CN106842112B - Sound localization method based on parametrization Bayes's dictionary learning under strong reverberant ambiance - Google Patents
Sound localization method based on parametrization Bayes's dictionary learning under strong reverberant ambiance Download PDFInfo
- Publication number
- CN106842112B CN106842112B CN201611251992.0A CN201611251992A CN106842112B CN 106842112 B CN106842112 B CN 106842112B CN 201611251992 A CN201611251992 A CN 201611251992A CN 106842112 B CN106842112 B CN 106842112B
- Authority
- CN
- China
- Prior art keywords
- formula
- dictionary
- room
- mirror image
- sound source
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S1/00—Beacons or beacon systems transmitting signals having a characteristic or characteristics capable of being detected by non-directional receivers and defining directions, positions, or position lines fixed relatively to the beacon transmitters; Receivers co-operating therewith
- G01S1/72—Beacons or beacon systems transmitting signals having a characteristic or characteristics capable of being detected by non-directional receivers and defining directions, positions, or position lines fixed relatively to the beacon transmitters; Receivers co-operating therewith using ultrasonic, sonic or infrasonic waves
- G01S1/76—Systems for determining direction or position line
Abstract
The invention discloses the sound localization methods based on parametrization Bayes's dictionary learning under a kind of strong reverberant ambiance, the technical issues of for solving existing sound localization method poor accuracy.Technical solution is to simplify dictionary based on microphone mirror image, under the frame of sparse Bayesian, realizes that auditory localization estimation synchronous with room reflections coefficient solves.Dictionary is precalculated out using the method for the present invention, the available promotion of efficiency for completing auditory localization and reflection coefficient estimation improves the precision of sound localization method in this way in the case where not being obviously improved calculation amount.The present invention has very big potential practical value to the auditory localization under complex indoor environment, also can be used as the method that parameters,acoustic is estimated in enclosure space.
Description
Technical field
The present invention relates to a kind of sound localization method, in particular to based on parametrization Bayes's word under a kind of strong reverberant ambiance
The sound localization method of allusion quotation study.
Background technique
Document " Structured sparsity models for reverberant speech separation,
Audio,Speech,and Language Processing,IEEE/ACM Transactions on,vol.22,pp.620-
633,2014 " by the multiple microphone signals acquired inside room, outer in the room that multiple grid point estimation room shape is arranged
Shape and multistage imaginary source, secondly estimate the acoustic absorptivity in room, finally using convex optimization using the false reputation source model in room
Data algorithm carries out the research of indoor multi-source separation.However, setting the meeting of mesh point number in space when room reverberation is stronger
It greatly increases, and the unknown parameters that absorb sound, this method is involved that predictor is more, and computational efficiency is low, derivation algorithm essence
Spend position that is not high, thus can not accurately calculating the indoor sound source of acquisition.Therefore, for indoor auditory localization, to unknown room
Parameters,acoustic carries out the precondition that accurate estimation is reasonable estimation room sound source position information.
Summary of the invention
In order to overcome the shortcomings of existing sound localization method poor accuracy, the present invention provides to be based under a kind of strong reverberant ambiance
Parameterize the sound localization method of Bayes's dictionary learning.This method is based on microphone mirror image and simplifies dictionary, in sparse Bayesian
Frame under, realize auditory localization synchronous with room reflections coefficient estimation solution.Dictionary is counted in advance using the method for the present invention
It calculates, completes the available promotion of efficiency of auditory localization and reflection coefficient estimation, be not obviously improved calculation amount in this way
In the case where, improve the precision of sound localization method.The present invention has very big dive to the auditory localization under complex indoor environment
In practical value, the method that parameters,acoustic is estimated in enclosure space also can be used as.
The technical solution adopted by the present invention to solve the technical problems: based on parametrization Bayes under a kind of strong reverberant ambiance
The sound localization method of dictionary learning, its main feature is that the following steps are included:
Step 1: localizing environment selection can be used in rectangle or directly acquire the room of virtual source according to shape.Selected room
Certain interior one-dimensional or two-dimensional surface carries out indoor grid dividing, it is assumed that indoor mesh point number is G.
Step 2: building signal model: it is the position of sound source in a room that the target information obtained is needed when auditory localization,
The quantization of this information is discrete to turn to the limited mesh point for being distributed in the interior space and outdoor mirror image space, each mesh point
Sound source amplitude be xg.Assuming that sound source is present in any one mesh point, the reception signal y (r, f) of any position is expressed as
Formula (1)
In formula, f is current frequency, and c is the velocity of sound in air, and G is indoor mesh point number, βgFor the reflection of corresponding virtual source
Coefficient, xgFor the vector for reflecting sound source position.On G all mesh points, xgOnly seldom some non-zero entry numbers, it is corresponding
In the practical sound source position of room space and the false reputation source position in mirror image space.When constructing dictionary, only portion is set in the room
Mesh point, each dictionary element are that the channel existing for sound source between intranet lattice point and receiving point mirror image is superimposed.Thus it obtains
yf=Afxf (4)
Formula (2) right end indicates total superposition of microphone and its contribution of mirror image acoustic energy, and formula (3) is that dictionary atom computing is public
Formula, N0For receiving point mirror image series,For ns,iThe corresponding sound pressure reflection coefficient of rank mirror image.Arrangement formula (3) obtains formula (4).Formula
In, yf∈CM×1,xf∈CG×1,Af∈CM×G。
Step 3: receiving point mirror image solves: for having a size of lx × ly × lz rectangular room, it is assumed that receiving point position is
(x0,y0,z0), the position of certain grade of receiving point mirror image is (xi,yi,zi), l, m, n are integer of the boundary between-∞ and+∞, then appoint
The space coordinate of meaning receiving point mirror image is expressed as
xi=2llx ± x0
yi=2mly ± y0 (5)
zi=2nlz ± z0
The reflection series of mirror image is given by
N=| 2l- Δl|+|2m-Δm|+|2n-Δn| (6)
X is sought in Δ l=1 expressioniX in formula0Before take it is negative;X is sought in Δ l=0 expressioniX in formula0Before take just;Y is sought in Δ m=1 expressioniFormula
Middle y0Before take it is negative;Y is sought in Δ m=0 expressioniY in formula0Before take just;Z is sought in Δ n=1 expressioniZ in formula0Before take it is negative;Z is sought in Δ n=0 expressioni
Z in formula0Before take just.
Step 4: construction solves dictionary: by the dictionary A under the different quantized values of formula (3) construction reflection coefficient βfAnd its
To the first derivative A' of βfWith matrix of second derivatives A "f。
Step 5: position vector xfSolve: when carrying out indoor acoustic fix ranging, the mesh point number G in room is not only much
Greater than microphone number M, and it is greater than sound source number.That is xfWith spatial sparsity, contained by the estimation of sparse Bayesian method
The x of sound source location information.Observation signal is obtained by formula (7) sampling.
IMFor the unit matrix of M dimension, xfIt is obtained by following formula:
xf| α~CN (0, Λ-1I) (8)
Wherein CN table in above formula
It gives instructions in reply Gaussian Profile, Gamma indicates gamma distribution.Above-mentioned probabilistic model is solved by variation Bayes-expectation-maximization algorithm.
Wherein xfPosterior probability obtained by formula (9):
Indicate the estimated value to dependent variable.
The method for providing each parameter Estimation directly below, whereinObeying mean value is
Gamma distribution.α0Obeying mean value isHeight
This distribution, the mark of Tr representing matrix.
Step 6: the estimation of parameter beta: including reflection coefficient β parameter in above-mentioned each dictionary, estimate the parameters equivalent in formula
(10) it minimizes:
The optimization problem is solved by Newton method, and iterative formula is as follows:
In formula, A'fIt indicates to corresponding parameter beta derivation, A "fIndicate second dervative.By in sparse Bayesian frame
Dictionary introduces reflection coefficient and solves the problems, such as this as parameter, and by Newton method iteration, is solved by substep, realization is being estimated
Sound source position xfWhile, estimate reflection coefficient β.On the one hand the estimation for changing parameter is room acoustics parameter Estimation, simultaneously
Also more accurate room acoustics information is obtained, better auditory localization effect is reached.
Step 7: parameter beta quantifies: in the case where room information is totally unknown, the initial value of β is set as 0.5, is joined by band
Several Bayes's Sparse methods are utilized respectively formula (9) and formula (10) to sound source position xfEstimated with reflection coefficient β, is reflected
Coefficient estimation and the estimation of sound source position carry out simultaneously, i.e., are repeated between formula (9) and formula (10), every step iteration is all
It needs to estimate reflection coefficient and immediate dictionary is found by quantization and carries out next step iteration, it is more that β difference will lead to dictionary
It is secondary to recalculate, bring huge calculation amount.
In order to avoid computing repeatedly dictionary, β is quantified as several discrete data points in its codomain range (0~1) first,
Then the corresponding dictionary of the value of these discrete points is calculated before iteration, the β being calculated is quantified to corresponding discrete
In value and call corresponding dictionary that can substantially reduce calculation amount.
The beneficial effects of the present invention are: this method, which is based on microphone mirror image, simplifies dictionary, under the frame of sparse Bayesian,
Realize that auditory localization estimation synchronous with room reflections coefficient solves.Dictionary is precalculated out using the method for the present invention, it is complete
At the available promotion of efficiency that auditory localization and reflection coefficient are estimated, in this way in the case where not being obviously improved calculation amount,
Improve the precision of sound localization method.The present invention has very big potential practical valence to the auditory localization under complex indoor environment
Value also can be used as the method that parameters,acoustic is estimated in enclosure space.
Different signal-to-noise ratio is done to obtained acquisition signal respectively to handle and position, select frequency range be 1kHz~
2kHz.Calculating time under statistics the method for the present invention and literature method respectively, table 1 show that correlation method is calculated at same
The time-consuming situation an of location tasks is completed, two kinds of time-consumings that the method for the present invention provides, one is not quantify reflection coefficient condition
Under (left side) is calculated, one is quantization reflection coefficient under the conditions of (right side) is calculated.
The comparison of 1 distinct methods time-consuming of table
Table 1 shows while accurately estimation reflection coefficient, calculates time-consuming fewer than literature algorithms time-consuming.The method of the present invention
Calculate the time is not influenced by RMR room reverb degree, it only has direct relationship, Grid dimension in the Grid dimension that room divides
Mesh is more, and solution vector dimension is higher, and it is more to expend the time.
It elaborates with reference to the accompanying drawings and detailed description to the present invention.
Detailed description of the invention
Fig. 1 is the process of the sound localization method based on parametrization Bayes's dictionary learning under strong reverberant ambiance of the invention
Figure.
Fig. 2 is the method for the present invention test macro connection block diagram.
Fig. 3 is the sound source information vector x iterative convergent process under 10*10 grid.
Fig. 4 is the sound source information vector x iterative convergent process under 20*20 grid.
It is bent that the reflection coefficient that Fig. 5 is different signal-to-noise ratio lower sensor numbers when being 10*10 iteratively solves mean square error (MSE)
Line.
It is bent that the reflection coefficient that Fig. 6 is different signal-to-noise ratio lower sensor numbers when being 20*20 iteratively solves mean square error (MSE)
Line.
In figure, MSEx, MSEbeta, MSEx0Respectively correspond x obtained by x, β and literature method0。
Specific embodiment
Referring to Fig.1-6.Based on the sound localization method tool of parametrization Bayes's dictionary learning under the strong reverberant ambiance of the present invention
Steps are as follows for body:
Step 1: localizing environment is chosen: this method, which is only available for rectangle or other, according to shape can directly acquire virtual source
Room.The indoor grid dividing of progress auditory localization in selected room in certain one-dimensional or two-dimensional surface, it is assumed here that indoor net
Lattice point number is G.
Step 2: building signal model: it is the position of sound source in a room that the target information obtained is needed when auditory localization,
The quantization of this information is discrete to turn to the limited mesh point for being distributed in the interior space and outdoor mirror image space, each mesh point
Possible sound source amplitude is xg.Assuming that sound source is likely to be present in any one mesh point, the reception signal y of any position (r,
F) it is represented by formula (1)
F is current frequency in formula, and c is the velocity of sound in air, and G is the total number of mesh point, βgFor the reflection of corresponding virtual source
Coefficient, xgFor the vector for reflecting sound source position.On all possible G mesh point, xgOnly seldom some non-zero entry numbers,
Corresponding to the practical sound source position of room space and the false reputation source position in mirror image space.When constructing dictionary, only portion in the room
Mesh point is set, each dictionary element is that sound source intranet lattice point that may be present is superimposed with the channel between receiving point mirror image.
Therefore it can obtain
yf=Afxf (4)
Formula (2) right end is expressed as total superposition of the acoustic energy contribution of microphone and its mirror image, and formula (3) is dictionary atom computing
Formula, N0 are receiving point mirror image series,For ns,iThe corresponding sound pressure reflection coefficient of rank mirror image.Arrangement formula (3) can obtain formula (4).
In formula, yf∈CM×1,xf∈CG×1,Af∈CM×G。
Step 3: receiving point mirror image solves: for having a size of lx × ly × lz rectangular room, it is assumed that receiving point position is
(x0,y0,z0), the position of certain grade of receiving point mirror image is (xi,yi,zi), l, m, n are integer of the boundary between-∞ and+∞, then appoint
The space coordinate of meaning receiving point mirror image is represented by
xi=2llx ± x0
yi=2mly ± y0 (5)
zi=2nlz ± z0
The reflection series of mirror image can be given by
N=| 2l- Δl|+|2m-Δm|+|2n-Δn| (6)
X is sought in Δ l=1 expressioniX in formula0Before take it is negative;X is sought in Δ l=0 expressioniX in formula0Before take just;Y is sought in Δ m=1 expressioniFormula
Middle y0Before take it is negative;Y is sought in Δ m=0 expressioniY in formula0Before take just;Z is sought in Δ n=1 expressioniZ in formula0Before take it is negative;Z is sought in Δ n=0 expressioni
Z in formula0Before take just.
Step 4: construction solves dictionary: by the dictionary A under the different quantized values of formula (3) construction reflection coefficient βfAnd its
To the first derivative A' of βfWith matrix of second derivatives A "f。
Step 5: position vector xfSolve: when carrying out indoor acoustic fix ranging, the mesh point number G in room is not only much
Greater than microphone number M, and it is greater than sound source number.That is xfWith spatial sparsity, contained by the estimation of sparse Bayesian method
The x of sound source location information.Observation signal is obtained by formula (7) sampling.
yf|xf,α0~CN (Afxf,α0 -1IM) (7)
IMFor the unit matrix of M dimension, xfIt is obtained by following formula:
xf| α~CN (0, Λ-1I) (8)
Wherein CN table in above formula
It gives instructions in reply Gaussian Profile, Gamma indicates gamma distribution.Above-mentioned probabilistic model can be asked by variation Bayes-expectation-maximization algorithm
Solution.Wherein xfPosterior probability can be obtained by formula (9):
Indicate the estimated value to dependent variable.
The method for providing each parameter Estimation directly below, whereinObeying mean value is
Gamma distribution.α0Obeying mean value isHeight
This distribution, the mark of Tr representing matrix.
Step 6: the estimation of parameter beta: including reflection coefficient β parameter in above-mentioned each dictionary, estimate that the parameter can be equivalent to
Formula (10) minimizes:
The optimization problem can be solved by Newton method, and iterative formula is as follows:
A' in formulafIt indicates to corresponding parameter beta derivation, A "fIndicate second dervative.By in sparse Bayesian frame
Dictionary introduces reflection coefficient and solves the problems, such as this as parameter, and by Newton method iteration, is solved by substep, it can be achieved that estimating
Count sound source position xfWhile, estimate reflection coefficient β.On the one hand the estimation for changing parameter is room acoustics parameter Estimation, together
When more accurate room acoustics information also can be obtained, can reach better auditory localization effect.
Step 7: parameter beta quantifies: in the case where room information is totally unknown, the initial value of β is traditionally arranged to be 0.5, passes through
Bayes's Sparse methods with parameter are utilized respectively formula (9) and formula (10) to sound source position xfEstimated with reflection coefficient β, instead
It penetrates coefficient estimation and the estimation of sound source position carries out simultaneously, i.e., be repeated between (9) and (10), every step iteration all needs
Estimate reflection coefficient and immediate dictionary is found by quantization and carries out next step iteration, it is multiple that β difference will lead to dictionary
It recalculates, brings huge calculation amount.
In order to avoid computing repeatedly dictionary, β is quantified as several discrete data points in its codomain range (0~1) first
(the uniform quantization method that the present invention takes 1000 points), then calculates the corresponding dictionary of the value of these discrete points before iteration
Come, by the β being calculated quantization in corresponding discrete value and calling corresponding dictionary that can substantially reduce calculation amount.
Application Example:
Room is an enclosing square space.Its length, width and height is respectively lx=4m, ly=5m, lz=3m.
Step 1: selecting sustained height inside closed room, arrange the linear microphone array of 15 array elements, array position
For x=1.44m, z=1.5m is divided into 0.2m between the direction y, and starting point microphone position is y=0.1m.
Step 2: according to room-sized, the plane for selecting z=1.5m is main reference zone, and the direction x, y respectively divides 10 sections
Uniform to carry out grid dividing, i.e. mesh point inside room is 100 (or 20 sections totally 400 points), sizing grid be 40cm ×
50cm(20cm×25cm).Mesh coordinate takes grid element center position coordinates.
Step 3: according to room wall surface material, when obtaining each wall surface acoustic absorptivity or directly measurement room reverberation by tabling look-up
Between determine room average sound pressure reflection coefficient β0。
Step 4: randomly selecting mesh point that one divides in advance as sound source position, connect computer, Yi Jixiang according to attached drawing 2
The acquisition equipment answered.
Step 5: opening all devices and drive sound source, persistently play scheduled white noise signal and set acquired time domain
The sample frequency of signal is simultaneously recorded, and data are saved.Short time discrete Fourier transform is done to each preservation data, obtains each microphone letter
Number frequency domain representation yf。
Step 6: selected coordinate reference points measure the relative position of each microphone in the room, according to room inner mesh
Point divides, and solves the location information r of each microphone mirror imageiAnd the reflection series n of each mirror images,i。
Step 7: according to each microphone mirror position ri, mirror reflection series ns,iAnd each wall surface reflection coefficient initial value β0It carries out
Quantization solves, and according to the corresponding dictionary atom of mesh point each in formula (5) solution room, forms the dictionary A under each Frequency pointfAnd
Its Jacobian matrix A'f、A”f。
Step 8: being based on yfAnd Af, the position vector x under each individual frequency is obtained according to formula (9)fAnd each frequency of joint
The position vector x that point data solvest, the estimation of room reflections coefficient is carried out according to formula (10), is repeated as many times as the step 1 straight
To position vector xtAnd reflection coefficient β is not changing.
Step 9: according to solving result xf、xtThe corresponding mesh point of middle maximum value is legal using we under the Frequency point
The sound source position that position obtains.
In this method reflection coefficient in different initial values and grid dividing the case where iteration convergence referring to Fig. 3, Fig. 4.
Wherein thick dashed line is true value, it can be seen that under different initial values in the case of 10*10,20*20 grid dividing, utilizes the method for the present invention
The true value of reflection coefficient can finally be converged to.
Claims (1)
1. under a kind of strong reverberant ambiance based on parametrization Bayes's dictionary learning sound localization method, it is characterised in that including with
Lower step:
Step 1: localizing environment selection can be used in rectangle or directly acquire the room of virtual source according to shape;Select certain in room
One-dimensional or two-dimensional surface carries out indoor grid dividing, it is assumed that indoor mesh point number is G;
Step 2: building signal model: it is the position of sound source in a room that the target information obtained is needed when auditory localization, this
The quantization of information is discrete to turn to the limited mesh point for being distributed in the interior space and outdoor mirror image space, the sound of each mesh point
Source amplitude is xg;Assuming that sound source is present in any one mesh point, the reception signal y (r, f) of any position is expressed as formula (1)
In formula, f is current frequency, and c is the velocity of sound in air, and G is indoor mesh point number, βgFor the reflection coefficient of corresponding virtual source,
xgReflect the sound source amplitude of g-th of sound source position;On G all mesh points, xgOnly seldom some non-zero entry numbers,
Corresponding to the practical sound source position of room space and the false reputation source position in mirror image space;When constructing dictionary, only portion in the room
Mesh point is set, each dictionary element is that the channel existing for sound source between intranet lattice point and receiving point mirror image is superimposed;Thus
It obtains
yf=Afxf (4)
Formula (2) right end indicates total superposition of microphone and its contribution of mirror image acoustic energy, and formula (3) is dictionary atom computing formula, N0For
Receiving point mirror image series,For ns,iThe corresponding sound pressure reflection coefficient of rank mirror image;Arrangement formula (3) obtains formula (4);In formula, yf∈
CM×1,xf∈CG×1,Af∈CM×G;
Step 3: receiving point mirror image solves: for having a size of lx × ly × lz rectangular room, it is assumed that receiving point position is (x0,
y0,z0), the position of certain grade of receiving point mirror image is (xi,yi,zi), l, m, n are integer of the boundary between-∞ and+∞, then arbitrarily connect
The space coordinate of sink mirror image is expressed as
The reflection series of mirror image is given by
N=| 2l- Δl|+|2m-Δm|+|2n-Δn| (6)
X is sought in Δ l=1 expressioniX in formula0Before take it is negative;X is sought in Δ l=0 expressioniX in formula0Before take just;Y is sought in Δ m=1 expressioniY in formula0
Before take it is negative;Y is sought in Δ m=0 expressioniY in formula0Before take just;Z is sought in Δ n=1 expressioniZ in formula0Before take it is negative;Z is sought in Δ n=0 expressioniIn formula
z0Before take just;
Step 4: construction solves dictionary: by the dictionary A under the different quantized values of formula (3) construction reflection coefficient βfAnd its to β's
First derivative A'fWith matrix of second derivatives A "f;
Step 5: position vector xfSolve: when carrying out indoor acoustic fix ranging, the mesh point number G in room is not only far longer than biography
Sound device number M, and it is greater than sound source number;That is xfWith spatial sparsity, sound source is contained by the estimation of sparse Bayesian method
The x of location information;Observation signal is obtained by formula (7) sampling;
IMFor the unit matrix of M dimension, xfIt is obtained by following formula:
WhereinIn above formulaIt indicates
Multiple Gauss distribution, Gamma indicate gamma distribution;Above-mentioned probabilistic model is solved by variation Bayes-expectation-maximization algorithm;Its
Middle xfPosterior probability obtained by formula (9):
Indicate the estimated value to dependent variable;
The method for providing each parameter Estimation directly below, whereinObeying mean value is
Gamma distribution;α0Obeying mean value isHeight
This distribution, the mark of Tr representing matrix;
Step 6: the estimation of parameter beta: including reflection coefficient β parameter in above-mentioned each dictionary, estimate the parameters equivalent in formula (10)
It minimizes:
The optimization problem is solved by Newton method, and iterative formula is as follows:
In formula, A'fIt indicates to corresponding parameter beta derivation, A "fIndicate second dervative;By to the dictionary in sparse Bayesian frame
It introduces reflection coefficient and solves the problems, such as this as parameter, and by Newton method iteration, solved by substep, realized in estimation sound source
Position xfWhile, estimate reflection coefficient β;On the one hand the estimation for changing parameter is room acoustics parameter Estimation, while also
To more accurate room acoustics information, reach better auditory localization effect;
Step 7: parameter beta quantifies: in the case where room information is totally unknown, the initial value of β is set as 0.5, by with parameter
Bayes's Sparse methods are utilized respectively formula (9) and formula (10) to sound source position xfEstimated with reflection coefficient β, reflection coefficient
Estimation and the estimation of sound source position carry out simultaneously, i.e., are repeated between formula (9) and formula (10), every step iteration requires
Estimation reflection coefficient simultaneously finds immediate dictionary by quantization and carries out next step iteration, and β difference will lead to dictionary and repeatedly weigh
It is new to calculate, bring huge calculation amount;
In order to avoid computing repeatedly dictionary, β is quantified as several discrete data points in its codomain range (0~1) first, then
The corresponding dictionary of the value of these discrete points is calculated before iteration, it will be in the β that be calculated quantization to corresponding discrete value
And call corresponding dictionary that can substantially reduce calculation amount.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611251992.0A CN106842112B (en) | 2016-12-30 | 2016-12-30 | Sound localization method based on parametrization Bayes's dictionary learning under strong reverberant ambiance |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611251992.0A CN106842112B (en) | 2016-12-30 | 2016-12-30 | Sound localization method based on parametrization Bayes's dictionary learning under strong reverberant ambiance |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106842112A CN106842112A (en) | 2017-06-13 |
CN106842112B true CN106842112B (en) | 2019-04-19 |
Family
ID=59113714
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611251992.0A Active CN106842112B (en) | 2016-12-30 | 2016-12-30 | Sound localization method based on parametrization Bayes's dictionary learning under strong reverberant ambiance |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106842112B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2019075135A1 (en) * | 2017-10-10 | 2019-04-18 | Google Llc | Joint wideband source localization and acquisition based on a grid-shift approach |
CN111007458B (en) * | 2018-10-08 | 2022-03-11 | 哈尔滨工业大学 | Radiation source direct positioning method combining arrival time difference and arrival angle information under sparse Bayesian framework |
CN110426569B (en) * | 2019-07-12 | 2021-09-21 | 国网上海市电力公司 | Noise reduction processing method for acoustic signals of transformer |
CN113534161B (en) * | 2021-05-31 | 2023-07-04 | 中国船舶重工集团公司第七一五研究所 | Beam mirror image focusing method for remotely positioning underwater sound source |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104168228A (en) * | 2014-08-06 | 2014-11-26 | 哈尔滨工业大学深圳研究生院 | Compressed sensing ultra-wide band channel estimation method and system based on cluster position set |
CN105044722A (en) * | 2015-08-03 | 2015-11-11 | 西安电子科技大学 | Full Bayes feature extraction method for synthesizing aperture radar object |
CN105389550A (en) * | 2015-10-29 | 2016-03-09 | 北京航空航天大学 | Remote sensing target detection method based on sparse guidance and significant drive |
CN105913063A (en) * | 2016-04-01 | 2016-08-31 | 中国地质大学(武汉) | Sparse expression acceleration method for image data set and device |
-
2016
- 2016-12-30 CN CN201611251992.0A patent/CN106842112B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104168228A (en) * | 2014-08-06 | 2014-11-26 | 哈尔滨工业大学深圳研究生院 | Compressed sensing ultra-wide band channel estimation method and system based on cluster position set |
CN105044722A (en) * | 2015-08-03 | 2015-11-11 | 西安电子科技大学 | Full Bayes feature extraction method for synthesizing aperture radar object |
CN105389550A (en) * | 2015-10-29 | 2016-03-09 | 北京航空航天大学 | Remote sensing target detection method based on sparse guidance and significant drive |
CN105913063A (en) * | 2016-04-01 | 2016-08-31 | 中国地质大学(武汉) | Sparse expression acceleration method for image data set and device |
Non-Patent Citations (1)
Title |
---|
Wideband DOA estimation by joint sparse representation under Bayesian learning framework;Lu Wang et al.;《2015 IEEE International Conference on Digital Signal Processing (DSP)》;20150724;全文 |
Also Published As
Publication number | Publication date |
---|---|
CN106842112A (en) | 2017-06-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106842112B (en) | Sound localization method based on parametrization Bayes's dictionary learning under strong reverberant ambiance | |
CN107247251B (en) | Three-dimensional sound source positioning method based on compressed sensing | |
Kuang et al. | A complete characterization and solution to the microphone position self-calibration problem | |
Gumerov et al. | Computation of the head-related transfer function via the fast multipole accelerated boundary element method and its spherical harmonic representation | |
US8532951B2 (en) | Method for calibrating a transducer array | |
EP3025130B1 (en) | Wide-band acoustic holography | |
KR20140099536A (en) | Apparatus and method for microphone positioning based on a spatial power density | |
Georgiou | Solution of the general moment problem via a one-parameter imbedding | |
CN110927669B (en) | CS multi-sound-source positioning method and system for wireless acoustic sensor network | |
CN112147571B (en) | Sound source azimuth angle estimation method based on regular orthogonal matching pursuit and bat algorithm | |
Traa et al. | Multichannel source separation and tracking with RANSAC and directional statistics | |
CN109343003B (en) | Method for identifying sound source formed by fast iterative shrinking wave beams | |
Pezzoli et al. | Sparsity-based sound field separation in the spherical harmonics domain | |
JP6106571B2 (en) | Sound source position estimating apparatus, method and program | |
JP4293986B2 (en) | Method and system for representing a sound field | |
Kristoffersen et al. | Deep sound field reconstruction in real rooms: introducing the isobel sound field dataset | |
Chen et al. | Broadband sound source localisation via non-synchronous measurements for service robots: A tensor completion approach | |
CN109521392B (en) | Underwater one-dimensional DOA estimation method and device based on non-circular signal and L-shaped linear array | |
Bellows et al. | Spherical harmonic expansions of high-resolution musical instrument directivities | |
US9578433B2 (en) | Method for self-calibration of a set of sensors, in particular microphones, and corresponding system | |
Yu et al. | Adaptive imaging of sound source based on total variation prior and a subspace iteration integrated variational Bayesian method | |
Roy et al. | Robust array calibration using time delays with application to ultrasound tomography | |
CN106842111B (en) | Indoor sound localization method based on microphone mirror image | |
Johnson et al. | Surface shape reconstruction from phaseless scattered acoustic data using a random forest algorithm | |
CN115278496A (en) | Sparse sound source identification method and system for microphone array measurement |
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 |