CN108600894A - A kind of earphone adaptive active noise control system and method - Google Patents
A kind of earphone adaptive active noise control system and method Download PDFInfo
- Publication number
- CN108600894A CN108600894A CN201810758793.1A CN201810758793A CN108600894A CN 108600894 A CN108600894 A CN 108600894A CN 201810758793 A CN201810758793 A CN 201810758793A CN 108600894 A CN108600894 A CN 108600894A
- Authority
- CN
- China
- Prior art keywords
- signal
- noise
- earphone
- analog
- matrix
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
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
- H04R1/10—Earpieces; Attachments therefor ; Earphones; Monophonic headphones
- H04R1/1083—Reduction of ambient noise
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04R—LOUDSPEAKERS, MICROPHONES, GRAMOPHONE PICK-UPS OR LIKE ACOUSTIC ELECTROMECHANICAL TRANSDUCERS; DEAF-AID SETS; PUBLIC ADDRESS SYSTEMS
- H04R2201/00—Details of transducers, loudspeakers or microphones covered by H04R1/00 but not provided for in any of its subgroups
- H04R2201/10—Details of earpieces, attachments therefor, earphones or monophonic headphones covered by H04R1/10 but not provided for in any of its subgroups
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Abstract
The invention discloses a kind of earphone adaptive active noise control system and methods, the system includes at least, one reference microphone for acquiring acquisition noise signal, one error microphone for being used for Acquisition Error noise signal, reference microphone is set to outside earphone, and error microphone is set to inside earphone；One loudspeaker unit, for playing secondary acoustical signal；Two analog-digital converter AD are respectively used to that noise signal will be referred to and error noise signal are converted to corresponding digital signal；One digital analog converter DA, for filtered output signals to be converted to analog signal and sub-loudspeaker of feeding；One processor, for updating filter factor and generating the input signal of sub-loudspeaker.The invention can solve the problems such as common adaptive algorithm real-time tracking scarce capacity and excessive computation delay, all there is preferable adaptive performance to different environment and different earphone wearing modes, improve the constringency performance and tracking performance of earphone self-adaptive active noise canceling.
Description
Technical field
The present invention relates to noises to eliminate field, and in particular to a kind of earphone adaptive active noise control system and method.
Background technology
The noise moment affects people’s lives, such as in cabin, subway and working environment, people all the time not by
Presence to the invasion of noise, noise makes the psychology of people become irritated and uneasy, how effectively to control noise, reduces it to people
Body and mind harm be always scientific research one of important goal.Scientific investigations showed that being exposed to strong noise environment for a long time
Under, it is easy to cause the permanent loss to all or part of hearing frequencies sensibility, or even cause deafness.Reducing noise mainly has master
Moving noise controls and two methods of passive noise control.Passive noise control mainly utilizes the reflection isolation sound of special material
Propagate or using material the decaying of porosity and viscosity by acoustic energy, passive noise control method compares high-frequency noise
Effectively.Active noise reduction earphone is to generate that one identical as noise amplitude and phase inside earphone using active noise control technique
Opposite acoustical signal so that the sound wave of secondary sound wave and original sound field is cancelled out each other at target, to reach the mesh of control noise
, active noise control technique is preferable to low-frequency noise effect.
A crucial problem is how to update filter coefficient in Active noise control.FxLMS(Filtered-
Xleast mean square) algorithm is the most popular adaptive algorithm for Active noise control, but the algorithm by
It is realized in time domain, computation complexity is very high, realizes and is restricted in the limited system of resource, in addition the algorithm substantially belongs to
In LMS class algorithms, convergence rate depends on the conditional number of input signal autocorrelation matrix.Frequency domain adaptive filtering algorithm due to
It has been widely used in echo cancelltion, Wave beam forming and has uttered long and high-pitched sounds with very low computation complexity and good constringency performance
The occasions such as inhibition.Also document has attempted to do Active noise control using Adaptive Algorithm.Such as：Document
《Q.Shen and A.S.Spanias,“Time and frequency domain xblock LMS algorithms for
single channel active noise control,”in Proceedings ofthe 2ndInternational
Congress ofRecent Developments inAir-and Structure-Borne SoundVibration,1992,
pp.353–360》It proposes frequency domain algorithm being applied to Active noise control, document first《Das,D.P.,G.Panda,and
S.M.Kuo."New block filtered-X LMS algorithms for active noisecontrol
systems."IET Signal Processing 1.2(2007):73-81》It is adaptive to be further discussed more thorough frequency domain
Answer active control technology.But these methods are handled by being then based on block, data buffer storage and processing introduce delay, the delay pair
The decline of performance will not be caused only algorithm to be brought to be delayed for the application of System Discrimination, but for Active noise control system
The influence of system is but very big, because it is non-causal that the delay, which may directly result in active system, especially for the application of earphone
In, the distance between feedforward microphone and sub-loudspeaker are very short, thus the delay that algorithm allows is very small.For example, big
In most circumaural earphones, the representative value of feedforward microphone and distances between loudspeakers is d=1 centimetres, it is assumed that sample rate fS
=16000Hz, then to ensure that the algorithm maximum delay that system causality allows is 4.7 sampling periods, and be directly realized by
Frequency domain algorithm would generally introduce the delay in such as 128 sampling periods, thus frequency domain algorithm is directly used for earphone active noise reduction
System is infeasible.In order to solve this problem, document《X.Qiu and C.H.Hansen,“Multidelay
adaptive filters for active noisecontrol,”in Proceedings of the 14th
International Congress on Sound andVibration(ICSV),Cairns,Australia,2007,
pp.724–732》Using no-delay algorithm, but the peak value complexity of this method is very high, because the algorithm is required one
The intensive operations of a series of complex degree such as Fourier transformation are completed in a sampling period, this is all one to most dsp chips
A prodigious challenge.
Another question using Adaptive Algorithm is that the selection of step-length must be in convergence rate and steady output rate
Between compromise.One larger step-length can ensure that algorithm has faster initial convergence speed, and when system changes
When with the ability quickly tracked again, but big step-length is easy to so that with larger imbalance when algorithm stable state, this is
The characteristic not being expected to.In the actual environment, the transmission function of main channel and secondary channel is all time-varying, such as different people
Ear it is of different sizes cause its secondary channel after wearing earphone to be different, the adjustment that wearer also can be frequent in practice
The elasticity of earphone causes the variation of above-mentioned two transmission function.Thus adaptive algorithm must have good real-time tracking
Energy.
Invention content
For deficiencies of the prior art, it is an object of the invention to：A kind of earphone adaptive active is provided to make an uproar
Acoustic control system and method, the problems such as common adaptive algorithm real-time tracking scarce capacity and excessive computation delay can be solved,
All there is preferable adaptive performance to different environment and different earphone wearing modes, improve earphone adaptive active and make an uproar
The constringency performance and tracking performance of acoustic control.
A kind of earphone adaptive active noise control system, the system include at least：
First audio collection device and the second audio collection device, the first audio collection device is mounted on outside earphone, for acquiring
With reference to noise signal；Second audio collection device is mounted on inside earphone, for acquiring the error noise signal after noise reduction；
First analog-digital converter AD and the second analog-digital converter AD, the first analog-digital converter AD and the first audio collection device electricity
Connection, the reference noise signal for acquiring the first audio collection device are converted to corresponding digital signal, the second analog-to-digital conversion
Device AD and the electrical connection of the second audio collection device, the error noise signal for acquiring the second audio collection device is converted to corresponding
Digital signal；
Processor unit, the first analog-digital converter AD are connected with processor unit first information input terminal, and the second modulus turns
Parallel operation AD is connected with the second information input terminal of processor unit；
Digital analog converter DA and sub-loudspeaker, processor unit information output are raised after digital analog converter DA with secondary
Sound device connects, and sub-loudspeaker is mounted in earphone, and digital analog converter DA is used to turn the output filtering signal of processor unit
It is changed to corresponding analog signal and is fed into sub-loudspeaker, sub-loudspeaker is for playing secondary audio signal.
Further, which further includes memory cell, and memory cell and processor unit two-way communication connect
It connects.
Further, which further includes power module, and power module is connected with processor unit feeder ear.
Further, which further includes control logic unit, and control logic unit is for controlling active noise reduction work(
The switch of energy.
A kind of earphone self-adaptive active noise canceling method, includes the following steps：
It is acquired in real time by the first audio collection device and refers to noise signal, and noise will be referred to by the first analog-digital converter and believed
It is transferred to processor unit after number being converted to corresponding digital signal x (n)；By the real-time Acquisition Error noise of the second audio collection device
Signal, and it is transferred to processor after error noise signal is converted to corresponding digital signal e (n) by the second analog-digital converter
Unit；
Filter coefficient vector is calculated by processor unitFiltering signal y (n) is exported with calculating, is adopted in frequency domain
Filter coefficient vector is calculated with Kalman filteringTo digital signal x (n) and time domain filter coefficients in time domain
It carries out convolution and obtains output filtering signal y (n)；Wherein, time domain filter coefficientsIt is filter coefficient vectorInverse Fu
In leaf transformation；
Output filtering signal y (n) is converted to corresponding analog signal and is fed into secondary by digital analog converter and is raised one's voice
Device, sub-loudspeaker is for playing secondary audio signal.
Further, filter coefficient vector is calculated using Kalman filtering in frequency domainThe step of it is as follows：
In frequency domain, the secondary channel with reference to the corresponding digital signal x (n) of noise signal by estimationFiltering obtains
Kalman filtering input signal v (n)；
Nearest 2L point Kalman filtering input signal v (n) sequences formed are done into Fourier transformation and obtain frequency domain vector v
(k), the element of frequency domain vector v (k) is sequentially placed in diagonal line composition diagonal matrix V (k)；
Nearest L point tolerance noise signal e (n) sequences formed are done into Fourier transformation and obtain frequency domain vector E (k)；
Update kalman gain matrix K (k)=P (k) VH(k)[V(k)P(k)VH(k)+2ψee(k)]-1, wherein P (k) is
State error matrix, ψee(k) it is observation noise error diagonal covariance matrix, subscript H represents conjugate transposition operation；
Update frequency domain filter coefficient vectorWherein
It is constraint matrix, ILBe dimension be L × L unit matrix, 0LIt is the null matrix that dimension is L × L, F is Fourier transform matrix, K
(k) it is kalman gain matrix；
Update state error matrix P (k)：Wherein, ψΔΔ(k)
It is process noise diagonal matrix, I2LIt is the unit matrix that dimension is 2L × 2L.
Further, in order to calculate kalman gain matrix K (k), it would be desirable to know ψee(k), heretofore described sight
Survey noise error matrix ψee(k) smooth power spectrum of error noise signal e (n) is used to be calculated.
Further, in order to calculate state error matrix P (k), we also need to calculating process noise covariance matrix ψΔΔ
(k), heretofore described process noise diagonal matrix ψΔΔ(k) diagonal entry ψΔΔ,i(k) computational methods areWherein,It is filter coefficient vectorI-th of element.
Compared with the prior art, the present invention has the following advantages：
A kind of earphone adaptive active noise control system disclosed by the invention and method are filtered in frequency domain using Kalman
Wave algorithm updates filter factor, since Kalman filtering algorithm has good tracking performance, to solve it is common from
The problem of adaptive algorithm ability of tracking deficiency so that different application environments and different earphone wearing modes can be met.Together
When, the filter factor of frequency domain is transformed into time domain, output filtering is directly calculated using convolution in time domain, so as to avoid
In the block delay that frequency-domain calculations introduce.By the control system and method, the constringency performance and tracking performance of calculating are improved.
Description of the drawings
Fig. 1 is the system block diagram of earphone adaptive active noise control system in the embodiment of the present invention；
Fig. 2 is the control schematic diagram of earphone adaptive active noise in the embodiment of the present invention；
Fig. 3 is the filtering flow chart of processor unit in the embodiment of the present invention；
Fig. 4 is the flow chart for using Kalman filtering in the embodiment of the present invention in frequency domain；
Fig. 5 is observation noise spectra calculation flow chart in the embodiment of the present invention.
Reference numeral：
102, the first audio collection device；104, the second audio collection device；106, sub-loudspeaker；108, the first analog-to-digital conversion
Device AD；110, digital analog converter DA；112, the second analog-digital converter AD；114, the secondary channel estimated；116, filter time domain
Coefficient；118, Kalman filtering；120, controller unit；150, memory cell；160, power module；170, control logic list
Member；180, earphone；202, shift unit；204, shift unit；206, shift unit；208, multiplier；210, multiplier；212、
Multiplier；214, multiplier；216, adder；218, adder；220, adder；302, Fourier transform unit；304, block
Germania gain matrix；306, Fourier transform unit；308, multiplier；310, inverse Fourier transform arithmetic element；312, sequence
Latter half zero setting；314, Fourier transformation；316, adder；318, delay unit；320, inverse Fourier transform；402, it takes altogether
Yoke module；404, multiplication unit；406, multiplication unit；408, addition unit；410, multiplication unit；412, delay unit.
Specific implementation mode
The embodiment of technical solution of the present invention is described in detail below in conjunction with attached drawing.Following embodiment is only used for
Clearly illustrate technical scheme of the present invention, therefore be intended only as example, and the protection of the present invention cannot be limited with this
Range.
Embodiment：
Referring to Figures 1 and 2, a kind of earphone adaptive active noise control system, the system include：
First audio collection device 102 and the second audio collection device 104, the first audio collection device are mounted on outside earphone 180,
Noise signal is referred to for acquiring；Second audio collection device is mounted on inside earphone, for acquiring the letter of the error noise after noise reduction
Number；
First analog-digital converter AD108 and the second analog-digital converter AD112, the first analog-digital converter AD and the first audio are adopted
Storage is electrically connected, and the reference noise signal for acquiring the first audio collection device is converted to corresponding digital signal x (n), the
Two analog-digital converter AD and the electrical connection of the second audio collection device, the error noise signal for acquiring the second audio collection device turn
It is changed to corresponding digital signal e (n)；
Processor unit 140, the first analog-digital converter AD are connected with processor unit first information input terminal, the second modulus
Converter AD is connected with the second information input terminal of processor unit；Digital signal x (n) and digital signal e (n) are input to processor
Unit obtains filter time domain coefficient after processor unit analysis116 and output filtering signal y (n)；Processor unit
140 have two parts function, first, utilizing the corresponding digital signal x (n) of reference noise signal number corresponding with error noise signal
Word signal e (n) updates filter coefficient vectorSecond is that calculating filtered output signals y (n).Wherein filter coefficient vectorUpdate obtains frame by frame in frequency domain, and filtered output signals y (n) is obtained in time domain node-by-node algorithm.
Digital analog converter DA110 and sub-loudspeaker 106, processor unit information output after digital analog converter DA and
Sub-loudspeaker connects, and sub-loudspeaker is mounted in earphone, and digital analog converter DA is used to filter the output of processor unit
Signal is converted to corresponding analog signal and is fed into sub-loudspeaker, and sub-loudspeaker is for playing secondary audio signal；It is defeated
Go out filtering signal y (n) and gives the broadcasting of sub-loudspeaker 106 after digital analog converter DA110.When it is implemented, the first audio
Microphone may be used in collector and the second audio collection device.
In the present embodiment, which further includes memory cell 150, memory cell and processor unit two-way
Letter connection.The control system further includes power module 160, and power module is connected with processor unit feeder ear.The control system
Further include control logic unit 170, control logic unit is used to control the switch of active noise reduction function.When it is implemented, described
DSP, ARM or other application specific processor chips may be used in processor unit.The memory cell be used for do storage program and
Variable.The control system further includes installation shell and circuit board, and circuit board installation cavity, second audio are equipped in mounting shell body
Collector, the first analog-digital converter, the second analog-digital converter, processor unit, digital analog converter and sub-loudspeaker are respectively mounted
On circuit boards.The installation housing outer surface is equipped with a mounting base, and the first audio collection device is mounted in mounting base.
With reference to Fig. 2~Fig. 5, a kind of earphone self-adaptive active noise canceling method includes the following steps：
It is acquired in real time by the first audio collection device and refers to noise signal, and noise will be referred to by the first analog-digital converter and believed
It is transferred to processor unit after number being converted to corresponding digital signal x (n)；By the real-time Acquisition Error noise of the second audio collection device
Signal, and it is transferred to processor after error noise signal is converted to corresponding digital signal e (n) by the second analog-digital converter
Unit；
Filter coefficient vector is calculated by processor unitFiltering signal y (n) is exported with calculating, is adopted in frequency domain
Filter coefficient vector is calculated with Kalman filtering 118To digital signal x (n) and time domain filter coefficients in time domainIt carries out convolution and obtains output filtering signal y (n)；Wherein, time domain filter coefficientsIt is filter coefficient vector's
Inverse Fourier transform；
Output filtering signal y (n) is converted to corresponding analog signal and is fed into secondary by digital analog converter and is raised one's voice
Device, sub-loudspeaker is for playing secondary audio signal.
In the present embodiment, filter coefficient vector is calculated using Kalman filtering in frequency domainThe step of it is as follows：
In frequency domain, the secondary channel with reference to the corresponding digital signal x (n) of noise signal by estimationFiltering
Obtain Kalman filtering input signal v (n)；
Nearest 2L point Kalman filtering input signal v (n) sequences formed are done into Fourier transformation and obtain frequency domain vector v
(k), the element of frequency domain vector v (k) is sequentially placed in diagonal line composition diagonal matrix V (k)；
Nearest L point tolerance noise signal e (n) sequences formed are done into Fourier transformation and obtain frequency domain vector E (k)；
Update kalman gain matrix K (k)=P (k) VH(k)[V(k)P(k)VH(k)+2ψee(k)]-1, wherein P (k) is
State error matrix, ψee(k) it is observation noise error diagonal covariance matrix, subscript H represents conjugate transposition operation；
Update frequency domain filter coefficient vectorWherein
It is constraint matrix, ILBe dimension be L × L unit matrix, 0LIt is the null matrix that dimension is L × L, F is Fourier transform matrix, K
(k) it is kalman gain matrix；
Update state error matrix P (k)：Wherein, ψΔΔ(k)
It is process noise diagonal matrix, I2LIt is the unit matrix that dimension is 2L × 2L.
When it is implemented, in order to calculate kalman gain matrix K (k), it would be desirable to know ψee(k), the sight in the present invention
Survey noise error matrix ψee(k) smooth power spectrum of error noise signal e (n) is used to be calculated.In order to calculate state error
Matrix P (k), we also need to calculating process noise covariance matrix ψΔΔ(k), the process noise diagonal matrix ψ in the present inventionΔΔ
(k) diagonal entry ψΔΔ,i(k) computational methods are：
Wherein,It is filter coefficient vectorI-th of element.Control
Device unit 120 is by the first analog-digital converter AD, digital analog converter DA, the second analog-digital converter AD, the secondary channel of estimation, filtering
Time-domain coefficients and Kalman filtering composition.
In the following, we illustrate the specific implementation of 116 modules in Fig. 2：With reference to Fig. 3, output filtering signal y (n)
Calculating process, wherein length be L filter vectorDigital signal x (n) and
Relational expression between output filtering signal y (n) is：
Wherein,Indicate time domain filter coefficients vectorI-th of element.Each new sampled data arrives
When, we will execute L multiplying according to expression formula (1) and L-1 sub-addition operations obtain convolution and obtain output filtering signal y
(n).Specifically, internal system safeguards that a buffering area or shift register are used to store the hits of current and past
According to.With reference to Fig. 3, when arriving new sampled data, old data pass through a series of shift units 202,204,206 and other
Unit etc. obtains x (n-1), x (n-2), until the total L element of x (n-L+1).Then these elements pass through multiplier unit and filter
Wave device weights are multiplied, that is, digital signal x (n) and weightsBe multiplied by multiplier 208, digital signal x (n-1) and
WeightsIt is multiplied by multiplier 210, digital signal x (n-2) and weightsIt is multiplied by multiplier 212, with such
It pushes away, last x (n-L+1) and weightsBe multiplied by multiplier 214, then all multiplication results adder 216,
218 are added to obtain output filtering signal y (n) with 220.The signal is exported when next sampling instant arrives to digital analog converter
DA110, that is, the delay of filter module is a sampling period Ts.In practice, we can be by improving digital display circuit
Sample frequency fsMake sampling period TsEnough is small, to which the delay can ignore not the influence of the causality of whole system
Meter.Certain sample frequency fsCan not infinitely it increase, because of fsIt is bigger, the exponent number L for the FIR filter that modeling controller needs
It is bigger, to which the calculation amount that expression formula (1) needs also increased dramatically.It needs to make Balancing selection in practical application.Processor
Arithmetic speed must be sufficiently fast, ensures executable expressions (1) as soon as the time needed is less than a sampling period, otherwise this is not one
Real-time system causes the performance of active guidance system to decline even failure.
We discuss another critical issue, the update of processor filter coefficient vector, most common adaptive calculation below
Method is FxLMS algorithms.Processor indicates with a transmission function W (z), the transmission of sub-loudspeaker to the second audio collection device
Function is written as S (z), and the transmission function from the first audio collection device to the second audio collection device is denoted as P (z).So when adaptive
When algorithmic statement is to stable state, we obtain W (z)=- P (z)/S (z), then the acoustic pressure at the second audio collection device is zero, to reach
To the purpose of perfect control noise.
In practical applications, main channel transmission function P (z) and secondary channel transmission function S (z) may be time-varying.Tool
In body to the application of earphone, there is everyone head sizes personalized and individual to wear elasticity that earphone is liked also not
Together, during wearing, the position for adjusting earphone and elasticity that wearer also can be frequently, this results in secondary channel transmission
Function S (z) is variant, to which processor transmission function W (z) is also time-varying.This requires used sef-adapting filters
With good tracking performance.In patent of the present invention, we are obtained 116 in Fig. 2 using frequency domain Kalman filtering algorithm
The filter factor of moduleModule 118 needs the function of realizing in namely Fig. 2.In order to enable technical staff preferably manages
The main thought of the present invention is solved, we specifically give the realization step of module 118 in Fig. 2 in Fig. 4.
However the variation of our controller transfer functions be difficult with accurate mathematical model be depicted come.For ease of description
Problem can meet actual demand again, we describe controller using simplified single order Markov model, that is,
W (k+1)=W (k)+Δ (k) (2)
In above formula, W (k) indicates that the filter coefficient vector of frequency domain, Δ (k) are indicated from kth frame to k+1 frame Time-Frequency Domain Filterings
The variation of coefficient, referred to as process noise.When Δ (k) is close to null vector, expression system has almost no change, when Δ (k) is larger
When, show that system has a greater change.In the language of Kalman filtering, expression formula (2) is referred to as state equation.Now we
To provide the frequency domain adaptive filtering algorithm based on Kalman filtering.
The input signal v (n) of Kalman filter is the secondary channel for inputting digital signal x (n) by estimationFilter
What wave obtained, that is, the expression formula that module 114 in Fig. 2 is realized is：
The operation can be completed in convolution or frequency domain Fourier transformation.It is also to be noted that in order to smoothly real
The method of the now invention needs to estimate secondary channel transmission functionThe estimation of the transmission function can utilize classics in advance
The method of System Discrimination obtain, such as simon He Jin classical treatise《S.Haykin,Adaptive FilterTheory,
5th Edition,Prentice Hall,2013》In describe using various adaptive algorithms can be used for estimate the transmission letter
Number.
With reference to Fig. 4, the specific implementation of Kalman filtering is illustrated.First, Fourier transform unit 302 is by nearest 2L points v
(n) sequence formed does Fourier transformation and obtains frequency domain vector v (k)=F [v (kL-2L+1) ..., v (kL)]T, wherein F is indicated
The element of v (k) is sequentially placed in diagonal line composition diagonal matrix V (k) by Fourier transform matrix.
Nearest L point e (n) sequences formed are done Fourier transformation and obtain frequency domain vector by Fourier transform unit 306
So according to expression formula (2), frequency domain Kalman filtering
Renewal equation is：
Wherein,It is frequency domain filter coefficient vector, actually it is time domain filter coefficients vector2L points
Fourier transformation, K (k) are kalman gain matrix 304,It is constraint matrix, ILBe dimension be L × L
Unit matrix, 0LIt is the null matrix that dimension is L × L.
Now, we illustrate the specific implementation of expression formula (4) according to Fig. 4.Unit 304 calculate card release according to input V (k)
Germania gain matrix K (k), multiplier 308 completes input matrix V (k) and gain matrix K (k) multiplication operations obtain diagonal matrix C
(k).Take the diagonal entry of diagonal matrix C (k) to obtain vectorial c (k), then inverse Fourier transform arithmetic element 310 complete to
The inverse Fourier transform of amount c (k) obtains the real vector a (k) that dimension is 2L × 1.Following sequence latter half zero setting module
312 realize constraint functions, that is, the subsequent L element of sequence a (k) are set to zero and L element of front remains unchanged
The sequence b (k) new to one.Then, fourier transformation module 314 executes Fourier transformation to sequence b (k), in adder 316
The output at place and delay unit 318Addition obtainsSimultaneously willInverse Fourier is carried out in unit 320
Transformation transformation obtainsFor being assigned to the module 116 of Fig. 2, the expression formula of kalman gain matrix K (k) in next frame
For：
K (k)=P (k) VH(k)[V(k)P(k)VH(k)+2ψee(k)]-1 (5)
Wherein, P (k) is state error matrix, ψee(k)=diag { [ψee,0(k),ψee,1(k),…,ψee,2L-1(k)]TBe
System noise error diagonal matrix, subscript H represent conjugate transposition operation.The computational methods of P (k) are：
Wherein, ψee(k)=diag { [ψee,0(k),ψee,1(k),…,ψee,2L-1(k)]TIt is process noise diagonal matrix, I
It is the unit matrix that dimension is 2L × 2L, that is, the implementation of 304 modules is described by expression formula (5) and expression formula (6) in Fig. 4.
The power spectrum ψ of systematic observation noise signalee,i(k) estimation is an important problem, we do not have in practice
There are a kind of means that can measure the system noise signal for being superimposed upon error microphone.But when filter is restrained to a certain extent
When, digital signal e (n) can preferable approximation system observation noise signal.Based on the fact that, in patent of the present invention
In, we replace observation noise power spectrum using the power spectrum of error signal.Namely we are using instantaneous to error signal
Power is smoothly obtained by a low-pass filter, such as following formula：
ψee,i(k)=α ψee,i(k-1)+(1-α)|Ei(k)|2 (7)
Wherein, α is smoothing factor, it is proposed that takes α=0.8.According to this formula, Fig. 5 gives computing system observation noise
The power spectrum ψ of signalee,i(k) specific block diagram, first in module 402 to inputting Ei(k) conjugate operation is taken, then by Ei(k) and
Its conjugation carries out multiplying in Unit 404, is then multiplied by factor 1- α 406 to the output of multiplication unit 404.Delay is single
412 couples of ψ of memberee,i(k) it carries out delay operation and obtains ψee,i(k-1), multiplying, multiplication then are carried out in unit 410 with factor-alpha
The output of unit 406 and the output of multiplication unit 410 sum to obtain ψ at addition unit 408ee,i(k)。
In order to enable above-mentioned algorithm smoothly executes, we also need to the covariance matrix ψ of estimation procedure noise delta (k)ΔΔ
(k).It is understood that the right Section 2 in expression formula (4) formula has reacted the fluctuation of filter coefficient vector to a certain extent
Situation, thus the present invention estimates actual process noise matrix using this：
So in the initial convergence phase of system, ψΔΔ,i(k) bigger value is taken, the receipts of filter can be accelerated in this way
It holds back；When algorithm reaches stable state and actual system fluctuation is smaller, ψΔΔ,i(k) value of very little is taken, this is conducive to algorithm steady
State reaches smaller imbalance；Once and system changes when needing quickly tracking, ψΔΔ,i(k) bigger value can be taken
To accelerate the tracking performance of algorithm.This explains method proposed by the present invention from principle has more than traditional frequency domain algorithm
Good performance.
Finally illustrate, the above examples are only used to illustrate the technical scheme of the present invention and are not limiting, although with reference to reality
Example is applied to describe the invention in detail, it will be understood by those of ordinary skill in the art that, it can be to the technical side of the present invention
Case is modified or replaced equivalently, and without departing from the objective and range of technical solution of the present invention, should all be covered in the present invention
Protection domain in.
Claims (8)
1. a kind of earphone adaptive active noise control system, which is characterized in that the system includes at least：
First audio collection device and the second audio collection device, the first audio collection device are mounted on outside earphone, for acquiring reference
Noise signal；Second audio collection device is mounted on inside earphone, for acquiring the error noise signal after noise reduction；
First analog-digital converter AD and the second analog-digital converter AD, the first analog-digital converter AD and the first audio collection device are electrically connected
It connects, the reference noise signal for acquiring the first audio collection device is converted to corresponding digital signal, the second analog-digital converter
AD and the electrical connection of the second audio collection device, the error noise signal for acquiring the second audio collection device are converted to corresponding number
Word signal；
Processor unit, the first analog-digital converter AD are connected with processor unit first information input terminal, the second analog-digital converter
AD is connected with the second information input terminal of processor unit；
Digital analog converter DA and sub-loudspeaker, processor unit information output is after digital analog converter DA and sub-loudspeaker
Connection, sub-loudspeaker are mounted in earphone, and digital analog converter DA is for being converted to the output filtering signal of processor unit
Corresponding analog signal is simultaneously fed into sub-loudspeaker, and sub-loudspeaker is for playing secondary audio signal.
2. earphone adaptive active noise control system according to claim 1, which is characterized in that the control system is also wrapped
Include memory cell, memory cell and processor unit two-way communication link.
3. earphone adaptive active noise control system according to claim 1, which is characterized in that the control system is also wrapped
Power module is included, power module is connected with processor unit feeder ear.
4. earphone adaptive active noise control system according to claim 1, which is characterized in that the control system is also wrapped
Control logic unit is included, control logic unit is used to control the switch of active noise reduction function.
5. a kind of earphone self-adaptive active noise canceling method, which is characterized in that include the following steps：
It is acquired in real time by the first audio collection device and refers to noise signal, and will turned with reference to noise signal by the first analog-digital converter
It is transferred to processor unit after being changed to corresponding digital signal x (n)；By the real-time Acquisition Error noise letter of the second audio collection device
Number, and it is transferred to processor list after error noise signal is converted to corresponding digital signal e (n) by the second analog-digital converter
Member；
Filter coefficient vector is calculated by processor unitFiltering signal y (n) is exported with calculating, using card in frequency domain
Kalman Filtering calculates filter coefficient vectorTo digital signal x (n) and time domain filter coefficients in time domainInto
Row convolution obtains output filtering signal y (n)；Wherein, time domain filter coefficientsIt is filter coefficient vectorInverse Fu in
Leaf transformation；
Output filtering signal y (n) is converted into corresponding analog signal by digital analog converter and is fed into sub-loudspeaker, it is secondary
Grade loud speaker is for playing secondary audio signal.
6. earphone self-adaptive active noise canceling method according to claim 5, which is characterized in that using card in frequency domain
Kalman Filtering calculates filter coefficient vectorThe step of it is as follows：
In frequency domain, the secondary channel with reference to the corresponding digital signal x (n) of noise signal by estimationFiltering obtains karr
Graceful filter input signal v (n)；
Nearest 2L point Kalman filtering input signal v (n) sequences formed are done into Fourier transformation and obtain frequency domain vector v (k),
The element of frequency domain vector v (k) is sequentially placed in diagonal line composition diagonal matrix V (k)；
Nearest L point tolerance noise signal e (n) sequences formed are done into Fourier transformation and obtain frequency domain vector E (k)；
Update kalman gain matrix K (k)=P (k) VH(k)[V(k)P(k)VH(k)+2ψee(k)]-1, wherein P (k) is state
Error matrix, ψee(k) it is observation noise error diagonal covariance matrix, subscript H represents conjugate transposition operation；
Update frequency domain filter coefficient vectorWhereinIt is about
Beam matrix, ILBe dimension be L × L unit matrix, 0LIt is the null matrix that dimension is L × L, F is Fourier transform matrix, K (k)
It is kalman gain matrix；
Update state error matrix P (k)：Wherein, ψΔΔ(k) it was
Journey noise diagonal matrix, I2LIt is the unit matrix that dimension is 2L × 2L.
7. earphone self-adaptive active noise canceling method according to claim 6, which is characterized in that the observation noise is missed
Poor matrix ψee(k) smooth power spectrum of error noise signal e (n) is used to be calculated.
8. earphone self-adaptive active noise canceling method according to claim 6, which is characterized in that the process noise pair
Angular moment battle array ψΔΔ(k) diagonal entry ψΔΔ, the computational methods of i (k) areWherein,
It is filter coefficient vectorI-th of element.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810758793.1A CN108600894B (en) | 2018-07-11 | 2018-07-11 | Earphone self-adaptive active noise control system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810758793.1A CN108600894B (en) | 2018-07-11 | 2018-07-11 | Earphone self-adaptive active noise control system and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108600894A true CN108600894A (en) | 2018-09-28 |
| CN108600894B CN108600894B (en) | 2023-07-04 |
Family
ID=63615309
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810758793.1A Active CN108600894B (en) | 2018-07-11 | 2018-07-11 | Earphone self-adaptive active noise control system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108600894B (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109658947A (en) * | 2018-11-18 | 2019-04-19 | 南京大学 | A kind of active noise controlling method of synchronous modeling and control |
| CN110166880A (en) * | 2019-07-02 | 2019-08-23 | 上海电机学院 | A kind of modified form adaptive noise reduction earphone and its noise-reduction method |
| CN110737456A (en) * | 2019-09-10 | 2020-01-31 | 中山市格美通用电子有限公司 | wireless earphone online upgrading method and earphone equipment |
| CN110896512A (en) * | 2019-12-13 | 2020-03-20 | 恒玄科技(上海)股份有限公司 | Noise reduction method and system for semi-in-ear earphone and semi-in-ear earphone |
| CN111063333A (en) * | 2019-12-19 | 2020-04-24 | 湖南国声声学科技股份有限公司 | Adaptive noise reduction method, adaptive noise reduction system, adaptive noise reduction device, and computer-readable storage medium |
| CN111341336A (en) * | 2020-03-16 | 2020-06-26 | 北京字节跳动网络技术有限公司 | Echo cancellation method, device, terminal equipment and medium |
| CN111554263A (en) * | 2020-04-30 | 2020-08-18 | 华南理工大学 | Active noise distributed control system and method for open space |
| CN112492438A (en) * | 2020-11-16 | 2021-03-12 | 上海电机学院 | Active noise reduction method for feedback type active noise reduction earphone in cabin |
| CN113115157A (en) * | 2021-04-13 | 2021-07-13 | 北京安声科技有限公司 | Active noise reduction method and device of earphone and semi-in-ear active noise reduction earphone |
| CN113132846A (en) * | 2021-04-13 | 2021-07-16 | 北京安声科技有限公司 | Active noise reduction method and device of earphone and semi-in-ear active noise reduction earphone |
| CN113488016A (en) * | 2021-06-30 | 2021-10-08 | 展讯通信（上海）有限公司 | Coefficient determination method and device |
| CN114040285A (en) * | 2021-09-26 | 2022-02-11 | 北京小米移动软件有限公司 | Method and device for generating parameters of feedforward filter of earphone, earphone and storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100014685A1 (en) * | 2008-06-13 | 2010-01-21 | Michael Wurm | Adaptive noise control system |
| US20110305345A1 (en) * | 2009-02-03 | 2011-12-15 | University Of Ottawa | Method and system for a multi-microphone noise reduction |
| JP2013120358A (en) * | 2011-12-08 | 2013-06-17 | Nippon Hoso Kyokai <Nhk> | Noise suppression device, noise suppression method and noise suppression program |
| CN103905959A (en) * | 2012-12-26 | 2014-07-02 | 上海航空电器有限公司 | Active noise control device based on pilot headset |
| US20160255446A1 (en) * | 2015-02-27 | 2016-09-01 | Giuliano BERNARDI | Methods, Systems, and Devices for Adaptively Filtering Audio Signals |
| CN107171735A (en) * | 2017-05-16 | 2017-09-15 | 浙江工业大学 | A kind of big line width CO OFDM phase noise compensation methods of time-frequency domain Kalman filtering |
| CN108172231A (en) * | 2017-12-07 | 2018-06-15 | 中国科学院声学研究所 | A kind of dereverberation method and system based on Kalman filtering |
-
2018
- 2018-07-11 CN CN201810758793.1A patent/CN108600894B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100014685A1 (en) * | 2008-06-13 | 2010-01-21 | Michael Wurm | Adaptive noise control system |
| US20110305345A1 (en) * | 2009-02-03 | 2011-12-15 | University Of Ottawa | Method and system for a multi-microphone noise reduction |
| JP2013120358A (en) * | 2011-12-08 | 2013-06-17 | Nippon Hoso Kyokai <Nhk> | Noise suppression device, noise suppression method and noise suppression program |
| CN103905959A (en) * | 2012-12-26 | 2014-07-02 | 上海航空电器有限公司 | Active noise control device based on pilot headset |
| US20160255446A1 (en) * | 2015-02-27 | 2016-09-01 | Giuliano BERNARDI | Methods, Systems, and Devices for Adaptively Filtering Audio Signals |
| CN107171735A (en) * | 2017-05-16 | 2017-09-15 | 浙江工业大学 | A kind of big line width CO OFDM phase noise compensation methods of time-frequency domain Kalman filtering |
| CN108172231A (en) * | 2017-12-07 | 2018-06-15 | 中国科学院声学研究所 | A kind of dereverberation method and system based on Kalman filtering |
Non-Patent Citations (3)
| Title |
|---|
| 刘洁波;黄纯;江亚群;汤涛;谢兴;: "基于强跟踪泰勒-卡尔曼滤波器的动态相量估计算法" * |
| 张谦;景占荣;: "一种过程噪声自适应调节的卡尔曼滤波算法" * |
| 郇战;戴永惠;王振海;陶亚辉;: "基于FxLMS算法和预测滤波器的数字耳机降噪研究" * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109658947A (en) * | 2018-11-18 | 2019-04-19 | 南京大学 | A kind of active noise controlling method of synchronous modeling and control |
| CN110166880A (en) * | 2019-07-02 | 2019-08-23 | 上海电机学院 | A kind of modified form adaptive noise reduction earphone and its noise-reduction method |
| CN110737456A (en) * | 2019-09-10 | 2020-01-31 | 中山市格美通用电子有限公司 | wireless earphone online upgrading method and earphone equipment |
| CN110737456B (en) * | 2019-09-10 | 2023-08-15 | 中山市格美通用电子有限公司 | Online upgrading method of wireless earphone and earphone equipment |
| CN110896512A (en) * | 2019-12-13 | 2020-03-20 | 恒玄科技(上海)股份有限公司 | Noise reduction method and system for semi-in-ear earphone and semi-in-ear earphone |
| CN111063333A (en) * | 2019-12-19 | 2020-04-24 | 湖南国声声学科技股份有限公司 | Adaptive noise reduction method, adaptive noise reduction system, adaptive noise reduction device, and computer-readable storage medium |
| CN111341336A (en) * | 2020-03-16 | 2020-06-26 | 北京字节跳动网络技术有限公司 | Echo cancellation method, device, terminal equipment and medium |
| CN111341336B (en) * | 2020-03-16 | 2023-08-08 | 北京字节跳动网络技术有限公司 | Echo cancellation method, device, terminal equipment and medium |
| CN111554263A (en) * | 2020-04-30 | 2020-08-18 | 华南理工大学 | Active noise distributed control system and method for open space |
| CN111554263B (en) * | 2020-04-30 | 2023-03-24 | 华南理工大学 | Active noise distributed control system and method for open space |
| CN112492438B (en) * | 2020-11-16 | 2022-07-26 | 上海电机学院 | Active noise reduction method for feedback type active noise reduction earphone in cabin |
| CN112492438A (en) * | 2020-11-16 | 2021-03-12 | 上海电机学院 | Active noise reduction method for feedback type active noise reduction earphone in cabin |
| CN113132846A (en) * | 2021-04-13 | 2021-07-16 | 北京安声科技有限公司 | Active noise reduction method and device of earphone and semi-in-ear active noise reduction earphone |
| CN113115157A (en) * | 2021-04-13 | 2021-07-13 | 北京安声科技有限公司 | Active noise reduction method and device of earphone and semi-in-ear active noise reduction earphone |
| CN113488016A (en) * | 2021-06-30 | 2021-10-08 | 展讯通信（上海）有限公司 | Coefficient determination method and device |
| CN114040285A (en) * | 2021-09-26 | 2022-02-11 | 北京小米移动软件有限公司 | Method and device for generating parameters of feedforward filter of earphone, earphone and storage medium |
| CN114040285B (en) * | 2021-09-26 | 2024-02-06 | 北京小米移动软件有限公司 | Method and device for generating feedforward filter parameters of earphone, earphone and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108600894B (en) | 2023-07-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108600894A (en) | A kind of earphone adaptive active noise control system and method | |
| CN108428445B (en) | Self-adaptive active noise reduction method of error-free microphone | |
| CN109727604A (en) | Frequency domain echo cancel method and computer storage media for speech recognition front-ends | |
| JP3216704B2 (en) | Adaptive array device | |
| CN100534221C (en) | Directional audio signal processing using an oversampled filterbank | |
| CN108900943A (en) | A kind of scene adaptive active denoising method and earphone | |
| US11245976B2 (en) | Earphone signal processing method and system, and earphone | |
| CN102740214B (en) | Howling suppression method based on feedback signal spectrum estimation | |
| EP2677765B1 (en) | Headphone for active noise suppression | |
| CN104883462B (en) | A kind of sef-adapting filter and filtering method for eliminating acoustic echo | |
| CN110402540B (en) | Active noise reduction method, device, chip, active control system and storage medium | |
| CN109817209A (en) | A kind of intelligent speech interactive system based on two-microphone array | |
| CN105810202B (en) | It is a kind of to drop hypoechoic method, apparatus and communication apparatus | |
| CN111031442A (en) | Earphone active noise reduction self-adaptive system based on DSP virtual sensing | |
| CN105430587A (en) | A Hearing Device Comprising A Gsc Beamformer | |
| CN110085246A (en) | Sound enhancement method, device, equipment and storage medium | |
| CN106161751A (en) | A kind of noise suppressing method and device | |
| CN109379652B (en) | Earphone active noise control secondary channel off-line identification method | |
| CN112562624B (en) | Active noise reduction filter design method, noise reduction method, system and electronic equipment | |
| EP3340653B1 (en) | Active occlusion cancellation | |
| CN110534125A (en) | A kind of real-time voice enhancing system and method inhibiting competitive noise | |
| CN103929704B (en) | The method and system that a kind of adaptive acoustic feedback based on transform domain is eliminated | |
| CN105356861A (en) | Active noise-reduction method and system | |
| CN112331226B (en) | Voice enhancement system and method for active noise reduction system | |
| CN104036783B (en) | MRI scanner adaptive voice strengthening system |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20230608 Address after: Building 11, Phase II, Smart Terminal Optoelectronics Industrial Park, Industrial Park, Pingliang City, Gansu Province 744000 Applicant after: Gansu Midi Acoustics Co.,Ltd. Address before: 402360 Building 1, No. 7, Tianxing Avenue, Shuangqiao Economic Development Zone, Dazu District, Chongqing (101 office buildings for standard plants) Applicant before: CHONGQING CHUANLE AUDIO TECHNOLOGY Co.,Ltd. |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |