CN108184192A - A kind of adaptive acoustic feedback suppressing method - Google Patents
A kind of adaptive acoustic feedback suppressing method Download PDFInfo
- Publication number
- CN108184192A CN108184192A CN201711448561.8A CN201711448561A CN108184192A CN 108184192 A CN108184192 A CN 108184192A CN 201711448561 A CN201711448561 A CN 201711448561A CN 108184192 A CN108184192 A CN 108184192A
- Authority
- CN
- China
- Prior art keywords
- adaptive
- frequency
- signal
- trapper
- algorithm
- 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.)
- Pending
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
- H04R3/00—Circuits for transducers, loudspeakers or microphones
- H04R3/02—Circuits for transducers, loudspeakers or microphones for preventing acoustic reaction, i.e. acoustic oscillatory feedback
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Processing of the speech or voice signal to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
- G10L21/0208—Noise filtering
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Processing of the speech or voice signal to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
- G10L21/0208—Noise filtering
- G10L21/0216—Noise filtering characterised by the method used for estimating noise
- G10L21/0224—Processing in the time domain
Abstract
The invention discloses a kind of adaptive acoustic feedback suppressing methods, by introducing a factor of momentum on the basis of LMS algorithm, that is prior uncertainty, in combination with NLMS algorithms, optimize the update step-length of adaptive algorithm, convergence rate is accelerated, and then adjusts the centre frequency of trapper faster, reduces the output signal distortion finally obtained.The present invention can track the change for frequency of uttering long and high-pitched sounds by adaptive notch method, automatically adjust the centre frequency of trapper.When sampled data is limited, in contrast accurate frequency estimation accuracy can be obtained, thus the distortion level of signal after trap can be reduced.Adaptive algorithm using recursive process, pertains only to Time Domain Processing, does not need to carry out FFT operations, greatly reduces computational complexity.The present invention carries out step-length adjustment by introducing prior uncertainty and variable convergence factor, has further speeded up the convergence rate of adaptive algorithm, algorithm is made to have the real-time that can meet signal processing while preferable acoustic feedback inhibition again.
Description
Technical field
The present invention relates to acoustic feedback algorithm in Speech processing, more particularly, to it is a kind of based on frequency-tracking from
Adapt to acoustic feedback suppressing method.
Background technology
Since sound reinforcement system is widely used, Speech processing is just perplexed by acoustic feedback problem always.Public address
The common trait of system is exactly that voice signal is amplified, to obtain larger output volume.In sound reinforcement system, sound by
Loud speaker is sent out, and when speaker output signal is coupled into microphone, is formed the reponse system of a closed loop, is worked as system
When gain is excessive, due to acoustic feedback phenomenon, ear-piercing howling will be generated.Acoustic feedback seriously limits carrying for system gain
It rises, distorted signals can be directly resulted in.Importantly, acoustic feedback may burn loud speaker and amplifier.Thus, effectively into
Row acoustic feedback inhibits, and is an important subject in sound reinforcement system signal processing.
Fig. 1 is the structure chart of general sound reinforcement system, and closed loop gain is:
The amplitude of generation of uttering long and high-pitched sounds and phase condition are:
The method that acoustic feedback inhibits is carried out from the phase and amplitude condition for destroying generation of uttering long and high-pitched sounds.At digital signal
The acoustic feedback suppressing method of reason is mainly the following:
In the fifties in last century, foreign countries just have document to propose to carry out chauvent's criterion using shift frequency method.Shift frequency method is to pass through liter
The Production conditions that frequency content destruction that is high or reducing input audio signal is uttered long and high-pitched sounds.The output signal for changing frequency enters back into and is
System will not be superimposed with original signal frequency again, so as to achieve the purpose that chauvent's criterion.It is easy that shift frequency method is realized, but can cause frequency
Rate is distorted, larger to sound quality damage, and human ear can differentiate this variation of low frequency signal.
Random phase method using people the sense of hearing it is insensitive to acoustical phase the characteristics of, in audio feedback circuit by addition with
Machine phase system makes the phase of acoustic feedback open-loop transfer function become time-varying, destroys the phase condition for generation of uttering long and high-pitched sounds.Typically
Random phase system is realized by all-pass filter, can guarantee in this way input signal by when only have change in phase.But
It is random phase method the problem is that when the phase condition for destroying a feedback paths, another feedback paths may be made
Phase condition meet, generate new frequency point of uttering long and high-pitched sounds, thus performance is not so good as other methods.
Wave trap method is that frequency point of uttering long and high-pitched sounds is inhibited into line level by notch filter on forward path, can be to a certain degree
On achieve the purpose that chauvent's criterion.First frequency point of uttering long and high-pitched sounds is detected before trap, this just needs to become by Short-time Fourier
It gets the power spectral information of signal in return, on the one hand increases the processing delay of system;On the other hand, inhibition is for frequency of uttering long and high-pitched sounds
The detection of point and notch bandwidth are provided with very big dependence.Particularly, for dsp system, if very few per frame sampling point,
The precision for frequency point detection of uttering long and high-pitched sounds may be not achieved.When input signal is handled by trapper, if the value for frequency point detection of uttering long and high-pitched sounds
It is larger to deviate actual value, along with notch bandwidth setting is unreasonable, does not have trap effect not only in this way, audio can be curbed instead
Other frequency contents, cause the distortion of bigger.
Therefore when carrying out acoustic feedback inhibition, a kind of real-time that not only can guarantee signal processing is designed, but also can
The method of the distortion level of audio is particularly important after reduction processing as possible.
Invention content
The purpose of the present invention is to provide a kind of adaptive acoustic feedback suppressing methods, can overcome the shortcomings of notch algorithm,
The processing delay of notch algorithm is reduced, while dependence of the notch algorithm for frequency point detection of uttering long and high-pitched sounds can be weakened again, is made by calculating
While method treated signal is completed acoustic feedback and inhibited, reduce distortion level as possible.
The present invention can track the change for frequency of uttering long and high-pitched sounds by adaptive notch method, automatically adjust the center frequency of trapper
Rate.When sampled data is limited, in contrast accurate frequency estimation accuracy can be obtained, thus signal after trap can be reduced
Distortion level.Adaptive algorithm using recursive process, pertains only to Time Domain Processing, does not need to carry out FFT operations, greatly reduce
Computational complexity.The present invention carries out step-length adjustment by introducing prior uncertainty and variable convergence factor, has further speeded up adaptive
Convergence speed of the algorithm is answered, algorithm is made to have the real-time that can meet signal processing while preferable acoustic feedback inhibition again.
Description of the drawings
Fig. 1 is sound reinforcement system structure chart;
Fig. 2 is system structure diagram of the embodiment of the present invention;
Fig. 3 is adaptive notch filter structure chart of the embodiment of the present invention;
Fig. 4 is the oscillogram containing signal of uttering long and high-pitched sounds;
Fig. 5 is that the embodiment of the present invention completes the oscillogram after chauvent's criterion.
Specific embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, with reference to the accompanying drawings and embodiments, it is right
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
Since B.Widrow in 1967 et al. proposition, adaptive filter algorithm is just widely used.Adaptively
Wave filter to reach the study and tracking to signal, while can be also not required to according to certain criterion, adjust automatically filter parameter
It will be to the priori statistical knowledge of signal.By adaptive algorithm, intelligent-tracking is uttered long and high-pitched sounds frequency point, it is possible to overcome wave trap method detection essence
The problem of spending.And due to only using the processing of time domain recursion, so can also reduce processing delay.But it needs to weigh wave filter ginseng
Several renewal speed and the distortion factor of audio.LMS algorithm is most basic adaptive filter algorithm, but there are convergence rate compared with
The problem of slow.
The present invention, by introducing a factor of momentum, i.e. prior uncertainty, calculates on the basis of LMS algorithm in combination with NLMS
Method optimizes the update step-length of adaptive algorithm, accelerates convergence rate.The centre frequency of trapper can be adjusted faster, most
The output signal distortion obtained eventually is also smaller.Implementation is as follows:
1st, the signal that microphone is collected into is AD converted by audio collection module, obtains digitized audio letter
Number;
2nd, according to adaptive notch algorithm, the centre frequency, the bandwidth of wave filter, state of first initial transmission function are needed
The constant of variable and adjusting step.When carrying out chauvent's criterion, it is desirable to which obtain maximum attenuation in frequency point of uttering long and high-pitched sounds believes output simultaneously
Number reduce as possible relative to the distortion of phonetic entry, then it is necessary to which the notch bandwidth near the Frequency point uttered long and high-pitched sounds is made to the greatest extent may be used
It can be small.Using fixed bandwidth in the present invention, and the frequency for generation of being uttered long and high-pitched sounds by adaptive iteration algorithm keeps track, it is updated with this sunken
The centre frequency of wave device.After adaptive algorithm restrains, frequency-tracking process is completed, just forms trapper accordingly to uttering long and high-pitched sounds
Frequency point carries out trap processing, the signal after the signal that trapper exports is chauvent's criterion.
3rd, the output signal after chauvent's criterion is completed by power amplifier and loud speaker, completes amplification output.
The principle that trapper carries out acoustic feedback inhibition is as follows:
According to the transfer function of second order IIR filter:
Wherein digital center frequency is:Notch bandwidth BW=π (1-r).When can accurately obtain in number
During frequency of heart, coefficient r is set relatively large to reduce notch filter, preferable chauvent's criterion effect can be obtained.
Regard the numerical frequency correlation function of second order IIR filter transfer function as adaptive-filtering coefficient, i.e.,It can be obtained by:
Using the least mean-square error of trapper output signal as criterion, i.e., the replacement criteria of adaptive algorithm is:mina(n)E
(|y(n)|2)。
Illustrate the step of adaptive notch acoustic feedback inhibits with reference to specific example:
Step 1:Signal acquisition
Structure diagram of the attached drawing 2 for entire adaptive acoustic feedback suppression system.Speaker output signal, by feeding back road
Diameter and voice coupling.Audio collection module is to the audio signal for carrying acoustic feedback with sample frequency fsSampling framing is carried out, if frame
A length of N using rectangular window, 0 is initialized as per frame data, then by the data deposit frame after sampling, obtained sample sequence is:
xi=[x (i*N), x (i* (N-1)) ..., x (i)] (i=1,2 ..., n), is then passed to adaptive notch algorithm
Module is handled;
Step 2:Adaptive acoustic feedback inhibits
After receiving the digital audio sequence in step 1, according to paper Waterschoot T V, Moonen M.A Pole-
Zero Placement Technique for Designing Second-Order IIR Parametric Equalizer
Filters[J].IEEE Transactions on Audio Speech&Language Processing,2007,15(8):
Definition in 2561-2565. about parameter in filter transfer function first initializes the parameter of adaptive notch algoritic module:
Centre frequency a (n)=0 of trapper,
Determine the parameter r=0.75 of notch filter,
Convergence factor convergence factor μ (n)=0.0025 for approaching step-length of adaptive algorithm is controlled,
Trapper state variable:T (n), u (n) are initialized as 0.
According to formula (2), filter transfer function is rewritten as:
It is write formula (3) as the two cascade forms of wave filter and can be obtained by the state variable of wave filter u (n) with defeated
Enter the relationship of outlet chamber.
According to the input signal x (n) that the centre frequency a (n) of current trapper and step 1 obtain, adaptive notch is updated
The state variable of device:T (n), u (n), frequency variableWith the factor mu (n) of control iteration convergence step-length, recursive process is such as
Under:
T (n+1)=u (n)-ra (n) t (n)-r2t(n-1)
U (n+1)=x (n)-ra (n) u (n)-r2u(n-1)
(take β=0.5, σ=2
The output of signal after being inhibited simultaneously:
Y (n)=u (n+1)+a (n) u (n)+u (n-1)
According to the centre frequency of the signal update trapper after state variable and chauvent's criterion:
A (n+1)=a (n) -2 μ (n) y (n-1) x (n)
Due toSo when the amplitude of a (n) is excessive by a (n) be limited to [- 2,2] it
Between.By the cascade expression of formula (3) and above-mentioned recursive process, structure such as Fig. 3 of wave filter may finally be determined.
After having handled 1 frame, by shifting processing, make next frame the state variable value t (n) that and then previous frame obtains, u
(n), output valve y (n), convergence factor μ (n) and trapper center frequency estimation value a (n) carry out recursion processing.
Step 3:The signal y (n) after acoustic feedback inhibits will be completed in step 2 after power amplifier amplifies by raising one's voice
Device exports.
The final result such as Fig. 4, Fig. 5.Fig. 4 is that audio is not passed through sound reinforcement system by adaptive notch algorithm
Waveform, it can be seen that the generation process uttered long and high-pitched sounds.Fig. 5 is obtained after algorithm process proposed by the present invention is added in sound reinforcement system
Output waveform, comparison diagram 4, Fig. 5 can be seen that through adaptive acoustic feedback suppressing method proposed by the present invention, can effectively inhibit
It utters long and high-pitched sounds the generation of signal, the distortion of synchronous signal is also smaller.Observe a time more forward part in Fig. 5, it can be seen that use
Method proposed by the present invention due to introducing prior uncertainty, in combination with variable step size convergence factor, optimizes adaptive algorithm
Update step-length, so the transient process of adaptive updates is very short, the voice of output hardly feels the presence of howling, tool
There is good real-time.
In short, the present invention is applicable to carry out the scene of acoustic feedback inhibition, tracking that can be relatively accurate in algorithm is maked a whistling sound
Under the premise of being frequency point, the convergence rate of adaptive algorithm is accelerated, reduces the distorted signals degree after acoustic feedback inhibits, energy
It is preferable to achieve the effect that acoustic feedback inhibits, while reduces the processing delay of algorithm again, there is preferable real-time.
The technical solution provided above the embodiment of the present invention is described in detail, specific case used herein
The principle of the present invention and embodiment are expounded, the explanation of above example is only intended to help to understand side of the invention
Method and its core concept;Meanwhile for those of ordinary skill in the art, thought according to the present invention, in specific embodiment
And there will be changes in application range, in conclusion the content of the present specification should not be construed as limiting the invention.
Claims (4)
1. a kind of adaptive acoustic feedback suppressing method, it is characterised in that:On the basis of LMS algorithm by introduce a momentum because
Son, i.e. prior uncertainty in combination with NLMS algorithms, optimize the update step-length of adaptive algorithm, accelerate convergence rate, and then
The centre frequency of adjustment trapper faster reduces the output signal distortion finally obtained.
2. a kind of adaptive acoustic feedback suppressing method according to claim 1, it is characterised in that include the following steps:
A is AD converted the signal that microphone is collected by audio collection module, obtains digitized audio signal;
B is according to adaptive notch algorithm, centre frequency, the bandwidth of wave filter, state variable and the tune of first initial transmission function
The constant of synchronizing length, using fixed bandwidth, the frequency for generation of being uttered long and high-pitched sounds by adaptive iteration algorithm keeps track updates trap with this
The centre frequency of device after adaptive algorithm restrains, completes frequency-tracking process, just forms trapper accordingly to frequency of uttering long and high-pitched sounds
Point carries out trap processing, the signal after the signal that trapper exports is chauvent's criterion;
Output signal after C completion chauvent's criterions passes through power amplifier and loud speaker, completes amplification output.
3. a kind of adaptive acoustic feedback suppressing method according to claim 2, it is characterised in that the step A includes:It is right
Audio signal with acoustic feedback is with sample frequency fsSampling framing is carried out, if frame length is N, using rectangular window, at the beginning of every frame data
Beginning turns to 0, then by the data deposit frame after sampling, obtained sample sequence is:
xi=[x (i*N), x (i* (N-1)) ..., x (i)] (i=1,2 ..., n).
4. a kind of adaptive acoustic feedback suppressing method according to claim 2, it is characterised in that step B includes:
B1 first initializes the parameter of adaptive notch algoritic module:Centre frequency a (n)=0 of trapper, bandwidth parameter r=
0.75, convergence factor μ (n)=0.0025, trapper state variable:T (n), u (n) are initialized as 0;
Filter transfer function is rewritten as by B2 according to the transfer function of second order IIR filter:
Being write as two cascade forms of wave filter, to can be obtained by factor of momentum u (n) in the state variable of wave filter defeated with inputting
Relationship between going out;
B3 updates the state of adaptive notch filter according to the centre frequency a (n) of current trapper and obtained input signal x (n)
Variable:Time t (n), factor of momentum u (n), frequency variableWith convergence factor μ (n), while signal after being inhibited
Export y (n);
B4 is according to the centre frequency of the signal update trapper after state variable and chauvent's criterion;
B5 makes the next frame state variable value that and then previous frame obtains by shifting processing:Time t (n), factor of momentum u
(n), output valve y (n), convergence factor μ (n) and trapper center frequency estimation value a (n) carry out recursion processing.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711448561.8A CN108184192A (en) | 2017-12-27 | 2017-12-27 | A kind of adaptive acoustic feedback suppressing method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711448561.8A CN108184192A (en) | 2017-12-27 | 2017-12-27 | A kind of adaptive acoustic feedback suppressing method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108184192A true CN108184192A (en) | 2018-06-19 |
Family
ID=62547765
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711448561.8A Pending CN108184192A (en) | 2017-12-27 | 2017-12-27 | A kind of adaptive acoustic feedback suppressing method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108184192A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110545350A (en) * | 2019-07-17 | 2019-12-06 | 浙江工业大学 | Method for increasing audio playing volume in mobile phone hands-free state |
CN111273230A (en) * | 2020-03-02 | 2020-06-12 | 开放智能机器(上海)有限公司 | Sound source positioning method |
WO2020150864A1 (en) * | 2019-01-21 | 2020-07-30 | 阿里巴巴集团控股有限公司 | Audio signal detection method and apparatus, and audio and video conference system |
CN111541437A (en) * | 2020-05-13 | 2020-08-14 | 上海航天控制技术研究所 | Filter algorithm suitable for elastic frequency time-varying characteristic |
CN112198911A (en) * | 2020-09-29 | 2021-01-08 | 上海大学 | Method and system for eliminating random line spectrum self-adaptive tracking |
CN116439913A (en) * | 2023-04-14 | 2023-07-18 | 中国人民解放军海军潜艇学院 | Marine active hearing protection type earmuff and protection method thereof |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103413556A (en) * | 2013-08-27 | 2013-11-27 | 广州顶毅电子有限公司 | Audio feedback signal suppression method |
CN103475980A (en) * | 2013-07-19 | 2013-12-25 | 杭州联汇数字科技有限公司 | Self-adaptive acoustic-feedback-eliminating method |
CN106454642A (en) * | 2016-09-23 | 2017-02-22 | 佛山科学技术学院 | Adaptive sub-band audio feedback suppression method |
WO2017138094A1 (en) * | 2016-02-09 | 2017-08-17 | 三菱電機株式会社 | Active noise control device |
CN107291663A (en) * | 2017-06-12 | 2017-10-24 | 华侨大学 | The variable step suppressed applied to acoustic feedback normalizes sub-band adaptive filtering method |
-
2017
- 2017-12-27 CN CN201711448561.8A patent/CN108184192A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103475980A (en) * | 2013-07-19 | 2013-12-25 | 杭州联汇数字科技有限公司 | Self-adaptive acoustic-feedback-eliminating method |
CN103413556A (en) * | 2013-08-27 | 2013-11-27 | 广州顶毅电子有限公司 | Audio feedback signal suppression method |
WO2017138094A1 (en) * | 2016-02-09 | 2017-08-17 | 三菱電機株式会社 | Active noise control device |
CN106454642A (en) * | 2016-09-23 | 2017-02-22 | 佛山科学技术学院 | Adaptive sub-band audio feedback suppression method |
CN107291663A (en) * | 2017-06-12 | 2017-10-24 | 华侨大学 | The variable step suppressed applied to acoustic feedback normalizes sub-band adaptive filtering method |
Non-Patent Citations (9)
Title |
---|
CHING-AN LAI: "NLMS algorithm with decreasing step size for adaptive IIR filters", 《SIGNAL PROCESSING》 * |
J. H. PARK等: "A NOVEL ADAPTIVE ALGORITHM WITH AN IIR FILTER", 《INTERNATIONAL JOURNAL OF AUTOMOTIVE TECHNOLOGY》 * |
TOON VAN WATERSCHOOT等: "A Pole-Zero Placement Technique for Designing Second-Order IIR Parametric Equalizer Filters", 《IEEE TRANSACTIONS ON AUDIO, SPEECH, AND LANGUAGE PROCESSING》 * |
刘杰等: "基于自适应IIR陷波器的声反馈抑制方法", 《电声技术》 * |
孙延龙: "基于双系统耦合理论的自适应滤波算法的研究", 《烟台大学硕士学位论文》 * |
杨阳等: "声学系统中反馈抑制器的研究和DSP实现", 《中国优秀硕士学位论文全文数据库,信息科技辑》 * |
袁高爽茜: "基于MATLAB的自适应啸叫抑制算法与评价标准的研究", 《中国优秀硕士学位论文全文数据库,信息科技辑》 * |
赵娟等: "基于变动量因子的常模盲均衡算法研究", 《荆楚理工学院学报》 * |
邵成华: "通信中的自适应噪声抵消方法研究", 《中国优秀硕士学位论文全文数据库,信息科技辑》 * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2020150864A1 (en) * | 2019-01-21 | 2020-07-30 | 阿里巴巴集团控股有限公司 | Audio signal detection method and apparatus, and audio and video conference system |
CN110545350A (en) * | 2019-07-17 | 2019-12-06 | 浙江工业大学 | Method for increasing audio playing volume in mobile phone hands-free state |
CN111273230A (en) * | 2020-03-02 | 2020-06-12 | 开放智能机器(上海)有限公司 | Sound source positioning method |
CN111273230B (en) * | 2020-03-02 | 2022-06-07 | 开放智能机器(上海)有限公司 | Sound source positioning method |
CN111541437A (en) * | 2020-05-13 | 2020-08-14 | 上海航天控制技术研究所 | Filter algorithm suitable for elastic frequency time-varying characteristic |
CN112198911A (en) * | 2020-09-29 | 2021-01-08 | 上海大学 | Method and system for eliminating random line spectrum self-adaptive tracking |
CN116439913A (en) * | 2023-04-14 | 2023-07-18 | 中国人民解放军海军潜艇学院 | Marine active hearing protection type earmuff and protection method thereof |
CN116439913B (en) * | 2023-04-14 | 2024-03-15 | 中国人民解放军海军潜艇学院 | Marine active hearing protection type earmuff and protection method thereof |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108184192A (en) | A kind of adaptive acoustic feedback suppressing method | |
CN109727604B (en) | Frequency domain echo cancellation method for speech recognition front end and computer storage medium | |
CN109065067B (en) | Conference terminal voice noise reduction method based on neural network model | |
CN106454642B (en) | Adaptive sub-band audio feedback suppression methods | |
DE60116255T2 (en) | NOISE REDUCTION DEVICE AND METHOD | |
JPH06189395A (en) | Hearing sensation auxiliary equipment, noise suppressor and feedback suppressor with adaptive filter function that is converged | |
JP2002541753A (en) | Signal Noise Reduction by Time Domain Spectral Subtraction Using Fixed Filter | |
CN108200522A (en) | A kind of change regularization ratio normalization sub-band adaptive filtering method | |
EP3025516B1 (en) | Automatic timbre, loudness and equalization control | |
Gil-Cacho et al. | Wiener variable step size and gradient spectral variance smoothing for double-talk-robust acoustic echo cancellation and acoustic feedback cancellation | |
CN111213359A (en) | Echo canceller and method for echo canceller | |
US11647344B2 (en) | Hearing device with end-to-end neural network | |
Wang et al. | NN3A: Neural network supported acoustic echo cancellation, noise suppression and automatic gain control for real-time communications | |
CN103929704B (en) | The method and system that a kind of adaptive acoustic feedback based on transform domain is eliminated | |
Zheng et al. | A deep learning solution to the marginal stability problems of acoustic feedback systems for hearing aids | |
CN112331226B (en) | Voice enhancement system and method for active noise reduction system | |
Sadjadi et al. | A comparison of front-end compensation strategies for robust LVCSR under room reverberation and increased vocal effort | |
JP2002541529A (en) | Reduction of signal noise by time domain spectral subtraction | |
CN110010150A (en) | Auditory Perception speech characteristic parameter extracting method based on multiresolution | |
Mishra et al. | Unsupervised noise-aware adaptive feedback cancellation for hearing aid devices under noisy speech framework | |
Gil-Cacho et al. | Regularized adaptive notch filters for acoustic howling suppression | |
CN115620737A (en) | Voice signal processing device, method, electronic equipment and sound amplification system | |
Lee et al. | Deep neural network-based speech separation combining with MVDR beamformer for automatic speech recognition system | |
Loetwassana et al. | Adaptive howling suppressor in an audio amplifier system | |
CN105744456A (en) | Digital hearing-aid self-adaptive sound feedback elimination method |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180619 |
|
RJ01 | Rejection of invention patent application after publication |