN105518775A (en) * | 2013-07-22 | 2016-04-20 | 弗朗霍夫应用科学研究促进协会 | In reduction of comb filter artifacts in multi-channel downmix with adaptive phase alignment |
CN103745724A (en) * | 2014-01-13 | 2014-04-23 | 电子科技大学 | Time-frequency mixed down-mixing method applied to multi-channel audio decoding |
CN104934032A (en) * | 2014-03-17 | 2015-09-23 | 华为技术有限公司 | Method and device for voice signal processing according to frequency domain energy |
CN104934032B (en) * | 2014-03-17 | 2019-04-05 | 华为技术有限公司 | The method and apparatus that voice signal is handled according to frequency domain energy |
CN104240712A (en) * | 2014-09-30 | 2014-12-24 | 武汉大学深圳研究院 | Three-dimensional audio multichannel grouping and clustering coding method and three-dimensional audio multichannel grouping and clustering coding system |
Non-Patent Citations (3)
Title |
---|
BAOZHEN WU等: "Downmix and coding of multichannel signals based on spatial correlation", 《2015 8TH INTERNATIONAL CONGRESS ON IMAGE AND SIGNAL PROCESSING (CISP)》 * |
王超: "多通道立体声虚拟重建技术研究", 《中国优秀硕士学位论文全文数据库》 * |
郭柯: "数字声音压缩编码的进展", 《通信学报》 * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110874626A (en) * | 2018-09-03 | 2020-03-10 | 华为技术有限公司 | Quantization method and device |
CN110874626B (en) * | 2018-09-03 | 2023-07-18 | 华为技术有限公司 | Quantization method and quantization device |
CN111345047A (en) * | 2019-04-17 | 2020-06-26 | 深圳市大疆创新科技有限公司 | Audio signal processing method, apparatus and storage medium |
CN111699700A (en) * | 2019-04-17 | 2020-09-22 | 深圳市大疆创新科技有限公司 | Audio signal processing method, apparatus and storage medium |
WO2020211017A1 (en) * | 2019-04-17 | 2020-10-22 | 深圳市大疆创新科技有限公司 | Audio signal processing method and device, and storage medium |
CN111128230A (en) * | 2019-12-31 | 2020-05-08 | 广州市百果园信息技术有限公司 | Voice signal reconstruction method, device, equipment and storage medium |
CN111128230B (en) * | 2019-12-31 | 2022-03-04 | 广州市百果园信息技术有限公司 | Voice signal reconstruction method, device, equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN107895580B (en) | 2021-06-01 |
WO2018058989A1 (en) | 2018-04-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107895580A (en) | The method for reconstructing and device of a kind of audio signal | |
CN104347067B (en) | Audio signal classification method and device | |
CN102667918B (en) | For making reverberator and the method for sound signal reverberation | |
CN107071686A (en) | The method and apparatus for audio playback is represented for rendering audio sound field | |
CN104854655A (en) | Method and apparatus for compressing and decompressing higher order ambisonics representation for sound field | |
TW200923913A (en) | Encoding and/or decoding digital content | |
US11521592B2 (en) | Small-footprint flow-based models for raw audio | |
CN111292768A (en) | Method and device for hiding lost packet, storage medium and computer equipment | |
CN111696580B (en) | Voice detection method and device, electronic equipment and storage medium | |
CN102831895A (en) | Method for achieving MFCC (Mel Frequency Cepstrum Coefficient) parameter extraction by field-programmable gate array | |
CN110379411A (en) | For the phoneme synthesizing method and device of target speaker | |
CN111261177A (en) | Voice conversion method, electronic device and computer readable storage medium | |
CN101646123B (en) | Filter bank simulating auditory perception model | |
CN110249384A (en) | The quantizer arranged with index coding and position | |
CN104064191B (en) | Sound mixing method and device | |
CN111696520A (en) | Intelligent dubbing method, device, medium and electronic equipment | |
CN110010153A (en) | A kind of mute detection method neural network based, terminal device and medium | |
CN106233112A (en) | Coding method and equipment and signal decoding method and equipment | |
CN114783459B (en) | Voice separation method and device, electronic equipment and storage medium | |
JP2023548707A (en) | Speech enhancement methods, devices, equipment and computer programs | |
CN109087654A (en) | To the method and apparatus of high-order clear stereo signal application dynamic range compression | |
Raj et al. | Multilayered convolutional neural network-based auto-CODEC for audio signal denoising using mel-frequency cepstral coefficients | |
CN1770256B (en) | Digital audio frequency mixing method based on transform domain | |
CN111653261A (en) | Speech synthesis method, speech synthesis device, readable storage medium and electronic equipment | |
CN104021793B (en) | Method and apparatus for processing audio signal |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |