CN110109058B - Planar array deconvolution sound source identification method - Google Patents
Planar array deconvolution sound source identification method Download PDFInfo
- Publication number
- CN110109058B CN110109058B CN201910366448.8A CN201910366448A CN110109058B CN 110109058 B CN110109058 B CN 110109058B CN 201910366448 A CN201910366448 A CN 201910366448A CN 110109058 B CN110109058 B CN 110109058B
- Authority
- CN
- China
- Prior art keywords
- sound source
- vector
- distribution
- planar array
- matrix
- 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.)
- Expired - Fee Related
Links
Images
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
- G01S5/00—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations
- G01S5/18—Position-fixing by co-ordinating two or more direction or position line determinations; Position-fixing by co-ordinating two or more distance determinations using ultrasonic, sonic, or infrasonic waves
- G01S5/20—Position of source determined by a plurality of spaced direction-finders
Abstract
The invention discloses a method for identifying a planar array deconvolution sound source, which comprises the following steps: 1. calculating a traditional beam forming output result; 2. establishing an equation set between a beam forming output result and sound source distribution; 3. and (3) iteratively solving the sound source distribution, wherein in the step, a source intensity distribution solving method suitable for the planar array is constructed by combining a sound source identification problem on the basis of generalized sparsity adaptive matching pursuit gSAMP. The invention has the technical effects that: the invention has high spatial resolution, can effectively remove side lobes, accurately positions each sound source, has positioning precision superior to that of the prior OMP-DAMAS method, and does not need prior knowledge of the sparsity of sound source signals.
Description
Technical Field
The invention belongs to the technical field of sound field identification, and particularly relates to a sound source identification method of a planar array.
Background
A sound source identification method based on beam forming of a microphone array can reliably identify a sound source, and has been widely used, among which The deconvolution imaging method for The mapping of acoustic sources (DAMAS) is The most classic. The documents "Orthogonal matching pursuit to the decoding of the acoustic source inverse [ J ]", Padois T, Berry a, Journal of the acoustic source of America,2015,138(6) ("Orthogonal matching pursuit is applied to the deconvolution acoustic source imaging inverse problem [ J ]", Padois T, Berry a, american acoustics, 2015,138(6): 3678.) propose the OMP-dam method, applying an Orthogonal matching pursuit algorithm (Orthogonal matching pursuit, OMP) to the solution of the inverse problem of deconvolution acoustic source imaging, improving the recognition performance of the deconvolution method, OMP selecting only one column of the most correlated residuals, i.e. the true columns of the acoustic source, to reconstruct the original signal to the mas. However, the existing OMP-DAMAS method requires a priori knowledge of the sparsity of the sound source (the a priori knowledge means that the number of noise sources needs to be determined in advance, and the number of iterations is determined according to the number, but the number is not easy to be determined in practical application), and when a multi-sound source is identified, a target column is easy to be mistakenly selected as a column adjacent to the target column, so that the identification accuracy is deteriorated, and the positioning is deviated.
Disclosure of Invention
Aiming at the problems in the prior art, the technical problem to be solved by the invention is to provide a planar array deconvolution sound source identification method, which does not use the prior knowledge of the sparsity of sound source signals and can improve the positioning accuracy of the sound source.
The technical problem to be solved by the invention is realized by the technical scheme that the technical scheme comprises
The calculation of the beamforming output is:
wherein C ═ Σr′C(r′)=∑r′q(r′)v*(r′)vT(r') is a full cross-spectrum matrix in the frequency domain;
r ' is a sound source position vector, q (r ') is the source intensity at r ', and superscripts T and x respectively represent transposition and conjugation; v ═ vm(r)]A steering column vector, v, representing the focus point rm(r) denotes a steering vector of the mth microphone; l is a matrix with all elements 1, w [ | v [ ]m|]2;
b=Aq
Where b ═ b (r) is an N-dimensional beamforming output column vector, and N is the number of grid points; a ═ psf (r | r') ] is a point spread function matrix of dimension N × N; q is an N-dimensional sound source distribution column vector;
2) Calculating the correlation coefficient h ═ AHrt-1The superscript H denotes the transpose conjugate, and the indices corresponding to the first S elements of the absolute value in H are chosen as { λ }i}i=1,2…S;
3) Let gamma bet=Γt-1∪{λi}i=1,2…S(ii) a Calculating a least squares solution:
θtis an N-dimensional source intensity distribution estimation vector to be solved,according to gammatThe included indexes select an intermediate matrix constructed by corresponding columns in the propagation function matrix A;
5) let Jt=Jt-1∪μ,Λt=Λt-1∪aμ,aμThe μ th column vector of A; updating the source intensity distribution estimation again:
7) If | rt‖2>E, enabling t to be t +1, returning to the step 2), and otherwise, outputting a result;
8) obtaining the estimated value q of the sound source distribution vector at JtHas a non-zero value of JtIs the final index set with a value of
Preferably, in step 3, 2), the parameter S is S < D and S < M/D, D is the number of sound sources, and M is the number of microphones.
Preferably, in step 3, 7), ε is selected to be 10-3To 10-6A value within the range.
The invention has the technical effects that:
the invention constructs a deconvolution sound source identification method suitable for a planar array on the basis of generalized sparsity adaptive matching pursuit. According to simulation and experimental verification: the invention has high spatial resolution, can effectively remove side lobes, accurately positions each sound source, has positioning precision superior to that of the prior OMP-DAMAS method, does not need prior knowledge of the sparsity of sound source signals, and has overall performance superior to that of the OMP-DAMAS method.
Drawings
The drawings of the invention are illustrated as follows:
FIG. 1 is a planar array sampling model of the present invention;
FIG. 2 is a flow chart of an iterative solution algorithm of the present invention;
FIG. 3 is a diagram of the sound source identification effect of the simulation of the present invention and OMP-DAMAS;
FIG. 4 is a graph of experimental identification imaging of the invention and OMP-DAMAS.
Detailed Description
The invention is further illustrated by the following examples in conjunction with the accompanying drawings:
the invention comprises the following steps:
As shown in fig. 1, assuming that sound sources to be measured all fall on a known sound source plane, a microphone plane array is located right in front of the sound source plane, and a sound pressure signal received by a microphone is changed into a full cross-spectrum matrix C in a frequency domain after being subjected to inverse fourier transform, which is expressed by the following formula:
C=∑r′C(r′)=∑r′q(r′)v*(r′)vT(r′) (1)
in the formula (1), r ' is a sound source position vector, q (r ') is the source intensity at the position of r ', and superscripts T and x respectively represent transposition and conjugation; v ═ vm(r)]A steering column vector, v, representing the focus point rm(r) denotes a steering vector of the mth microphone, which is expressed as follows:
in the formula (2), the reaction mixture is,denotes a complex unit, k ═ 2 π f/c denotes a wave number, f denotes an acoustic frequency, and c denotes a wave velocity. r denotes the location vector from the center of the array to the focal point, rmRepresenting the location vector from the center of the array to the mth microphone.
According to the document "four typical beam forming sound source identification and clarification methods [ J ]", Yangye, Church, data collection and processing, 2014,29(2): 316-:
in formula (3), l is a matrix in which all elements are 1, and w [ | v [ ]m|]2。
Substituting formula (1) for formula (3) to obtain:
in equation (4), psf (r | r ') is called an array point propagation function, and indicates a contribution amount of a point sound source having unit source intensity to beam-forming sound pressure generated at the r position when only the point sound source exists at the r' position, and when r 'is r', psf is 1. It can be seen that the output quantity of the beam forming at the r position can be expressed as the sum of the beam forming sound pressure contribution of each sound source point at the point multiplied by the corresponding source intensity, and a linear equation set among the beam forming output result, the array point propagation function and the sound source distribution is constructed according to the deconvolution algorithm:
b=Aq (5)
in formula (5), b ═ b (r) is an N-dimensional beamforming output column vector, and N is the number of mesh points; a ═ psf (r | r') ] is a point spread function matrix of dimension N × N; q is an N-dimensional sound source distribution column vector; where A, b is known and q is unknown.
For equation (5), the existing DAMAS method adopts the gaussian seidel iteration method to solve the sound source intensity distribution, and OMP-DAMAS reconstructs q by an orthogonal matching pursuit algorithm by using the sparse characteristic of the sound source distribution vector.
In the literature, "generalized sparsity adaptive matching pursuit algorithm for CS [ J/OL]"Mayushuang, Liu Cui Xiang, Guo Shitao, Wangbaozhu, computer engineering and application (first network thesis, release address ishttp://kns.cnki.net/kcms/ detail/11.2127.TP.20181115.1637.006.htmlAnd the network initial date is 2018-11-19), on the basis of the generalized sparsity adaptive matching pursuit gSAMP, in combination with the sound source identification problem, the iterative solving process of the invention is shown in FIG. 2, and the steps are as follows:
1) 1, the initialization t, and the residual rt-1B, supporting setMatrix Λt-1=0;Λt-1Is a matrix with elements of 0, and Γ and J are both empty sets/matrices, which are intermediate quantities in the iterative process.
2) Calculating the correlation coefficient h ═ AHrt-1The superscript H denotes the transpose conjugate, and the indices corresponding to the first S elements of the absolute value in H are chosen as { λ }i}i=1,2…S;
3) Let gamma bet=Γt-1∪{λi}i=1,2…S(ii) a Calculating a least squares solution:
θtis an N-dimensional source intensity distribution estimation vector to be solved,according to gammatThe included indexes select an intermediate matrix constructed by corresponding columns in the propagation function matrix A;
this equation represents solving for a vector θtSo thatTo a minimum (b andknown), the calculation result isθtIt is in fact an unknown in the equation.Is calculated as a pair oftAn estimate of (d).
5) let Jt=Jt-1∪μ,Λt=Λt-1∪aμ,aμThe μ th column vector of A; updating the source intensity distribution estimation again:
7) If | rt‖2>E, enabling t to be t +1, returning to the step 2), and otherwise, outputting a result;
8) obtaining the estimated value q of the sound source distribution vector at JtHas a non-zero value of JtIs the final index set with a value of
Each iteration is at JtThis set is added with a new index, which can be regarded as the sound source coordinates, i.e. JtIs the set of coordinates that ultimately identifies all sound sources.
The above-mentioned iterative formula involves the selection of two parameters S and epsilon; wherein S is the number of atoms selected in each iteration, and compared with the selection of OMP-DAMAS single atoms (the atoms refer to columns in a point spread function matrix, namely S is always 1), the probability of successfully recovering the original signal can be effectively improved by increasing the number of the atoms selected in each iteration. A recommended formula (S < D and S < M/D) is given for the selection of the parameter S in the method, D is the number of sound sources, and M is the number of microphones. In practical application, S only needs to be a small integer (e.g. 2) greater than 1 to achieve the effect of improving the sound source positioning accuracy.
Epsilon is a threshold value for stopping iteration when residual error rtWhen the 2 norm is attenuated to a certain value, the original signal can be approximated by the selected atomsLike a sparse representation. When epsilon is selected to be a large value, iteration can be quickly stopped after a main sound source is found; when epsilon is selected to be smaller, iteration can be stopped after finding a secondary sound source, and 10 is generally selected-3To 10-6A value within the range.
Simulation test
In order to verify the accuracy of the method, the method is compared with OMP-DAMAS, the performance improvement of the method is explored, and sound source identification simulation is carried out.
At a specific position, a point sound source of radiation intensity and frequency sound wave is set, a plane area 2m × 2m in front is identified by an array, the sound source plane is 1m away from the array, and the sound source plane is discretized into 21 × 21 grid points. The simulation assumes that the sound source is 6 incoherent monopole point sound sources, the coordinates of which are (-0.6,0,1), (0.6,0,1), (-0.2,0.5,1), (0.2,0.5,1), (-0.2, -0.5,1), (0.2, -0.5,1) m, the imaging results of the sound source under the conditions of 2000Hz, 4000Hz and 6000Hz are respectively calculated, the effective sound pressure is 0.02Pa, the corresponding sound pressure level is 60dB, the cross-spectrum matrix of the sound pressure signals received by each microphone is calculated according to the formula (1) in the forward direction, and Gaussian white noise with the signal-to-noise ratio of 10dB is added; and setting a focusing sound source plane, reconstructing sound source intensity distribution and imaging respectively based on the OMP-DAMAS and the method. Wherein OMP-DAMAS sets the iteration number to be 6 according to the prior knowledge. The method sets the maximum iteration times to be 20 times, takes S to be 4 and epsilon to be 10-6. The imaging result is converted in decibels with reference to the reference sound pressure, and the display dynamic range is set to be 10 dB.
The effect of the simulation is shown in fig. 3, where the position of the real sound source is marked by a white cross in fig. 3, and the position of the sound source is indicated by a high-amplitude black dot as the "main lobe".
As can be seen from FIG. 3, OMP-DAMAS has two sources whose locations deviate at a source frequency of 2000Hz, and at a source frequency of 6000Hz, OMP-DAMAS loses one source and reconstructs a virtual pseudolobe at a non-source location. The method accurately positions the sound source position under three frequencies, and obtains a clear sound source imaging image.
Verification test
For testing artificial junctionsAnd if the result is correct, taking the loudspeaker excited by the steady-state signal as a sound source to carry out experimental verification. By usingA 36 channel Combo array of company, 0.65m array diameter, integrated 4958 microphone samples the sound pressure signal. Sound pressure signals received by the microphones are simultaneously acquired by a PULSE 3560D type data acquisition system and transmitted to a PULSE LABSHOP for spectrum analysis to obtain a cross-spectrum matrix of the sound pressure signals, the sampling frequency is set to be 16384Hz, a Hanning window is added to the signals, an average of 64 sections and a 66.7% overlapping rate are adopted, the time length of each section is 0.25s, and the corresponding frequency resolution is 4 Hz.
The sound pressure signals are sampled by the array, frequency domain signals with the frequency of 2000Hz are extracted and are led into a post-processing program written by MATLAB for imaging, and FIG. 4 is an identification imaging graph of a sound source of a test loudspeaker.
As seen from FIG. 4, the position of a sound source in the OMP-DAMAS imaging graph deviates from a grid point, the method accurately identifies the positions of four loudspeakers, and the method improves the positioning accuracy of the sound source.
Claims (3)
1. A planar array deconvolution sound source identification method comprises the following steps,
step 1, calculating the output result of the traditional beam forming
The calculation of the beamforming output is:
wherein C ═ Σr′C(r′)=∑r′q(r′)v*(r′)vT(r') is a full cross-spectrum matrix in the frequency domain;
r ' is a sound source position vector, q (r ') is the source intensity at r ', and superscripts T and x respectively represent transposition and conjugation; v ═ vm(r)]A steering column vector, v, representing the focus point rm(r) denotes a steering vector of the mth microphone; l is a matrix with all elements 1,w=[|vm|]2;
Step 2, establishing an equation set between a beam forming output result and sound source distribution
b=Aq
Where b ═ b (r) is an N-dimensional beamforming output column vector, and N is the number of grid points; a ═ psf (r | r') ] is a point spread function matrix of dimension N × N; q is an N-dimensional sound source distribution column vector;
the method is characterized in that:
step 3, iterative solution of sound source distribution
2) Calculating the correlation coefficient h ═ AHrt-1The superscript H denotes the transpose conjugate, and the indices corresponding to the first S elements of the absolute value in H are chosen as { λ }i}i=1,2...S;
3) Let gamma bet=Γt-1∪{λi}i=1,2...S(ii) a Calculating a least squares solution:
θtis an N-dimensional source intensity distribution estimation vector to be solved,according to gammatThe included indexes select an intermediate matrix constructed by corresponding columns in the propagation function matrix A;
5) let Jt=Jt-1∪μ,At=At-1∪aμ,aμThe μ th column vector of A; updating the source intensity distribution estimation again:
7) If rt||2If the value is more than epsilon, making t equal to t +1 return to the step 2), otherwise, outputting the result;
8) obtaining the estimated value q of the sound source distribution vector at JtHas a non-zero value of JtIs the final index set with a value of θt。
2. The planar array deconvolution sound source identification method of claim 1, wherein: in step 3, 2), the parameters S are S < D and S < M/D, D is the number of sound sources, and M is the number of microphones.
3. The planar array deconvolution sound source identification method of claim 1 or 2, characterized by: in step 3 7) above, ε is selected to be 10-3To 10-6A value within the range.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910366448.8A CN110109058B (en) | 2019-05-05 | 2019-05-05 | Planar array deconvolution sound source identification method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910366448.8A CN110109058B (en) | 2019-05-05 | 2019-05-05 | Planar array deconvolution sound source identification method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110109058A CN110109058A (en) | 2019-08-09 |
CN110109058B true CN110109058B (en) | 2021-04-06 |
Family
ID=67488200
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910366448.8A Expired - Fee Related CN110109058B (en) | 2019-05-05 | 2019-05-05 | Planar array deconvolution sound source identification method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110109058B (en) |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111551943B (en) * | 2020-05-19 | 2022-07-12 | 中国科学院声学研究所 | DAMAS 2-based sparse array high-resolution three-dimensional acoustic imaging method and system |
CN112198476B (en) * | 2020-10-16 | 2023-10-27 | 昆明理工大学 | Three-dimensional positioning method of mobile sound source based on stereoscopic vision and beam forming |
CN113658606B (en) * | 2021-08-19 | 2023-08-08 | 中国人民解放军海军工程大学 | Beam forming method based on self-adaptive compressed sensing under low signal-to-noise ratio condition |
CN113962250B (en) * | 2021-09-14 | 2023-05-12 | 西南交通大学 | Self-adaptive minimum correlation generalized deconvolution method |
CN115113139B (en) * | 2022-05-12 | 2024-02-02 | 苏州清听声学科技有限公司 | Sound source identification method and device based on microphone array and electronic equipment |
CN114741652A (en) * | 2022-06-10 | 2022-07-12 | 杭州兆华电子股份有限公司 | Deconvolution high-resolution imaging method and system based on acoustic image instrument |
CN116008911B (en) * | 2022-12-02 | 2023-08-22 | 南昌工程学院 | Orthogonal matching pursuit sound source identification method based on novel atomic matching criteria |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6178141B1 (en) * | 1996-11-20 | 2001-01-23 | Gte Internetworking Incorporated | Acoustic counter-sniper system |
CN106443587A (en) * | 2016-11-18 | 2017-02-22 | 合肥工业大学 | High-resolution rapid deconvolution sound source imaging algorithm |
CN106483503A (en) * | 2016-10-08 | 2017-03-08 | 重庆大学 | The quick Deconvolution Method of medicine ball array three-dimensional identification of sound source |
CN107063675A (en) * | 2017-06-09 | 2017-08-18 | 中国航发湖南动力机械研究所 | Apparatus for rotating vane vibration measuring accuracy computational methods and system |
CN108596215A (en) * | 2018-04-04 | 2018-09-28 | 中国航发湖南动力机械研究所 | Multi-modal signal resolution separation method, device, equipment and storage medium |
CN108964737A (en) * | 2018-07-02 | 2018-12-07 | 浙江大学 | A kind of super directive property subsurface communication receiver and communication means based on circular array |
CN109409194A (en) * | 2018-08-30 | 2019-03-01 | 中国航发湖南动力机械研究所 | Multi-modal time-domain signal modal separation, damping parameter discrimination method and storage medium |
CN109631756A (en) * | 2018-12-06 | 2019-04-16 | 重庆大学 | A kind of whir source discrimination based on mixing time-frequency domain |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109633551A (en) * | 2019-01-08 | 2019-04-16 | 中国电子科技集团公司第三研究所 | A kind of acoustic array of detectable a variety of acoustic targets |
-
2019
- 2019-05-05 CN CN201910366448.8A patent/CN110109058B/en not_active Expired - Fee Related
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6178141B1 (en) * | 1996-11-20 | 2001-01-23 | Gte Internetworking Incorporated | Acoustic counter-sniper system |
CN106483503A (en) * | 2016-10-08 | 2017-03-08 | 重庆大学 | The quick Deconvolution Method of medicine ball array three-dimensional identification of sound source |
CN106443587A (en) * | 2016-11-18 | 2017-02-22 | 合肥工业大学 | High-resolution rapid deconvolution sound source imaging algorithm |
CN107063675A (en) * | 2017-06-09 | 2017-08-18 | 中国航发湖南动力机械研究所 | Apparatus for rotating vane vibration measuring accuracy computational methods and system |
CN108596215A (en) * | 2018-04-04 | 2018-09-28 | 中国航发湖南动力机械研究所 | Multi-modal signal resolution separation method, device, equipment and storage medium |
CN108964737A (en) * | 2018-07-02 | 2018-12-07 | 浙江大学 | A kind of super directive property subsurface communication receiver and communication means based on circular array |
CN109409194A (en) * | 2018-08-30 | 2019-03-01 | 中国航发湖南动力机械研究所 | Multi-modal time-domain signal modal separation, damping parameter discrimination method and storage medium |
CN109631756A (en) * | 2018-12-06 | 2019-04-16 | 重庆大学 | A kind of whir source discrimination based on mixing time-frequency domain |
Also Published As
Publication number | Publication date |
---|---|
CN110109058A (en) | 2019-08-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110109058B (en) | Planar array deconvolution sound source identification method | |
US8254654B2 (en) | Sidelobe suppression in ultrasound imaging using dual apodization with cross-correlation | |
US8848933B2 (en) | Signal enhancement device, method thereof, program, and recording medium | |
McCowan | Microphone arrays: A tutorial | |
US7867166B2 (en) | Statistical estimation of ultrasonic propagation parameters for aberration correction | |
CN110133579B (en) | Spherical harmonic order self-adaptive selection method suitable for sound source orientation of spherical microphone array | |
Zeng et al. | Eigenspace-based minimum variance beamformer combined with Wiener postfilter for medical ultrasound imaging | |
CN107765221B (en) | Deconvolution sound source imaging method suitable for identifying coherent and incoherent sound sources | |
Bai et al. | Application of convex optimization to acoustical array signal processing | |
KR20210042907A (en) | Method and system for non-invasive characterization of heterogeneous media using ultrasound | |
CN109343003B (en) | Method for identifying sound source formed by fast iterative shrinking wave beams | |
CN111208213A (en) | Spectral seeking sub-band minimum variance ultrasonic imaging algorithm fused with alternative multiplier iteration | |
Morgan et al. | Multi-covariate imaging of sub-resolution targets | |
Blacodon | Spectral estimation method for noisy data using a noise reference | |
Agarwal et al. | Improving spatial resolution using incoherent subtraction of receive beams having different apodizations | |
Jensen et al. | The iterative adaptive approach in medical ultrasound imaging | |
Deylami et al. | Iterative minimum variance beamformer with low complexity for medical ultrasound imaging | |
Salari et al. | User parameter-free minimum variance beamformer in medical ultrasound imaging | |
Schlunk et al. | Combining ADMIRE and MV to improve image quality | |
CN107997784B (en) | Ultrasonic beam synthesis method and system based on sound velocity adaptive correction | |
CN115662383B (en) | Method and system for deleting main sound source, method, system and device for identifying multiple sound sources | |
Hald | Cross-spectral matrix denoising for beamforming in wind tunnels | |
Buerger et al. | The spatial coherence of noise fields evoked by continuous source distributions | |
Mars et al. | Beamforming techniques using microphone arrays | |
Jia et al. | A joint time-space-frequency filtering framework for multichannel speech enhancement via complex-valued tensor representations |
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 | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20210406 |