CN114915875A - Adjustable beam forming method, electronic equipment and storage medium - Google Patents

Adjustable beam forming method, electronic equipment and storage medium Download PDF

Info

Publication number
CN114915875A
CN114915875A CN202210844400.5A CN202210844400A CN114915875A CN 114915875 A CN114915875 A CN 114915875A CN 202210844400 A CN202210844400 A CN 202210844400A CN 114915875 A CN114915875 A CN 114915875A
Authority
CN
China
Prior art keywords
array
fir
complex
filter
sub
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
Application number
CN202210844400.5A
Other languages
Chinese (zh)
Other versions
CN114915875B (en
Inventor
陈华伟
张展
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing University of Aeronautics and Astronautics
Original Assignee
Nanjing University of Aeronautics and Astronautics
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Nanjing University of Aeronautics and Astronautics filed Critical Nanjing University of Aeronautics and Astronautics
Priority to CN202210844400.5A priority Critical patent/CN114915875B/en
Publication of CN114915875A publication Critical patent/CN114915875A/en
Application granted granted Critical
Publication of CN114915875B publication Critical patent/CN114915875B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; DEAF-AID SETS; PUBLIC ADDRESS SYSTEMS
    • H04R1/00—Details of transducers, loudspeakers or microphones
    • H04R1/20—Arrangements for obtaining desired frequency or directional characteristics
    • H04R1/32—Arrangements for obtaining desired frequency or directional characteristics for obtaining desired directional characteristic only
    • H04R1/40—Arrangements for obtaining desired frequency or directional characteristics for obtaining desired directional characteristic only by combining a number of identical transducers
    • H04R1/406—Arrangements for obtaining desired frequency or directional characteristics for obtaining desired directional characteristic only by combining a number of identical transducers microphones
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L21/00—Processing of the speech or voice signal to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
    • G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
    • G10L21/0208—Noise filtering
    • G10L21/0216—Noise filtering characterised by the method used for estimating noise
    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
    • H03H17/00—Networks using digital techniques
    • H03H17/02—Frequency selective networks
    • G—PHYSICS
    • G10—MUSICAL INSTRUMENTS; ACOUSTICS
    • G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
    • G10L21/00—Processing of the speech or voice signal to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
    • G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
    • G10L21/0208—Noise filtering
    • G10L21/0216—Noise filtering characterised by the method used for estimating noise
    • G10L2021/02161—Number of inputs available containing the signal or the noise to be suppressed
    • G10L2021/02166—Microphone arrays; Beamforming
    • H—ELECTRICITY
    • H03—ELECTRONIC CIRCUITRY
    • H03H—IMPEDANCE NETWORKS, e.g. RESONANT CIRCUITS; RESONATORS
    • H03H17/00—Networks using digital techniques
    • H03H2017/0072—Theoretical filter design
    • H03H2017/0081—Theoretical filter design of FIR filters
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; DEAF-AID SETS; PUBLIC ADDRESS SYSTEMS
    • H04R2430/00—Signal processing covered by H04R, not provided for in its groups
    • H04R2430/20—Processing of the output signals of the acoustic transducers of an array for obtaining a desired directivity characteristic

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Signal Processing (AREA)
  • Otolaryngology (AREA)
  • Acoustics & Sound (AREA)
  • Mathematical Physics (AREA)
  • Quality & Reliability (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Human Computer Interaction (AREA)
  • Computational Linguistics (AREA)
  • Multimedia (AREA)
  • Computer Hardware Design (AREA)
  • Circuit For Audible Band Transducer (AREA)
  • Measurement Of Velocity Or Position Using Acoustic Or Ultrasonic Waves (AREA)

Abstract

The invention discloses an adjustable beam forming method, which utilizes the symmetrical characteristic of a complex weighting coefficient of a channel of an FIR sub-filter to equivalently convert the traditional frequency domain design problem into a new dimension reduction design problem, obtains a result of an optimization problem as a completely independent complex weighting value, quickly recovers through a symmetrical relation according to the obtained completely independent complex weighting value to obtain all complex weighting values, and obtains a time domain tap coefficient for hardware realization through fitting; only completely independent tap coefficients are stored according to the symmetrical characteristic of the tap coefficients, the hardware structure is simplified, the times of multipliers are reduced, and the hardware efficiency is improved. The invention obtains the maximum improvement effect under the design of the adjustable beam former with the annular array polynomial structure of even number of microphones, can greatly reduce the problem dimension of the tap coefficient stage of the FIR filter, and obviously quickens the optimization speed; and the storage consumption is lower on the hardware realization, and the hardware consumption and the times of multiplying units are less.

Description

Adjustable beam forming method, electronic equipment and storage medium
Technical Field
The invention belongs to the field of circular microphone arrays, and particularly relates to an adjustable beam forming method, electronic equipment and a storage medium.
Background
In a three-dimensional space, the annular microphone array has 360-degree omnidirectional spatial resolution which is not possessed by a general linear array in the horizontal direction in the same plane as the array. A broadband beamformer of a polynomial structure (see article [2 ]1] Wang, T., Chen, H. Robust design of Farrow-structure-based steerable broadband beamformers with sparse tap weights via convex optimization. J AUDIO SPEECH MUSIC PROC.2015, 14 (2015), https:// doi.org/10.1186/s 13636-015-. Therefore, by adopting a polynomial structure, the function that the pointing angle is continuously adjustable in a space of 360 degrees can be realized by the annular microphone array broadband beam former.
There are many design methods available for the design of wideband beamformers, and the more classical is based on constrained least squares design (see document [2 ]] E. Mabande, A. Schad and W. Kellermann, "Design of robust superdirective beamformers as a convex optimization problem," 2009 IEEE International Conference on Acoustics, Speech and Signal Processing2009, pp. 77-80, doi: 10.1109/icassp.2009.4959524)), which is designed by applying some constraints on the beam shape or array WNG, and minimizing the sum of the squares of the errors between the actual response and the expected response of the array as an objective function. Usually the objective function is a convex problem and the constraints are also convex, so the problem can be solved efficiently using Matlab and CVX toolbox. However, in the broadband beam former with the polynomial structure, because the array element is connected with a plurality of sub-filters in the rear, the array scale is large, the dimension of the required optimization variable is large, the design calculation time is obviously improved, and simultaneously, the hardware cost is increased geometrically and occupies more resources.
Disclosure of Invention
In order to solve the technical problems mentioned in the background art, the present invention provides an adjustable beamforming method.
In order to achieve the technical purpose, the technical scheme of the invention is as follows:
a method of tunable beamforming comprising the steps of: step 1, constructing a problem model for a determined uniform circular ring-shaped microphone array structure, and selecting a reference coordinate system;
step 2, solving the problem of the complex weighting coefficient of the FIR sub-filter channel according to the symmetry of the complex weighting coefficient of the FIR sub-filter channel to carry out dimension reduction design; after dimension reduction, a least square method is used for solving a completely independent FIR sub-filter channel complex weighting coefficient;
step 3, restoring all FIR sub-filter channel complex weighting coefficients through a symmetry relation according to the completely independent FIR sub-filter channel complex weighting coefficients obtained by dimension reduction design in the step 2; according to all the restored FIR sub-filter channel complex weighting coefficients, all FIR sub-filter channel tap coefficients are obtained through least square fitting;
step 4, combining the channels with repeated tap coefficients by using the symmetry of the tap coefficients of the channels of the FIR sub-filter obtained in the step 1, deleting the channels with zero tap values, only storing independent tap coefficients during storage, and multiplexing the tap coefficients in a mode of folding a delay line;
and 5, changing the original array model according to the method in the step 4, so that the wave beam output by the changed array model is consistent with the wave beam parameter output by the original array model.
Preferably, the step 1 of constructing a problem model for a determined uniform circular-ring-shaped microphone array structure specifically includes: for a circular microphone array with an array element number of M, an order of P and a tap length of N and a symmetrical expected response, a problem model is constructed as shown in the following formula:
Figure DEST_PATH_IMAGE001
wherein
Figure 549572DEST_PATH_IMAGE002
Is a vector containing all of the complex weight values,
Figure DEST_PATH_IMAGE003
indicating continuous operating frequency
Figure 404396DEST_PATH_IMAGE004
The bandwidth is discretized intoQAfter one point of the middleqThe frequency at the individual discrete points of the signal,q=1,2,…Qg is a matrix in which the steering vectors of the array are grouped together at different angles of incidence at different pointing angles, b des Vectors representing expected response contributions at different angles of incidence at different pointing angles,
Figure DEST_PATH_IMAGE005
the expression is taken as a two-norm,
Figure 121816DEST_PATH_IMAGE006
indicating a continuous pointing angle
Figure DEST_PATH_IMAGE007
Is discretized intoIAfter the point isiThe value of the angle at each point is, i=1,2,…,I(ii) a s.t. represents a constraint, the first constraint being an array response distortionless constraint,
Figure 224682DEST_PATH_IMAGE008
is composed of
Figure DEST_PATH_IMAGE009
At an angle of incidence of
Figure 976737DEST_PATH_IMAGE009
The vector of the direction of the time of flight,
Figure 205725DEST_PATH_IMAGE010
representing a group delay in the passband of
Figure DEST_PATH_IMAGE011
The second constraint is an array white noise gain constraint, wherein
Figure 410441DEST_PATH_IMAGE012
Is composed of
Figure 343762DEST_PATH_IMAGE009
At an angle of incidence of
Figure 714438DEST_PATH_IMAGE009
The steering vector of the time-domain array to the white noise gain,
Figure DEST_PATH_IMAGE013
a minimum threshold representing the white noise gain that the array is expected to satisfy in the design;
in the optimization problem of the above design method, the variables to be optimized are those containing complex weights of all FIR sub-filter channels
Figure 973381DEST_PATH_IMAGE014
And, according to the symmetry of the array and the expected response, the elements therein have the relationship shown below:
Figure DEST_PATH_IMAGE016A
in the formula (I), the compound is shown in the specification,
Figure 71918DEST_PATH_IMAGE017
indicates that this is the secondmAfter an array elementpComplex weight values of sub-filter channels of order in whichm=1,2,…,M，p=0,1,…P。
Preferably, for a circular microphone array with an array element number of M, an order of P, and a tap length of N, the symmetry of the complex weighting coefficient of the FIR sub-filter channel specifically refers to: when the desired response is to pass the part of the band with the delay term
Figure DEST_PATH_IMAGE018
Time, complex weighting coefficient
Figure 746613DEST_PATH_IMAGE019
The following symmetry relationship occurs:
Figure DEST_PATH_IMAGE020
in the formula (I), the compound is shown in the specification,
Figure 706217DEST_PATH_IMAGE021
representing the taking of conjugate complex numbers;
taking the value as when the expected response delay is expressed as
Figure DEST_PATH_IMAGE022
In the process, the complex-weighted relation real number weight also has the result that the front and back orders of the corresponding channel are opposite, and the result is represented as:
Figure 339323DEST_PATH_IMAGE023
Figure DEST_PATH_IMAGE024
denotes the firstmAfter an array elementpIn the sub-filter path of ordernA tap coefficient of whichm=1,2,...,M，p=1,2,...,P，n=0,1,...,N-1。
Preferably, the step 2 of performing dimension reduction design on the problem of solving the complex weighting coefficient of the FIR sub-filter channel specifically refers to: substituting the symmetrical relation in the step 1 into the original design problem to enable symmetrical matrixes or vectors to be combined or mutually offset, and solving the rewritten optimization problem, wherein the expression is as follows:
Figure 518632DEST_PATH_IMAGE025
in the formula (I), the compound is shown in the specification,
Figure DEST_PATH_IMAGE026
、
Figure 59335DEST_PATH_IMAGE027
and
Figure DEST_PATH_IMAGE028
respectively, indicate the wheat located in the x-axis forward direction, the y-axis forward direction and in the first quadrantThe FIR sub-filter channel connected behind the microphone is weighted in complex number, the center of the array is selected as the origin, the first microphone is placed under the coordinate system in the positive direction of the x axis,
Figure 640489DEST_PATH_IMAGE029
respectively representing the set of guide vectors of all FIR sub-filter channels connected behind the microphone on the positive direction and the negative direction of the x axis in different incidence angle directions under different pointing angles after symmetry combination and simplification,
Figure 146294DEST_PATH_IMAGE031
the guide vectors of all FIR sub-filter channels connected with the microphone in the positive direction of the y-axis are combined and simplified through symmetry and are collected in different incidence angle directions under different pointing angles, and
Figure DEST_PATH_IMAGE032
respectively representing the direction sets of the guide vectors of the microphones positioned in the first quadrant and the second quadrant under different incidence angles after being combined by symmetry; b des Vectors representing expected response contributions at different angles of incidence at different pointing angles,
Figure 140795DEST_PATH_IMAGE033
indicating that the array is at the same pointing angle and incident angle
Figure DEST_PATH_IMAGE034
The response of the array in time is,
Figure 157293DEST_PATH_IMAGE035
representing a group delay in the passband of
Figure DEST_PATH_IMAGE036
In the form of a desired response to the user,
Figure 858532DEST_PATH_IMAGE037
is composed of
Figure DEST_PATH_IMAGE038
At an angle of incidence of
Figure 567862DEST_PATH_IMAGE038
The steering vector of the time-domain array to the white noise gain,
Figure 252922DEST_PATH_IMAGE039
representing the lowest threshold for white noise gain that the array is expected to satisfy in the design.
Preferably, in step 3, all FIR sub-filter channel complex weighting coefficients are restored through the symmetry relationship, and all complex weighting coefficients are restored through the symmetry relationship by using a part of complex weighting values obtained by the dimension reduction design formula in step 2, and the expression is as follows:
Figure DEST_PATH_IMAGE040
in the formula (I), the compound is shown in the specification,
Figure 589223DEST_PATH_IMAGE041
、
Figure DEST_PATH_IMAGE042
and
Figure 144970DEST_PATH_IMAGE043
respectively representing the complex weights of the FIR sub-filter channels connected after the microphones located in the x-axis forward direction, the y-axis forward direction and in the first quadrant,
Figure DEST_PATH_IMAGE044
is a vector containing all of the complex weight values,
Figure 25201DEST_PATH_IMAGE045
is the group delay contained by the expected response; u is one dimension
Figure DEST_PATH_IMAGE046
The matrix of (a) is,
Figure 604081DEST_PATH_IMAGE047
represents rounding up, whereiniGo to the firstjThe elements of the column are respectively
Figure DEST_PATH_IMAGE048
(ii) a The other two matrices are represented as:
Figure 460916DEST_PATH_IMAGE049
Figure DEST_PATH_IMAGE050
after the complex weight value is obtained, a tap coefficient is obtained through least square fitting.
Preferably, the step 4 specifically refers to: after all tap coefficients are obtained, a symmetrical simplified hardware structure is utilized, all FIR filters behind M +2-M array elements with M larger than M/2+1 array elements are removed, the M +2-M array elements are added or subtracted and merged to the back of the M array element, meanwhile, the FIR filters behind the M/4+1 array elements can adopt a structure of folding delay lines, and finally the 1 st and M/2+1 st array element odd-order FIR filters are removed.
An electronic device, comprising: a memory storing a computer program executable by the processor and a processor implementing the above adjustable beamforming method when the computer program is executed by the processor.
A storage medium having a computer program stored thereon, which when read and executed, implements the adjustable beamforming method described above.
Adopt the beneficial effect that above-mentioned technical scheme brought:
compared with the design technology of the existing adjustable broadband beam former with the circular array polynomial structure, the design method has the advantages that the dimension of the required optimization variable is obviously reduced through equivalent dimension reduction design on the premise of not influencing the design result and not sacrificing the performance, so that the problem optimization time in the design process is obviously reduced, and particularly, the design time is greatly shortened under the condition that the number of array elements and the order number of sub-filters are more.
In addition, according to the symmetry relationship of the tap coefficients, when the tap coefficients are stored, only repeated parts can be stored, and the storage space is also remarkably saved. And the tap coefficients of partial channels behind the array elements of the beam former are the same or zero, or have a symmetrical relation in front and back, so that the hardware circuit structure can be simplified, the times of multipliers can be obviously reduced, the hardware cost is reduced, and the functions of reducing the energy consumption of equipment and improving the real-time performance and the stability of a system can be realized.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
fig. 2a and 2b are beam patterns obtained after the original design and the dimension reduction design, respectively, and fig. 2c is a WNG comparison diagram;
FIG. 3 is a schematic diagram of a polynomial wideband beamformer architecture;
FIGS. 4a and 4b are even and odd order diagrams, respectively, after optimizing a hardware structure using tap coefficient symmetry;
FIG. 5 is a comparison between the original design and the reduced dimension design.
Detailed Description
The technical scheme of the invention is explained in detail in the following with the accompanying drawings.
The invention discloses an adjustable beam forming method, which designs an adjustable planar broadband beam forming design with an 8-array element 6-order 512-tap polynomial structure and a tap radius of 0.035m, wherein the sound velocity is 340m/s, the sampling frequency is 8000Hz, the working frequency is 300-3400Hz, WNG constraint is more than or equal to-10 dB, and the expected response group delay is 255.5. The specific process is shown in fig. 1, after selecting a proper coordinate system, solving the problem of the complex weighting coefficient of the channel of the FIR sub-filter according to the symmetry of the complex weighting coefficient of the channel of the FIR sub-filter to perform dimension reduction design; after dimension reduction, the completely independent FIR sub-filter complex weighting coefficient is solved by using the least square rule and the like, and then all complex weighting coefficients can be quickly recovered by using the result of dimension reduction design through the symmetry relation: as shown in the following formula:
Figure 136748DEST_PATH_IMAGE051
therein is in addition to
Figure DEST_PATH_IMAGE052
The complex weights are the optimization results from the dimension reduction design, and the remaining quantities and matrices are all known. After the complex weighted value is obtained, the tap coefficient can be obtained by least square fitting or inverse Fourier transform, so that the design part of the adjustable broadband beam former with the polynomial structure is completed, and the tap coefficients correspondingly contain symmetry.
And finally, respectively solving and obtaining all tap coefficients of the original design and the dimensionality reduction design. After the tap coefficients are obtained, the tap coefficients can be used to simulate the beam pattern of the beam former and the frequency-averaged WNG pattern as an evaluation index.
The originally designed beam pattern is shown in fig. 2a and the reduced-dimension designed beam pattern is shown in fig. 2 b. From the comparison of the beam patterns and the frequency averaged WNG at different angles in fig. 2c, it can be seen that the results obtained after the dimension reduction design are exactly the same, which means that the dimension reduction design does not sacrifice the performance of any beam former. While the design time is reduced from the original time of about 1773 seconds to 488 seconds, which reduces the time of about 73%.
The original circuit structure of the beamformer in hardware is shown in fig. 3, and by using the symmetry of tap coefficients mentioned in the dimension reduction design in step 5, we can use the optimized structures shown in fig. 4a and fig. 4b for even-order and odd-order FIR sub-filter channels in fig. 3, respectively, to replace the original structure shown in fig. 3, and structurally, it can be seen that a considerable portion of the circuit is saved for each FSU.
The savings in specific hardware that result from replacing the original structure with the optimized structure are shown in the following table:
original structure Optimized structure Saving ratio
Number of FIR tap coefficient storage 28672 7424 About 74 percent
FIR channel (with tap and multiplier) 56 channels About 26 channels About 46 percent
In the example of this embodiment, the original polynomial structure has 8 × 5 × 512=20480 tap coefficients to be stored, and the storage space of the tap number can be reduced by about 75% by using the dimension reduction design and symmetry rule of the present invention. In the aspect of hardware structure implementation, due to the adoption of the optimized structure, about 53 percent of the number of taps and the number of multipliers are directly saved, and hardware resources are saved to a greater extent. And the performance can be completely guaranteed not to be influenced, and on the premise, the hardware overhead, the equipment power consumption, the real-time calculation speed and the like can be obviously improved.
The present invention enumerates the situation as the number of array elements varies from 6 to 20. It can be found that when the array scale becomes larger, the advantage of saving design time in dimension reduction design will be more obvious, and fig. 5 shows that as the number of array elements increases, the reduction of dimension reduction design time can be up to more than 80%, and a beam former can be designed more quickly and conveniently. By calculation, the saving ratio in the aspect of hardware implementation is also close to that of the 8-array element in the embodiment, the storage amount of the tap coefficient is reduced by about 75%, and the saving amount of the multiplier is about 50%, which is huge for the saving amount brought by the array with higher scale.
The embodiments are only for illustrating the technical idea of the present invention, and the technical idea of the present invention is not limited thereto, and any modifications made on the basis of the technical scheme according to the technical idea of the present invention fall within the scope of the present invention.

Claims (8)

1. A method of adjustable beamforming, comprising the steps of: step 1, constructing a problem model for a determined uniform circular ring-shaped microphone array structure, and selecting a reference coordinate system;
step 2, solving the problem of the complex weighting coefficient of the FIR sub-filter channel according to the symmetry of the complex weighting coefficient of the FIR sub-filter channel to carry out dimension reduction design; after dimension reduction, a least square method is used for solving a completely independent FIR sub-filter channel complex weighting coefficient;
step 3, restoring all FIR sub-filter channel complex weighting coefficients through a symmetry relation according to the completely independent FIR sub-filter channel complex weighting coefficients obtained by dimension reduction design in the step 2; according to all the restored FIR sub-filter channel complex weighting coefficients, all FIR sub-filter channel tap coefficients are obtained through least square fitting;
step 4, combining the channels with repeated tap coefficients by using the symmetry of the tap coefficients of the channels of the FIR sub-filter obtained in the step 1, deleting the channels with zero tap values, only storing independent tap coefficients during storage, and multiplexing the tap coefficients in a mode of folding a delay line;
and 5, changing the original array model according to the method in the step 4, so that the wave beam output by the changed array model is consistent with the wave beam parameter output by the original array model.
2. The method according to claim 1, wherein the step 1 of constructing a problem model for a determined uniform circular ring shaped microphone array structure specifically comprises: for a circular microphone array with an array element number of M, an order of P and a tap length of N and a symmetrical expected response, a problem model is constructed as shown in the following formula:
Figure 951787DEST_PATH_IMAGE002
wherein
Figure 961331DEST_PATH_IMAGE004
Is a vector containing all of the complex weight values,
Figure 73644DEST_PATH_IMAGE006
indicating continuous operating frequency
Figure 877652DEST_PATH_IMAGE008
The bandwidth is discretized intoQAfter the point is inqThe frequency at the individual discrete points of the signal,q=1,2,…Qg is a matrix in which the steering vectors of the array are grouped together at different angles of incidence at different pointing angles, b des Vectors representing expected response contributions at different angles of incidence at different pointing angles,
Figure 872153DEST_PATH_IMAGE010
the expression is taken as a two-norm,
Figure 419809DEST_PATH_IMAGE012
indicating a continuous pointing angle
Figure 386628DEST_PATH_IMAGE014
Is discretized intoIAfter the point isiThe value of the angle at each point is, i=1,2,…,I(ii) a s.t. represents a constraint, the first constraint being an array response distortionless constraint,
Figure 860072DEST_PATH_IMAGE016
is composed of
Figure 545131DEST_PATH_IMAGE018
At an angle of incidence of
Figure 630899DEST_PATH_IMAGE019
The vector of the direction of the time of flight,
Figure 452224DEST_PATH_IMAGE021
representing a group delay in the passband of
Figure 863614DEST_PATH_IMAGE023
The second constraint is an array white noise gain constraint, wherein
Figure 567128DEST_PATH_IMAGE025
Is composed of
Figure 456587DEST_PATH_IMAGE026
At an angle of incidence of
Figure 397998DEST_PATH_IMAGE027
The steering vector of the array to the white noise gain,
Figure 449130DEST_PATH_IMAGE029
a minimum threshold representing the white noise gain that the array is expected to satisfy in the design;
in the optimization problem of the above design method, the variables to be optimized are those containing complex weights of all FIR sub-filter channels
Figure 341738DEST_PATH_IMAGE031
And, according to the symmetry of the array and the expected response, the elements therein have the relationship shown below:
Figure 34887DEST_PATH_IMAGE033
in the formula (I), the compound is shown in the specification,
Figure 830805DEST_PATH_IMAGE035
indicates that this is the secondmAfter an array elementpComplex weight values of sub-filter channels of order in whichm=1,2,…,M，p=0,1,…P。
3. The method as claimed in claim 1, wherein for a circular microphone array with M array elements, P order and N tap length, the symmetry of the complex weighting coefficients of the FIR sub-filter channels specifically refers to: when the desired response is a portion of the passband, the response with the delay term is expressed as
Figure 380735DEST_PATH_IMAGE037
Time, complex weighting coefficient
Figure 262103DEST_PATH_IMAGE039
The following symmetry relationship occurs:
Figure 493365DEST_PATH_IMAGE041
in the formula (I), the compound is shown in the specification,
Figure 143789DEST_PATH_IMAGE043
representing the taking of conjugate complex numbers;
taking the value as when the expected response delay is expressed as
Figure 802303DEST_PATH_IMAGE045
In the process, the complex-weighted relation real number weight also has the result that the front and back orders of the corresponding channel are opposite, and the result is represented as:
Figure 170968DEST_PATH_IMAGE047
Figure 710314DEST_PATH_IMAGE049
is shown asmFirst after the array elementpIn the sub-filter path of ordernA tap coefficient of whichm=1,2,...,M，p=1,2,...,P，n=0,1,...,N-1。
4. The method according to claim 1, wherein the step 2 of performing dimension reduction design on the problem of solving the complex weighting coefficients of the FIR sub-filter channels specifically refers to: substituting the symmetrical relation in the step 1 into the original design problem, so that symmetrical matrixes or vectors can be merged or mutually offset, and the rewritten optimization problem is represented as follows:
Figure 11983DEST_PATH_IMAGE051
in the formula (I), the compound is shown in the specification,
Figure 106978DEST_PATH_IMAGE053
、
Figure 962938DEST_PATH_IMAGE055
and
Figure 270423DEST_PATH_IMAGE057
respectively representing the complex weights of FIR sub-filter channels connected behind the microphones positioned in the positive direction of the x-axis, the positive direction of the y-axis and in the first quadrant, taking the center of the array as the origin, placing the first microphone under the coordinate system in the positive direction of the x-axis,
Figure 895439DEST_PATH_IMAGE059
the guide vectors respectively representing all FIR sub-filter channels connected with the microphone in the positive direction and the negative direction of the x axis are simplified after being combined by symmetryThe collection in different incident angle directions at different pointing angles,
Figure 161335DEST_PATH_IMAGE061
the guide vectors of all FIR sub-filter channels connected with the microphone in the positive direction of the y-axis are combined and simplified through symmetry and are collected in different incidence angle directions under different pointing angles, and
Figure 239013DEST_PATH_IMAGE063
respectively representing the direction sets of the guide vectors of the microphones positioned in the first quadrant and the second quadrant under different incidence angles after being combined by symmetry; b des Vectors representing expected response contributions at different angles of incidence at different pointing angles,
Figure 678084DEST_PATH_IMAGE065
indicating that the array is at the same pointing angle and incident angle
Figure 157607DEST_PATH_IMAGE067
The response of the array in time is,
Figure 92940DEST_PATH_IMAGE069
represents the group delay within the passband of
Figure 392334DEST_PATH_IMAGE071
In the form of a desired response to the user,
Figure 307201DEST_PATH_IMAGE073
is composed of
Figure 906809DEST_PATH_IMAGE074
At an angle of incidence of
Figure 311246DEST_PATH_IMAGE075
The steering vector of the time-domain array to the white noise gain,
Figure 363515DEST_PATH_IMAGE077
representing the lowest threshold for white noise gain that the array is expected to satisfy in the design.
5. The method according to claim 1, wherein in step 3, all FIR sub-filter channel complex weighting coefficients are restored through a symmetry relationship, and all complex weighting coefficients are restored through a symmetry relationship by using a part of complex weighting values obtained by the dimension reduction design formula in step 2, and the expression is as follows:
Figure 82073DEST_PATH_IMAGE079
in the formula (I), the compound is shown in the specification,
Figure 536188DEST_PATH_IMAGE081
、
Figure 49209DEST_PATH_IMAGE083
and
Figure 87310DEST_PATH_IMAGE084
respectively representing the complex weights of the FIR sub-filter channels connected after the microphones located in the x-axis forward direction, the y-axis forward direction and in the first quadrant,
Figure 609558DEST_PATH_IMAGE086
is a vector containing all of the complex weight values,
Figure 714917DEST_PATH_IMAGE088
is the group delay contained by the expected response; u is one dimension
Figure 664419DEST_PATH_IMAGE090
The matrix of (a) is,
Figure 425701DEST_PATH_IMAGE092
represents rounding up, whereiniGo to the firstjThe elements of the column are respectively
Figure 486061DEST_PATH_IMAGE094
(ii) a The other two matrices are represented as:
Figure DEST_PATH_IMAGE095
Figure 118031DEST_PATH_IMAGE096
after the complex weight value is obtained, a tap coefficient is obtained by least square fitting.
6. The method according to claim 1, wherein step 4 specifically refers to: after all tap coefficients are obtained, a hardware structure is simplified by using symmetry, all FIR filters behind M +2-M array elements, M of which is larger than M/2+1 array elements are removed, the M +2-M array elements are added or subtracted and merged to the back of the M array element, meanwhile, the FIR filters behind the M/4+1 array elements can adopt a structure of folding delay lines, and finally the 1 st and M/2+1 array element odd-order FIR filters are removed.
7. An electronic device, comprising: a memory storing a computer program executable by the processor, and a processor implementing the method of tunable beamforming according to any of the preceding claims 1-6 when executing the computer program.
8. A storage medium having stored thereon a computer program which, when read and executed, implements the adjustable beamforming method of any of claims 1-6.
CN202210844400.5A 2022-07-18 2022-07-18 Adjustable beam forming method, electronic equipment and storage medium Active CN114915875B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210844400.5A CN114915875B (en) 2022-07-18 2022-07-18 Adjustable beam forming method, electronic equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210844400.5A CN114915875B (en) 2022-07-18 2022-07-18 Adjustable beam forming method, electronic equipment and storage medium

Publications (2)

Publication Number Publication Date
CN114915875A true CN114915875A (en) 2022-08-16
CN114915875B CN114915875B (en) 2022-10-21

Family

ID=82772052

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210844400.5A Active CN114915875B (en) 2022-07-18 2022-07-18 Adjustable beam forming method, electronic equipment and storage medium

Country Status (1)

Country Link
CN (1) CN114915875B (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CA2499033A1 (en) * 2004-03-02 2005-09-02 Microsoft Corporation A system and method for beamforming using a microphone array
CN104768100A (en) * 2014-01-02 2015-07-08 中国科学院声学研究所 Time domain broadband harmonic region beam former and beam forming method for ring array
US10536302B1 (en) * 2018-09-05 2020-01-14 Raytheon Company Beamspace nonlinear equalization for spur reduction
CN112020864A (en) * 2018-04-13 2020-12-01 伯斯有限公司 Smart beam control in microphone arrays

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CA2499033A1 (en) * 2004-03-02 2005-09-02 Microsoft Corporation A system and method for beamforming using a microphone array
CN104768100A (en) * 2014-01-02 2015-07-08 中国科学院声学研究所 Time domain broadband harmonic region beam former and beam forming method for ring array
CN112020864A (en) * 2018-04-13 2020-12-01 伯斯有限公司 Smart beam control in microphone arrays
US10536302B1 (en) * 2018-09-05 2020-01-14 Raytheon Company Beamspace nonlinear equalization for spur reduction

Also Published As

Publication number Publication date
CN114915875B (en) 2022-10-21

Similar Documents

Publication Publication Date Title
CN100534221C (en) Directional audio signal processing using an oversampled filterbank
CN100477704C (en) Method and device for acoustic echo cancellation combined with adaptive wavebeam
US20080310646A1 (en) Audio signal processing method and apparatus for the same
CN105590631A (en) Method and apparatus for signal processing
Yan et al. Time-domain implementation of broadband beamformer in spherical harmonics domain
US20130044894A1 (en) System and method for efficient sound production using directional enhancement
CN111415676A (en) Blind source separation method and system based on separation matrix initialization frequency point selection
KR20080073936A (en) Apparatus and method for beamforming reflective of character of actual noise environment
CN108447498A (en) Sound enhancement method applied to microphone array
Rosen et al. FIR-based symmetrical acoustic beamformer with a constant beamwidth
CN107105111A (en) One kind combination proportional affine projection echo cancel method of step-length
Zotter A linear-phase filter-bank approach to process rigid spherical microphone array recordings
CN112201273A (en) Noise power spectral density calculation method, system, equipment and medium
CN114915875B (en) Adjustable beam forming method, electronic equipment and storage medium
CN113889136A (en) Pickup method, pickup device and storage medium based on microphone array
Zhang et al. Selective frequency invariant uniform circular broadband beamformer
JP7254938B2 (en) Combined source localization and separation method for acoustic sources
Málek et al. Sparse target cancellation filters with application to semi-blind noise extraction
CN113763984B (en) Parameterized noise elimination system for distributed multi-speaker
CN112820312A (en) Voice separation method and device and electronic equipment
CN112420068B (en) Quick self-adaptive beam forming method based on Mel frequency scale frequency division
JP7182168B2 (en) Sound information processing device and program
CN116017230A (en) Microphone array, signal processing method, device, equipment and medium thereof
CN112488048A (en) Acoustic signal difference feature fusion method based on deep learning and random array
Pepe et al. Digital filters design for personal sound zones: A neural approach

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