CN107170462A - Hidden method for acoustic based on MVDR - Google Patents
Hidden method for acoustic based on MVDR Download PDFInfo
- Publication number
- CN107170462A CN107170462A CN201710163190.2A CN201710163190A CN107170462A CN 107170462 A CN107170462 A CN 107170462A CN 201710163190 A CN201710163190 A CN 201710163190A CN 107170462 A CN107170462 A CN 107170462A
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- omega
- mtd
- signal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
- 238000000034 method Methods 0.000 title claims abstract description 52
- 238000005516 engineering process Methods 0.000 claims description 25
- 239000011159 matrix material Substances 0.000 claims description 21
- 230000006854 communication Effects 0.000 claims description 10
- 230000000694 effects Effects 0.000 claims description 8
- 238000012545 processing Methods 0.000 claims description 6
- 230000017105 transposition Effects 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 5
- 125000006850 spacer group Chemical group 0.000 claims description 5
- 230000009466 transformation Effects 0.000 claims description 4
- 238000005303 weighing Methods 0.000 claims description 4
- 238000010521 absorption reaction Methods 0.000 claims description 3
- 238000005562 fading Methods 0.000 claims description 3
- 230000002969 morbid Effects 0.000 claims description 3
- 238000003672 processing method Methods 0.000 claims description 3
- 230000002123 temporal effect Effects 0.000 claims description 3
- 238000013519 translation Methods 0.000 claims description 3
- 230000005236 sound signal Effects 0.000 claims description 2
- 235000013399 edible fruits Nutrition 0.000 claims 1
- 230000004044 response Effects 0.000 description 10
- 230000006870 function Effects 0.000 description 8
- 230000009467 reduction Effects 0.000 description 8
- 238000004891 communication Methods 0.000 description 7
- 230000015572 biosynthetic process Effects 0.000 description 4
- 206010027336 Menstruation delayed Diseases 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 210000000988 bone and bone Anatomy 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000000926 separation method Methods 0.000 description 2
- 238000003786 synthesis reaction Methods 0.000 description 2
- PEDCQBHIVMGVHV-UHFFFAOYSA-N Glycerine Chemical compound OCC(O)CO PEDCQBHIVMGVHV-UHFFFAOYSA-N 0.000 description 1
- 241001062009 Indigofera Species 0.000 description 1
- 206010044565 Tremor Diseases 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 230000005764 inhibitory process Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000002955 isolation Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000001208 nuclear magnetic resonance pulse sequence Methods 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000001953 sensory effect Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000000527 sonication Methods 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech or voice signal processing techniques 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
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech or voice signal processing techniques 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
- G10L21/0224—Processing in the time domain
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech or voice signal processing techniques 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
Landscapes
- Engineering & Computer Science (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Computational Linguistics (AREA)
- Quality & Reliability (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Circuit For Audible Band Transducer (AREA)
- Soundproofing, Sound Blocking, And Sound Damping (AREA)
Abstract
The invention discloses a kind of hidden method for acoustic based on MVDR, this method is first with Short Time Fourier Transform, the time-domain signal that sensor array is received resolves into frequency domain sub-band signal, the hidden acoustic filter based on MVDR is constructed on each subband, so that the acoustical signal of sound source is undamped by hidden acoustic filter within from hidden throw, obtain estimating signal eventually through anti-STFT.The present invention has preferable hidden sound effective value.
Description
Technical field
The present invention relates to the hidden audio technology of microphone array, and in particular to a kind of hidden method for acoustic based on MVDR.
Background technology
The research of hidden audio technology has had very long history, and people mainly explore two kinds of sides in acoustical signal sensory field
Method：Bone-conduction microphone and ultrasonic microphone, signal separation techniques and difference microphone array are explored in Underwater Acoustic channels field
Row.
Bone-conduction microphone is the slight vibration of caused incidence bone when talking using people to collect voice signal
Get up to switch to electric signal.Because it is different from conventional microphone by air transmitted pickup sound, so in very noisy environment
In the outflow of sound high-resolution can also be come.Early in before the centuries, people have had many real to bone conduction technology
Using, but the speech quality of early stage bone-conduction microphone is not fine, the speech quality of especially high frequency is poor, so early
The bone-conduction microphone of phase is only used for the conventional microphone of auxiliary, for example, do speech terminals detection using bone-conduction microphone,
To improve the performance of single-channel voice noise reduction.Nearly ten or twenty year, bone-conduction microphone starts really to be paid close attention to by people, its performance
Have and greatly improve.At present, in the market has occurred in that communication headsets of many moneys based on bone-conduction microphone.Wherein, 2013 5
The moon, a kind of bone-conduction microphone was invented by Beijing MeiErSiTong Science Development stock Co., Ltd, realized the product independent research state
The interior breakthrough of zero, has put goods on the market at present.Recently, attention is also obtain similar to other sonic transducers of osteoacusis principle, such as
DAIKIN-D Talk Mic headsets, its operation principle is to pick up speaker's laryngograph signal using highly sensitive microphone,
And it is converted into electric signal.This pickup mode has many similar parts with bone-conduction microphone.Although bone-conduction microphone technology
Breakthrough development has been obtained, but it is not also ideal that its communication headset is promoted at present, is primarily present Railway Project：1) move
State effect is poor；2) cost is high；3) packaging effect is poor；4) lower tone.
Before more than ten years, a collection of scientist of AT＆T Labs of the U.S. devises a kind of ultrasonic microphone.This microphone by
The conventional microphone composition in one small ultrasonic transmitter and one big broadband.During work, ultrasonic transmitter sends a cycle
The wideband pulse sequence of property.The frequency of this wideband pulse signal is between 20kHz to 70kHz, the sound channel reflection through speaker
Afterwards, reflected signal is received by microphone, then the Digital Signal Processing link of rear end using transmission signal and reflected signal come
Estimate the form parameter of sound channel, and then synthesize the voice described in speaker.The maximum technical characterstic of this microphone be operated in it is super
Audio frequency range, therefore do not disturbed by audio signal in the range of sense of hearing perceived frequency, available in solution class cocktail party environment
Voice communication problem.The scientist of AT＆T Labs of the U.S. constructs a prototype system that can be worked, this prototype system
The vowel for receiving and being synthesized is sent in a speech recognition system, 95% discrimination can be obtained.Preliminary hearing
Experiment also confirms that the quality of the vowel of synthesis can reach the speech quality of conventional microphone substantially.Certainly, this microphone is true
Positive practical preceding also many problems need to solve, and maximum technical problem is exactly the sound not protruded for track characteristics such as nasal sound
Element, the speech quality and intelligibility of synthesis be not high.
Hidden audio technology theoretically can be regarded as Signal separator or strengthen a subproblem of problem.In a complexity
Acoustic enviroment in, the signal from some sound source is picked up using microphone, the signal can be contaminated with few exceptions.Root
Noise in the mechanism produced according to pollution, Speech processing is divided into 4 classes：Ambient noise, echo, reverberation and interference.In order to incite somebody to action
Sound source and noise are separated, and are handled for each noise like with specific method：
Ambient noise (Noise)：Ambient noise can not be avoided and ubiquitous, and its presence can have a strong impact on voice letter
Number the perception to spatial information of speech quality, intelligibility and human ear.Ambient noise is generally all relatively stable, that is to say, that
The statistical property of current time spot noise can be replaced with the noise statisticses in historical time.According to the system of signals with noise
Characteristic and the statistical property of noise are counted, wave filter can be designed observation signal is filtered, and then strengthens voice signal, is suppressed
Ambient noise, this technology is referred to as noise reduction technology.Noise reduction technology can utilize single channel pickup system, can also utilize multichannel
Pickup system, they correspond respectively to single channel noise reduction technology and multichannel noise reduction technology.Single channel noise reduction technology is being made an uproar
Sound can cause speech distortion while suppression, Comparatively speaking, and multichannel noise reduction technology is while identical output signal-to-noise ratio is obtained
Speech distortion can be reduced.
Echo (Echo)：Acoustic echo is that the acoustical coupling between microphone and loudspeaker is produced.The presence meeting of echo
Have a strong impact on multi-party duplex interaction.The characteristics of echo is maximum is that sound-source signal is known, as long as can estimate from loudspeaker
To the acoustic propagation channel between microphone, the echo composition in observation signal with regard to microphone pickup can be estimated, by this composition
Estimation subtracted from the signal picked up, it is possible to realize echo cancellor, this technology is referred to as echo cancellation technology.
Reverberation (Reverberation)：Reverberation is due to that the interface reflection (multipath effect) in room environment is caused.Instead
Penetrate and be divided into early reflection and late period reflection.Early reflection (within usual 40ms) can typically carry useful information, such as pass through analysis
The structure of early reflection, can size up the room.In addition, early reflection can also strengthen the harmonic components of music, raising is listened
Sense.But late period reflection can cause Spectrum Distortion, and then cause speech quality, the decline of intelligibility, and sound source can be obscured
Positional information.In voice communication system, late period reflection causes reverberation, so that the quality of voice communication is had a strong impact on, so needing
Want dereverberation technology.A kind of dereverberation technology is to carry out blind estimate to channel first, then recycles balancing technique to realize and goes to mix
Ring；Another technology for suppressing reverberation is super sensing array beamses formation technology, and its general principle is to extract the sound of desired orientation
While source signal, suppress the signal from other directions.Reverberation be from from all directions, therefore it is super point to array can be with one
Suppress reverberation with determining degree.
Interference (Interference) signal from other sound sources：Interference signal is caused by a source noise, in being space
Noise from some direction.In voice communication, surrounding often has many people, and there are other sound sources, therefore each logical
Believe end, the situation of multi-acoustical is inevitable, can be interfered with each other between the signal from different sound sources.The allusion quotation of AF panel
Type technology is beam-forming technology, and its basic thought is the wave filter for being initially formed a spatial domain, then wave filter is responded into maximum
The size of array response that is depended on to the direction of quasiexpectation sound source, the inhibition level to interference on interference radiating way of direction.Language
The target of sound noise reduction, Sound seperation and Wave beam forming is all to separate desired useful signal with other interference signals, so
These technologies may be used to hidden sonication.But the separating property that current isolation technics can be obtained is also very limited, it is impossible to meet
The demand of hidden sound application.
Aforementioned signal separation techniques need to use microphone array (microphone array).For microphone
The research of array has had the history of more than 40 years, and in this four ten years, people have had been developed for many Array Designs
With processing method.According to response theory of the array to sound field, these arrays can be divided into two major classes：Plus type array [additive
Microphone array (AMA)] and difference array [differential microphone array (DMA)].Additivity array
Stock size is larger, and what each microphone was measured is the acoustic pressure of sound field, and whole array beamses formation is also that sonic pressure field is rung
Should.Substantial amounts of work is all the processing method on additivity array on additivity array in current document.In contrast, it is poor
Subarray is that the space differentiation of sonic pressure field is responded, with array sizes are small, beam pattern frequency invariance is preferable, to
The features such as determining that in the case of array element number array directive property can be maximized.
The content of the invention
The technical problem to be solved in the present invention is to provide a kind of hidden method for acoustic based on MVDR, with good stealthy effect
Really.
In order to solve the above technical problems, the present invention is adopted the following technical scheme that：Hidden method for acoustic based on MVDR, this method is
First with Short Time Fourier Transform, the time-domain signal that sensor array is received is resolved into subband signal, the structure on each subband
Make the hidden acoustic filter based on MVDR so that the acoustical signal of sound source is undamped by hidden acoustic filter within from hidden throw,
Obtain estimating signal eventually through anti-STFT.
Further, this method comprises the following steps：
S1：According to parameters such as the structure of array, array element number, the positions of sound source, steering vector is constructed
The signal y that sensor in microphone array is receivedm(k)=xm(k)+vm(k), m=1,2 ..., M, which is divided into, to be had
Certain short time frame for overlapping ratio, frame length can from several milliseconds to tens milliseconds, then in M passage each passage it is every
One frame carries out Short Time Fourier Transform, obtains Ym(ω, i), wherein i represent the i-th frame, then construct
Y (ω, i)=[Y1(ω,i) Y2(ω,i) … YM(ω,i)]T,
S2：With Short Time Fourier Transform, the time-domain signal that sensor array is received is resolved into subband signal；
S3：On the subband that frequency is ω, the hidden acoustic filter based on MVDR is constructed
S4：On subband, the signal of the i-th frame is handled using hidden acoustic filter h (ω)
S5：Using inverse Fourier transform in short-term and overlap-add method, to Z, (ω, i) enters line translation, so as to obtain wave beam shape
Time-domain signal Z (k) after.
Further, with Short Time Fourier Transform, the time-domain signal that sensor array is received is resolved into subband signal
Comprise the following steps that：
Assuming that the spacing between two neighboring microphone is δ, because hidden audio technology is used to pick up closely sound source, therefore assume
There is preferable near-field sound source and interference effect in acoustic enviroment on the microphone array, the sound source distance away from each microphone point
Wei not rs,1, rs,2..., rs,M, the center of array is defined as reference point, distance of the sound source away from reference point is rs, incidence angle is θs,
Then distance of the sound source away from m-th of microphone can be expressed as
Wherein,
Discrete time k is located at, the signal that sound source is sent is x (k), if ignoring the absorption loss in communication process, m-th
Also there is the amplitude fading being inversely proportional with distance in the signal that microphone is picked up, can relative to sound-source signal only phase delay
It is expressed as：
Wherein, xm(k) sound-source signal that m-th of microphone is picked up, v are representedm(k) represent what m-th of microphone was picked up
Noise signal；τmRepresent time delay of m-th of microphone compared to sound source；Because the wave surface of the preferable sound source near field is spherical, τmCan
It is expressed as：
Wherein c represents the velocity of sound in air,
Formula (1) is changed into a frequency domain：
Wherein,Wave number is represented, the π f of ω=2 represent angular frequency；F represents temporal frequency,Represent imaginary number
Unit, Ym(ω)、Xm(ω)、Vm(ω) represents y respectivelym(k)、xm(k)、vm(k) Fourier transformation.
Further, according to parameters such as the structure of array, array element number, the positions of sound source, construction length is sweared for M guiding
Amount：Superscript T represents vectorial transposition computing, makes r=rs, θ=θs, then,The signal that M microphone is picked up is with vector representation：
Y (ω)=[Y1(ω) Y2(ω) … YM(ω)]T
=ds(ω,rs,θs)X(ω)+v(ω),
Y (ω) passes through ARRAY PROCESSING, and obtained output signal is:
Wherein, Z (ω) is sound-source signal X (ω) estimation；Subscript H represents conjugate transposition computing, h (ω)=[H1(ω)
H2(ω) … HM(ω)]TRepresent wave filter of the microphone array to the weighing vector of input signal, that is, microphone array
Coefficient.
Further, in S3 on the subband that frequency is ω, hidden acoustic filter method of the construction based on MVDR is as follows：Do not wane
Subtract closely sound-source signal while the variance for minimizing array output end residual noise has just obtained the hidden acoustic filters of MVDR, mathematics
Expression formula is as follows：
Formula (2) can be solved using method of Lagrange multipliers, is obtained
Further, it is assumed that noise is isotropic noise, (m, n) individual element of its normalized correlation matrix can be with
It is written as form：
Wherein,There are two kinds of extreme situations：If 1) ω τ0It is very big, i.e., high frequency or greatly
Under spacer conditions, the noise signal that two sensors are received is close to uncorrelated, and isotropic noise is close to space white noise；2)
If ω τ0Very small, i.e., under low frequency or small spacer conditions, the noise signal that two sensors are received is just close relevant,
Isotropic noise points of proximity source noise；Hidden acoustic filter based on MVDR has the computing of a matrix inversion, works as microphone array
When first number is more, morbid state occurs in matrix, and extremely unstable situation occurs in wave filter, and in order to avoid inverting, shakiness is pledged love
The generation of condition, using traditional way：Diagonal loading technique, above-mentioned matrix is added with pair of horns matrix, after diagonal loading
Matrix can be expressed as [Γdn(ω)+ε Ι], wherein Ι is the unit matrix that size is M, and ε is loading coefficient, now based on MVDR
Hidden acoustic filter coefficient be：
Beneficial effects of the present invention：The present invention is minitype microphone array in platform, and its core is first with Fu in short-term
Leaf transformation, subband signal is resolved into by the time-domain signal that sensor array is received, and appropriate hidden sound filter is constructed on each subband
Ripple device so that the acoustical signal of sound source is undamped by hidden acoustic filter within from hidden throw, and method of the invention has very
Good hidden sound effective value.
Brief description of the drawings
In order to more clearly illustrate the technical scheme in the embodiment of the present invention, below by using required in embodiment
Accompanying drawing is simply introduced, it should be apparent that, drawings in the following description are only some embodiments described in the present invention, for
For those of ordinary skill in the art, on the premise of not paying creative work, other can also be obtained according to these accompanying drawings
Accompanying drawing.
Fig. 1 is hidden acoustic model figure.
Fig. 2 is the system schematic that minitype microphone array handles voice.
Fig. 3 is that uniform linear array is schemed to the pickup of near field acoustical signal and processing.
Fig. 4 is the hidden acoustic attenuation curve map of the hidden method for acoustic based on 8 array element MVDR.
Fig. 5 is the hidden method for acoustic based on 8 array element MVDR that frequency is 1300Hz, 3300Hz
Beam pattern (indigo plant, red, purple, black represent r=5cm, 10cm, 30cm, 50cm respectively).
Fig. 6 is sound source and end of the hidden method for acoustic based on 8 array element MVDR to end-on direction 0cm
Penetrate the signal interference ratio figure of the interference in direction.
Embodiment
Technical scheme will be clearly and completely described by embodiment below.
The hidden method for acoustic based on MVDR of the present invention, this method is first with Short Time Fourier Transform, by sensor array
The time-domain signal received resolves into subband signal, appropriate hidden acoustic filter is constructed on each subband so that from hidden throw
The acoustical signal of sound source is undamped by hidden acoustic filter within, obtains estimating signal eventually through anti-STFT.
The method of the present invention comprises the following steps：
S1：According to parameters such as the structure of array, array element number, the positions of sound source, steering vector is constructed
S2：The signal y that sensor in microphone array is receivedm(k)=xm(k)+vm(k), m=1,2 ..., M is divided into
There is certain short time frame for overlapping ratio, frame length can be from several milliseconds to tens milliseconds, then to each passage in M passage
Each frame carry out Short Time Fourier Transform, obtain Ym(ω, i), wherein i represent the i-th frame, then construct
Y (ω, i)=[Y1(ω,i) Y2(ω,i) … YM(ω,i)]T；
S3：With Short Time Fourier Transform, the time-domain signal that sensor array is received is resolved into subband signal；Assuming that phase
Spacing between adjacent two microphones is δ, because hidden audio technology is used to pick up closely sound source, therefore assumes exist in acoustic enviroment
Preferable near-field sound source and interference effect are on the microphone array, and distance of the sound source away from each microphone is respectively rs,1,
rs,2..., rs,M, the center of array is defined as reference point, distance of the sound source away from reference point is rs, incidence angle is θs, then sound source away from
The distance of m-th of microphone can be expressed as
Wherein,
Discrete time k is located at, the signal that sound source is sent is x (k), if ignoring the absorption loss in communication process, m-th
Also there is the amplitude fading being inversely proportional with distance in the signal that microphone is picked up, can relative to sound-source signal only phase delay
It is expressed as：
Wherein, xm(k) sound-source signal that m-th of microphone is picked up, v are representedm(k) represent what m-th of microphone was picked up
Noise signal；τmRepresent time delay of m-th of microphone compared to sound source；
Because the wave surface of the preferable sound source near field is spherical, τmIt is represented by：
Wherein c represents the velocity of sound in air,
Formula (1) is changed into a frequency domain：
Wherein,Wave number is represented, the π f of ω=2 represent angular frequency；F represents temporal frequency,Represent imaginary number
Unit, Ym(ω)、Xm(ω)、Vm(ω) represents y respectivelym(k)、xm(k)、vm(k) Fourier transformation.
According to parameters such as the structure of array, array element number, the positions of sound source, the steering vector that construction length is M：Superscript T represents vectorial transposition computing, makes r=rs, θ=θs, then,The signal that M microphone is picked up is with vector representation：
Y (ω)=[Y1(ω) Y2(ω) … YM(ω)]T
=ds(ω,rs,θs)X(ω)+v(ω),
Y (ω) passes through ARRAY PROCESSING, and obtained output signal is:
Wherein, Z (ω) is sound-source signal X (ω) estimation；Subscript H represents conjugate transposition computing, h (ω)=[H1(ω)
H2(ω) … HM(ω)]TRepresent wave filter of the microphone array to the weighing vector of input signal, that is, microphone array
Coefficient.
Before hidden acoustic filter is derived, the several important indicators for weighing Stealth Fighter are first introduced.
The composition relevant with sound-source signal is in array output：hH(ω)ds(ω,rs,θs)X(ω).Therefore, array is to sound
The response of source signal is：
H(ω,rs,θs)=hH(ω)ds(ω,rs,θs)。
Response of the array to sound-source signal has three variables：ω,rs,θs.Fixed two of which variable, array is believed sound source
Number response generate three indexs with another variable change：1) fixed ω, θs, array is to the response of sound-source signal with rs
The index of change is referred to as hidden acoustic attenuation function；2) fixed rs,θsThe index that response of the array to sound-source signal changes with ω is referred to as hidden
Acoustic frequency receptance function；3) fixed ω, rs, array is to the response of sound-source signal with θsThe index of change is referred to as beam pattern.
Hidden acoustic attenuation function
Hidden acoustic attenuation function describes gain of the array to single-frequency sound-source signal at different distance, its mathematical definition
For：
H(rs)=hH(ω)ds(rs)。
Hidden acoustic frequency receptance function
Hidden acoustic frequency receptance function describes gain of the array to broadband signal, and its mathematical definition is：
H (ω)=hH(ω)ds(ω)。
Beam pattern
Beam pattern describes susceptibility of the array to different directions incoming signal, and its mathematical definition is：
H(θs)=hH(ω)ds(θs)
Signal interference ratio and signal interference ratio gain
Signal to noise ratio weighs the relative size of sound-source signal and noise signal.By contrasting input and output signal to noise ratio, Ke Yiheng
Measure the performance of Beam-former.The present invention concerns interference noise.The input signal interference ratio of array is defined as：
Wherein, X0(ω) is the sound-source signal of reference position pickup,V0(ω) is reference position
The interference signal of pickup,rnRepresent interference noise to the distance of array center.φX0(ω)、φV0
(ω) represents X respectively0(ω),V0The power of (ω).
The output signal interference ratio of array is expressed as：
Wherein θnRepresent direction, the φ of interference noiseX(ω)、φV(ω)X (ω), V (ω) power are represented respectively.
Therefore, signal interference ratio gain is:
S4：On the subband that frequency is ω, the hidden acoustic filter based on MVDR is constructed；Unattenuated closely sound-source signal is same
When minimize array output end residual noise variance just obtained the hidden acoustic filters of MVDR, mathematic(al) representation is as follows：
Formula (2) can be solved using method of Lagrange multipliers, is obtained
Assuming that noise is isotropic noise, (m, n) individual element of its normalized correlation matrix can be written as
Form:
Wherein,There are two kinds of extreme situations：If 1) ω τ0It is very big, i.e., high frequency or greatly
Under spacer conditions, the noise signal that two sensors are received is close to uncorrelated, and isotropic noise is close to space white noise；2)
If ω τ0Very small, i.e., under low frequency or small spacer conditions, the noise signal that two sensors are received is just close relevant,
Isotropic noise points of proximity source noise；
Hidden acoustic filter based on MVDR has the computing of a matrix inversion, when microphone array element number is more, matrix
Morbid state occurs, extremely unstable situation occurs in wave filter, the occurrence of in order to avoid inverting unstable, using tradition
Way：Diagonal loading technique, above-mentioned matrix is added with pair of horns matrix, and the matrix after diagonal loading can be expressed as [Γdn
(ω)+ε Ι], wherein Ι is the unit matrix that size is M, and ε is loading coefficient, and now the hidden acoustic filter coefficient based on MVDR is：
S5：On subband, the signal of the i-th frame is handled using hidden acoustic filter h (ω)
S6：Using inverse Fourier transform in short-term and overlap-add method, to Z, (ω, i) enters line translation, so as to obtain wave beam shape
Time-domain signal Z (k) after.
In order to show the effect of the present invention, provide following specific example to verify the correct of the algorithm that this patent is proposed
Property.Wherein, hidden acoustic attenuation function, hidden acoustic frequency receptance function and beam pattern are MATLAB simulation results, and signal interference ratio is in northwest
Polytechnical university's intelligent acoustic and the experimental result that sound darkroom is tested that totally disappeared for facing border communication speech research center.Experiment condition
Set as follows：
Microphone array even structure linear array, microphone number M=8, array element spacing δ=1.1cm.
Sound source position：θs=0 °, rs=5cm.
Fig. 4 (a) is the hidden acoustic attenuation curve map of the hidden method for acoustic based on 8 array element MVDR.It can be seen that：Point source signal exists
Decay quickly within 20cm, and it is faster apart from nearlyer decay；More than 20cm, attenuation is almost inversely proportional with distance.Illustrate to work as sound
During the distance no more than 20cm of source, the hidden method for acoustic based on 8 array element MVDR can at a distance be disturbed with hidden sound.Explanation：Hidden sound radius r0
≈ 20cm, when sound source is apart from rs＜ r0When, the hidden method for acoustic based on 8 array element MVDR can be with hidden sound r0Outer interference.
Fig. 4 (b) is the hidden acoustic frequency response curve of the hidden method for acoustic based on 8 array element MVDR.It can be seen that：2000Hz's
Putting response of the source signal at 5cm, 10cm, 30cm, 50cm is about successively：0dB、19dB、33dB、38dB.Explanation：Sound source distance
During for 5cm, the hidden method for acoustic based on 8 array element MVDR can be with the interference at hidden sound end-on direction 10cm, 30cm, 50cm.
Fig. 5 is the beam pattern for the hidden method for acoustic based on 8 array element MVDR that frequency is 1300Hz, 3300Hz.It can be seen that：
On different frequency, sound source distance be 5cm when, the hidden method for acoustic based on 8 array element MVDR can with hidden sound any direction 10cm, 30cm,
Interference at 50cm, the hidden sound effective value to the interference of non-end-on direction is more preferable.
Fig. 6 is that the hidden method for acoustic based on 8 array element MVDR is done to the letter of the interference of end-on direction 0cm sound source and end-on direction
Than figure.It can be seen that：When interference is at 50cm, it is right in about 10dB, Fig. 6 (b) that the signal interference ratio that voice is disturbed in Fig. 6 (a) is improved
The signal interference ratio of white Gaussian noise interference is improved in about 6dB, Fig. 6 (c) improves about 7dB to the signal interference ratio of chirp waveform,
About 7dB is improved to the signal interference ratio of mono-tone interference in Fig. 6 (d).Explanation：Hidden method for acoustic based on 8 array element MVDR has hidden well
Sound effective value.
Embodiment described above is only that the preferred embodiment of the present invention is described, not to the design of the present invention
It is defined with scope, on the premise of design concept of the present invention is not departed from, ordinary skill technical staff is to this hair in this area
The all variations and modifications that bright technical scheme is made all should fall into protection scope of the present invention, claimed skill of the invention
Art content, is all documented in technical requirements book.
Claims (10)
1. the hidden method for acoustic based on MVDR, it is characterised in that：This method is first with Short Time Fourier Transform, by sensor array
The time-domain signal received resolves into subband signal, the hidden acoustic filter based on MVDR is constructed on each subband so that from hidden
Throw acoustical signal of sound source within is undamped by hidden acoustic filter, obtains estimating signal eventually through anti-STFT.
2. the hidden method for acoustic according to claim 1 based on MVDR, it is characterised in that：This method comprises the following steps：
S1：According to parameters such as the structure of array, array element number, the positions of sound source, steering vector is constructed
<mrow>
<mi>d</mi>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>,</mo>
<msub>
<mi>r</mi>
<mi>s</mi>
</msub>
<mo>,</mo>
<mi>&theta;</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msup>
<mfenced open = "[" close = "]">
<mtable>
<mtr>
<mtd>
<mfrac>
<msup>
<mi>e</mi>
<mrow>
<mo>-</mo>
<msub>
<mi>jkr</mi>
<mrow>
<mi>s</mi>
<mo>,</mo>
<mn>1</mn>
</mrow>
</msub>
</mrow>
</msup>
<msub>
<mi>r</mi>
<mrow>
<mi>s</mi>
<mo>,</mo>
<mn>1</mn>
</mrow>
</msub>
</mfrac>
</mtd>
<mtd>
<mfrac>
<msup>
<mi>e</mi>
<mrow>
<mo>-</mo>
<msub>
<mi>jkr</mi>
<mrow>
<mi>s</mi>
<mo>,</mo>
<mn>2</mn>
</mrow>
</msub>
</mrow>
</msup>
<msub>
<mi>r</mi>
<mrow>
<mi>s</mi>
<mo>,</mo>
<mn>2</mn>
</mrow>
</msub>
</mfrac>
</mtd>
<mtd>
<mn>...</mn>
</mtd>
<mtd>
<mfrac>
<msup>
<mi>e</mi>
<mrow>
<mo>-</mo>
<msub>
<mi>jkr</mi>
<mrow>
<mi>s</mi>
<mo>,</mo>
<mi>M</mi>
</mrow>
</msub>
</mrow>
</msup>
<msub>
<mi>r</mi>
<mrow>
<mi>s</mi>
<mo>,</mo>
<mi>M</mi>
</mrow>
</msub>
</mfrac>
</mtd>
</mtr>
</mtable>
</mfenced>
<mi>T</mi>
</msup>
</mrow>
S2：The signal y that sensor in microphone array is receivedm(k)=xm(k)+vm(k), m=1,2 ..., M, which is divided into, one
Surely overlap ratio short time frame, frame length can from several milliseconds to tens milliseconds, then in M passage each passage it is every
One frame carries out Short Time Fourier Transform, obtains Ym(ω, i), wherein i represent the i-th frame, then construct
Y (ω, i)=[Y1(ω,i) Y2(ω,i) … YM(ω,i)]T；
S3：With Short Time Fourier Transform, the time-domain signal that sensor array is received is resolved into subband signal；
S4：On the subband that frequency is ω, the hidden acoustic filter based on MVDR is constructed
<mrow>
<msub>
<mi>h</mi>
<mrow>
<mi>M</mi>
<mi>V</mi>
<mi>D</mi>
<mi>R</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<msubsup>
<mi>&Gamma;</mi>
<mi>v</mi>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msubsup>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<mi>d</mi>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>,</mo>
<msub>
<mi>r</mi>
<mi>s</mi>
</msub>
<mo>,</mo>
<msub>
<mi>&theta;</mi>
<mi>s</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msup>
<mi>d</mi>
<mi>H</mi>
</msup>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>,</mo>
<msub>
<mi>r</mi>
<mi>s</mi>
</msub>
<mo>,</mo>
<msub>
<mi>&theta;</mi>
<mi>s</mi>
</msub>
<mo>)</mo>
</mrow>
<msubsup>
<mi>&Gamma;</mi>
<mi>v</mi>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msubsup>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<mi>d</mi>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>,</mo>
<msub>
<mi>r</mi>
<mi>s</mi>
</msub>
<mo>,</mo>
<msub>
<mi>&theta;</mi>
<mi>s</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>;</mo>
</mrow>
S5：On subband, the signal of the i-th frame is handled using hidden acoustic filter h (ω)
<mfenced open = "" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<mi>Z</mi>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>,</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msup>
<mi>h</mi>
<mi>H</mi>
</msup>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<mi>y</mi>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>,</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>m</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>M</mi>
</munderover>
<msubsup>
<mi>H</mi>
<mi>m</mi>
<mo>*</mo>
</msubsup>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<msub>
<mi>Y</mi>
<mi>m</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>,</mo>
<mi>i</mi>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
S6：Using inverse Fourier transform in short-term and overlap-add method, to Z, (ω, i) enters line translation, so as to obtain after Wave beam forming
Time-domain signal Z (k).
3. the hidden method for acoustic according to claim 1 based on MVDR, it is characterised in that：With Short Time Fourier Transform, it will pass
Sensor array received to time-domain signal resolve into comprising the following steps that for subband signal：
Assuming that the spacing between two neighboring microphone is δ, because hidden audio technology is used to pick up closely sound source, therefore acoustics is assumed
There is preferable near-field sound source and interference effect in environment on the microphone array, distance of the sound source away from each microphone is respectively
rs,1, rs,2..., rs,M, the center of array is defined as reference point, distance of the sound source away from reference point is rs, incidence angle is θs, then sound
Distance of the source away from m-th of microphone can be expressed as
<mrow>
<msub>
<mi>r</mi>
<mrow>
<mi>s</mi>
<mo>,</mo>
<mi>m</mi>
</mrow>
</msub>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<msqrt>
<mrow>
<msubsup>
<mi>r</mi>
<mi>s</mi>
<mn>2</mn>
</msubsup>
<mo>-</mo>
<mn>2</mn>
<msub>
<mi>r</mi>
<mi>s</mi>
</msub>
<msub>
<mi>d</mi>
<mi>m</mi>
</msub>
<mi>c</mi>
<mi>o</mi>
<mi>s</mi>
<mi>&theta;</mi>
<mo>+</mo>
<msubsup>
<mi>d</mi>
<mi>m</mi>
<mn>2</mn>
</msubsup>
</mrow>
</msqrt>
<mo>,</mo>
</mrow>
</mtd>
<mtd>
<mrow>
<mi>m</mi>
<mo>&le;</mo>
<mfrac>
<mrow>
<mi>M</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
<mn>2</mn>
</mfrac>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msqrt>
<mrow>
<msubsup>
<mi>r</mi>
<mi>s</mi>
<mn>2</mn>
</msubsup>
<mo>+</mo>
<mn>2</mn>
<msub>
<mi>r</mi>
<mi>s</mi>
</msub>
<msub>
<mi>d</mi>
<mi>m</mi>
</msub>
<mi>cos</mi>
<mi>&theta;</mi>
<mo>+</mo>
<msubsup>
<mi>d</mi>
<mi>m</mi>
<mn>2</mn>
</msubsup>
</mrow>
</msqrt>
<mo>,</mo>
</mrow>
</mtd>
<mtd>
<mrow>
<mi>m</mi>
<mo>></mo>
<mfrac>
<mrow>
<mi>M</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
<mn>2</mn>
</mfrac>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
</mrow>
Wherein,
Discrete time k is located at, the signal that sound source is sent is x (k), if ignoring the absorption loss in communication process, m-th of Mike
Also there is the amplitude fading being inversely proportional with distance, can represent in the signal that wind is picked up relative to sound-source signal only phase delay
For：
<mrow>
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>y</mi>
<mi>m</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msub>
<mi>x</mi>
<mi>m</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<msub>
<mi>v</mi>
<mi>m</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<msub>
<mi>r</mi>
<mrow>
<mi>s</mi>
<mo>,</mo>
<mi>m</mi>
</mrow>
</msub>
</mfrac>
<mi>x</mi>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>-</mo>
<msub>
<mi>&tau;</mi>
<mi>m</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>+</mo>
<msub>
<mi>v</mi>
<mi>m</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>k</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<mi>m</mi>
<mo>=</mo>
<mn>1</mn>
<mo>,</mo>
<mn>2</mn>
<mo>,</mo>
<mn>...</mn>
<mo>,</mo>
<mi>M</mi>
<mo>,</mo>
</mrow>
</mtd>
</mtr>
</mtable>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, xm(k) sound-source signal that m-th of microphone is picked up, v are representedm(k) noise that m-th of microphone is picked up is represented
Signal；τmRepresent time delay of m-th of microphone compared to sound source；
Because the wave surface of the preferable sound source near field is spherical, τmIt is represented by：
<mrow>
<msub>
<mi>&tau;</mi>
<mi>m</mi>
</msub>
<mo>=</mo>
<mfrac>
<msub>
<mi>r</mi>
<mrow>
<mi>s</mi>
<mo>,</mo>
<mi>m</mi>
</mrow>
</msub>
<mi>c</mi>
</mfrac>
<mo>,</mo>
<mi>m</mi>
<mo>=</mo>
<mn>1</mn>
<mo>,</mo>
<mn>2</mn>
<mo>,</mo>
<mo>...</mo>
<mo>,</mo>
<mi>M</mi>
<mo>,</mo>
</mrow>
Wherein c represents the velocity of sound in air,
Formula (1) is changed into a frequency domain：
<mrow>
<msub>
<mi>Y</mi>
<mi>m</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<msup>
<mi>e</mi>
<mrow>
<mo>-</mo>
<msub>
<mi>jkr</mi>
<mrow>
<mi>s</mi>
<mo>,</mo>
<mi>m</mi>
</mrow>
</msub>
</mrow>
</msup>
<msub>
<mi>r</mi>
<mrow>
<mi>s</mi>
<mo>,</mo>
<mi>m</mi>
</mrow>
</msub>
</mfrac>
<mi>X</mi>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<msub>
<mi>V</mi>
<mi>m</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
<mi>m</mi>
<mo>=</mo>
<mn>1</mn>
<mo>,</mo>
<mn>2</mn>
<mo>,</mo>
<mo>...</mo>
<mo>,</mo>
<mi>M</mi>
<mo>,</mo>
</mrow>
Wherein,Wave number is represented, the π f of ω=2 represent angular frequency；F represents temporal frequency,Represent imaginary number list
Position, Ym(ω)、Xm(ω)、Vm(ω) represents y respectivelym(k)、xm(k)、vm(k) Fourier transformation.
4. the hidden method for acoustic according to claim 3 based on MVDR, it is characterised in that：According to the structure of array, array number
The parameters such as mesh, the position of sound source, the steering vector that construction length is M：
Superscript T represents vectorial transposition computing, makes r=rs, θ=θs, that
,The signal that M microphone is picked up is with vector representation：
Y (ω)=[Y1(ω) Y2(ω) … YM(ω)]T
=ds(ω,rs,θs)X(ω)+v(ω),
Y (ω) passes through ARRAY PROCESSING, and obtained output signal is:
<mfenced open = "" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<mi>Z</mi>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>m</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>M</mi>
</munderover>
<msubsup>
<mi>H</mi>
<mi>m</mi>
<mo>*</mo>
</msubsup>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<msub>
<mi>Y</mi>
<mi>m</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mo>=</mo>
<msup>
<mi>h</mi>
<mi>H</mi>
</msup>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<msub>
<mi>d</mi>
<mi>s</mi>
</msub>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>,</mo>
<msub>
<mi>r</mi>
<mi>s</mi>
</msub>
<mo>,</mo>
<msub>
<mi>&theta;</mi>
<mi>s</mi>
</msub>
<mo>)</mo>
</mrow>
<mi>X</mi>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<msup>
<mi>h</mi>
<mi>H</mi>
</msup>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<mi>v</mi>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<mo>,</mo>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
Wherein, Z (ω) is sound-source signal X (ω) estimation；Subscript H represents conjugate transposition computing, h (ω)=[H1(ω) H2
(ω) … HM(ω)]TRepresent wave filter system of the microphone array to the weighing vector of input signal, that is, microphone array
Number.
5. the hidden method for acoustic according to claim 1 based on MVDR, it is characterised in that：In the subband that frequency is ω in S3
On, hidden acoustic filter method of the construction based on MVDR is as follows：Unattenuated closely sound-source signal minimizes array output end simultaneously
The variance of residual noise has just obtained the hidden acoustic filters of MVDR, and mathematic(al) representation is as follows：
<mrow>
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>h</mi>
<mrow>
<mi>M</mi>
<mi>V</mi>
<mi>D</mi>
<mi>R</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mi>arg</mi>
<munder>
<mi>min</mi>
<mrow>
<mi>h</mi>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
</mrow>
</munder>
<msup>
<mi>h</mi>
<mi>H</mi>
</msup>
<msub>
<mi>R</mi>
<mrow>
<mi>v</mi>
<mi>v</mi>
</mrow>
</msub>
<mi>h</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<mi>s</mi>
<mi>u</mi>
<mi>b</mi>
<mi>j</mi>
<mi>e</mi>
<mi>c</mi>
<mi>t</mi>
<mi> </mi>
<mi>t</mi>
<mi>o</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<msup>
<mi>h</mi>
<mi>H</mi>
</msup>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<mi>d</mi>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>,</mo>
<msub>
<mi>r</mi>
<mi>s</mi>
</msub>
<mo>,</mo>
<msub>
<mi>&theta;</mi>
<mi>s</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mn>1</mn>
</mrow>
</mtd>
</mtr>
</mtable>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
</mrow>
Formula (2) can be solved using method of Lagrange multipliers, is obtained
<mrow>
<msub>
<mi>h</mi>
<mrow>
<mi>M</mi>
<mi>V</mi>
<mi>D</mi>
<mi>R</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<msubsup>
<mi>&Gamma;</mi>
<mi>v</mi>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msubsup>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<mi>d</mi>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>,</mo>
<msub>
<mi>r</mi>
<mi>s</mi>
</msub>
<mo>,</mo>
<msub>
<mi>&theta;</mi>
<mi>s</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msup>
<mi>d</mi>
<mi>H</mi>
</msup>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>,</mo>
<msub>
<mi>r</mi>
<mi>s</mi>
</msub>
<mo>,</mo>
<msub>
<mi>&theta;</mi>
<mi>s</mi>
</msub>
<mo>)</mo>
</mrow>
<msubsup>
<mi>&Gamma;</mi>
<mi>v</mi>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msubsup>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<mi>d</mi>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>,</mo>
<msub>
<mi>r</mi>
<mi>s</mi>
</msub>
<mo>,</mo>
<msub>
<mi>&theta;</mi>
<mi>s</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>.</mo>
</mrow>
6. the hidden method for acoustic according to claim 5 based on MVDR, it is characterised in that：Assuming that noise is made an uproar for isotropism
Sound, (m, n) individual element of its normalized correlation matrix can be written as form:
Wherein,There are two kinds of extreme situations：1) such as s fruit ω τ0It is very big, i.e., high frequency or spacing greatly
In the case of, the noise signal that two sensors are received is close to uncorrelated, and isotropic noise is close to space white noise；If 2)
ωτ0It is very small, i.e., under low frequency or small spacer conditions, the noise signal that two sensors are received just close to relevant, respectively to
Same sex noise points of proximity source noise；
Hidden acoustic filter based on MVDR has the computing of a matrix inversion, and when microphone array element number is more, matrix can go out
Extremely unstable situation occurs in existing morbid state, wave filter, the occurrence of in order to avoid inverting unstable, is done using traditional
Method：Diagonal loading technique, above-mentioned matrix is added with pair of horns matrix, and the matrix after diagonal loading can be expressed as [Γdn(ω)
+ ε Ι], wherein Ι is the unit matrix that size is M, and ε is loading coefficient, and now the hidden acoustic filter coefficient based on MVDR is：
<mrow>
<msub>
<mi>h</mi>
<mrow>
<mi>M</mi>
<mi>V</mi>
<mi>D</mi>
<mi>R</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<msup>
<mrow>
<mo>&lsqb;</mo>
<msub>
<mi>&Gamma;</mi>
<mrow>
<mi>d</mi>
<mi>n</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<mi>&epsiv;</mi>
<mi>I</mi>
<mo>&rsqb;</mo>
</mrow>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msup>
<mi>d</mi>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>,</mo>
<mo>-</mo>
<msub>
<mi>r</mi>
<mi>s</mi>
</msub>
<mo>,</mo>
<msub>
<mi>&theta;</mi>
<mi>s</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msup>
<mi>d</mi>
<mi>H</mi>
</msup>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>,</mo>
<msub>
<mi>r</mi>
<mi>s</mi>
</msub>
<mo>,</mo>
<msub>
<mi>&theta;</mi>
<mi>s</mi>
</msub>
<mo>)</mo>
</mrow>
<msup>
<mrow>
<mo>&lsqb;</mo>
<msub>
<mi>&Gamma;</mi>
<mrow>
<mi>d</mi>
<mi>n</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<mo>+</mo>
<mi>&epsiv;</mi>
<mi>I</mi>
<mo>&rsqb;</mo>
</mrow>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msup>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>)</mo>
</mrow>
<mi>d</mi>
<mrow>
<mo>(</mo>
<mi>&omega;</mi>
<mo>,</mo>
<msub>
<mi>r</mi>
<mi>s</mi>
</msub>
<mo>,</mo>
<msub>
<mi>&theta;</mi>
<mi>s</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
<mo>.</mo>
</mrow>
7. the as claimed in claim 1 hidden method for acoustic for being based on MVDR, it is characterised in that the microphone array include but
It is not limited to homogenous linear microphone array, non-homogeneous linear microphone array and circular microphone array.
8. the hidden method for acoustic as claimed in claim 1 based on MVDR, it is characterised in that the microphone array is small-sized Mike
One kind in wind array and large-scale microphone array.
9. the hidden method for acoustic as claimed in claim 1 based on MVDR, it is characterised in that the audio signal processing method is used for
Handle narrow band signal or broadband signal.
10. the hidden method for acoustic as claimed in claim 1 based on MVDR, it is characterised in that the hidden sound of the microphone array half
Footpath increases with the increase of microphone number and array element spacing.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710163190.2A CN107170462A (en) | 2017-03-19 | 2017-03-19 | Hidden method for acoustic based on MVDR |
| CN201810221809.5A CN108597532A (en) | 2017-03-19 | 2018-03-18 | Hidden method for acoustic based on MVDR |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710163190.2A CN107170462A (en) | 2017-03-19 | 2017-03-19 | Hidden method for acoustic based on MVDR |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107170462A true CN107170462A (en) | 2017-09-15 |
Family
ID=59848862
Family Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710163190.2A Withdrawn CN107170462A (en) | 2017-03-19 | 2017-03-19 | Hidden method for acoustic based on MVDR |
| CN201810221809.5A Pending CN108597532A (en) | 2017-03-19 | 2018-03-18 | Hidden method for acoustic based on MVDR |
Family Applications After (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810221809.5A Pending CN108597532A (en) | 2017-03-19 | 2018-03-18 | Hidden method for acoustic based on MVDR |
Country Status (1)
| Country | Link |
|---|---|
| CN (2) | CN107170462A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110018465A (en) * | 2018-01-09 | 2019-07-16 | 中国科学院声学研究所 | One kind being based on the pretreated MVDR Beamforming Method of all phase |
| WO2019205797A1 (en) * | 2018-04-27 | 2019-10-31 | 深圳市沃特沃德股份有限公司 | Noise processing method, apparatus and device |
| CN112420068A (en) * | 2020-10-23 | 2021-02-26 | 四川长虹电器股份有限公司 | Quick self-adaptive beam forming method based on Mel frequency scale frequency division |
| CN116013239A (en) * | 2022-12-07 | 2023-04-25 | 广州声博士声学技术有限公司 | Active noise reduction algorithm and device for air duct |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110148420A (en) * | 2019-06-30 | 2019-08-20 | 桂林电子科技大学 | A kind of audio recognition method suitable under noise circumstance |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103856866B (en) * | 2012-12-04 | 2019-11-05 | 西北工业大学 | Low noise differential microphone array |
| CN104464739B (en) * | 2013-09-18 | 2017-08-11 | 华为技术有限公司 | Acoustic signal processing method and device, Difference Beam forming method and device |
| CN103491397B (en) * | 2013-09-25 | 2017-04-26 | 歌尔股份有限公司 | Method and system for achieving self-adaptive surround sound |
| EP2916320A1 (en) * | 2014-03-07 | 2015-09-09 | Oticon A/s | Multi-microphone method for estimation of target and noise spectral variances |
| DK2916321T3 (en) * | 2014-03-07 | 2018-01-15 | Oticon As | Processing a noisy audio signal to estimate target and noise spectral variations |
| EP2928210A1 (en) * | 2014-04-03 | 2015-10-07 | Oticon A/s | A binaural hearing assistance system comprising binaural noise reduction |
-
2017
- 2017-03-19 CN CN201710163190.2A patent/CN107170462A/en not_active Withdrawn
-
2018
- 2018-03-18 CN CN201810221809.5A patent/CN108597532A/en active Pending
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110018465A (en) * | 2018-01-09 | 2019-07-16 | 中国科学院声学研究所 | One kind being based on the pretreated MVDR Beamforming Method of all phase |
| CN110018465B (en) * | 2018-01-09 | 2020-11-06 | 中国科学院声学研究所 | MVDR beam forming method based on full-phase preprocessing |
| WO2019205797A1 (en) * | 2018-04-27 | 2019-10-31 | 深圳市沃特沃德股份有限公司 | Noise processing method, apparatus and device |
| CN112420068A (en) * | 2020-10-23 | 2021-02-26 | 四川长虹电器股份有限公司 | Quick self-adaptive beam forming method based on Mel frequency scale frequency division |
| CN112420068B (en) * | 2020-10-23 | 2022-05-03 | 四川长虹电器股份有限公司 | Quick self-adaptive beam forming method based on Mel frequency scale frequency division |
| CN116013239A (en) * | 2022-12-07 | 2023-04-25 | 广州声博士声学技术有限公司 | Active noise reduction algorithm and device for air duct |
| CN116013239B (en) * | 2022-12-07 | 2023-11-17 | 广州声博士声学技术有限公司 | Active noise reduction algorithm and device for air duct |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108597532A (en) | 2018-09-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107170462A (en) | Hidden method for acoustic based on MVDR | |
| KR101340215B1 (en) | Systems, methods, apparatus, and computer-readable media for dereverberation of multichannel signal | |
| CN107248413A (en) | Hidden method for acoustic based on Difference Beam formation | |
| CN106782590B (en) | Microphone array beam forming method based on reverberation environment | |
| Khalil et al. | Microphone array for sound pickup in teleconference systems | |
| Zhao et al. | Design of robust differential microphone arrays | |
| US8351554B2 (en) | Signal extraction | |
| CN105869651A (en) | Two-channel beam forming speech enhancement method based on noise mixed coherence | |
| Peled et al. | Linearly-constrained minimum-variance method for spherical microphone arrays based on plane-wave decomposition of the sound field | |
| Ryan et al. | Application of near-field optimum microphone arrays to hands-free mobile telephony | |
| Jeub et al. | Binaural dereverberation based on a dual-channel wiener filter with optimized noise field coherence | |
| Shabtai | Optimization of the directivity in binaural sound reproduction beamforming | |
| Jarrett et al. | Dereverberation performance of rigid and open spherical microphone arrays: Theory & simulation | |
| Geng et al. | A speech enhancement method based on the combination of microphone array and parabolic reflector | |
| Kallinger et al. | A spatial filtering approach for directional audio coding | |
| Kowalczyk | Raking early reflection signals for late reverberation and noise reduction | |
| Li et al. | A two-microphone noise reduction method in highly non-stationary multiple-noise-source environments | |
| Leng et al. | On speech enhancement using microphone arrays in the presence of co-directional interference | |
| CN112017684B (en) | Closed space reverberation elimination method based on microphone array | |
| Ma et al. | A time-domain nearfield frequency-invariant beamforming method | |
| Habets | Towards multi-microphone speech dereverberation using spectral enhancement and statistical reverberation models | |
| Kwan et al. | Speech separation algorithms for multiple speaker environments | |
| Lotter et al. | A stereo input-output superdirective beamformer for dual channel noise reduction. | |
| Kowalczyk | Multichannel Wiener filter with early reflection raking for automatic speech recognition in presence of reverberation | |
| Yermeche | Soft-Constrained Subband Beamforming for Speech Enhancement |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20170915 |
|
| WW01 | Invention patent application withdrawn after publication |