CN109817234B - Target speech signal enhancement method, system and storage medium based on continuous noise tracking - Google Patents
Target speech signal enhancement method, system and storage medium based on continuous noise tracking Download PDFInfo
- Publication number
- CN109817234B CN109817234B CN201910168105.0A CN201910168105A CN109817234B CN 109817234 B CN109817234 B CN 109817234B CN 201910168105 A CN201910168105 A CN 201910168105A CN 109817234 B CN109817234 B CN 109817234B
- Authority
- CN
- China
- Prior art keywords
- noise
- speech
- signal
- voice
- fourier transform
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 25
- 238000001228 spectrum Methods 0.000 claims abstract description 24
- 238000011084 recovery Methods 0.000 claims abstract description 4
- 238000004422 calculation algorithm Methods 0.000 claims description 12
- 238000004590 computer program Methods 0.000 claims description 6
- 230000003595 spectral effect Effects 0.000 claims description 6
- 230000006870 function Effects 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 2
- 230000009286 beneficial effect Effects 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 2
- 230000001629 suppression Effects 0.000 description 2
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000009432 framing Methods 0.000 description 1
- 230000002401 inhibitory effect Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
Images
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
-
- 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
- G10L21/0272—Voice signal separating
-
- 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
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/03—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters
-
- 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
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/45—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of analysis window
Landscapes
- Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Signal Processing (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Quality & Reliability (AREA)
- Circuit For Audible Band Transducer (AREA)
- Noise Elimination (AREA)
Abstract
The invention provides a target speech signal enhancement method, a system and a storage medium based on continuous noise tracking, wherein the target speech signal enhancement method comprises the following steps: step 1: receiving a voice signal with noise, performing frame division and windowing processing on the voice signal with noise, and obtaining a time-frequency domain relation by using short-time Fourier transform; step 2: estimating a noise power spectrum; and step 3: estimating a speech power spectrum; and 4, step 4: estimating a speech signal by a speech estimator; and 5: inverse fourier transform, windowing and using overlap-add techniques to achieve speech recovery. The invention has the beneficial effects that: the invention effectively separates the target voice signal, greatly reduces the noise residual amount in the voice signal and greatly improves the quality of the target signal. This is very important for applications such as automatic speech recognition, speaker recognition, man-machine interface and hearing aids.
Description
Technical Field
The present invention relates to the field of speech processing technologies, and in particular, to a target speech signal enhancement method and system based on continuous noise tracking, and a storage medium.
Background
Noise exists everywhere in life, and the aim of the voice enhancement algorithm is to improve the quality and intelligibility of target voice signals polluted by the noise. Existing speech enhancement algorithms typically employ a speech activity detector to estimate background noise and thus achieve target signal enhancement, and these algorithms perform well in stationary noise environments and high signal-to-noise ratio conditions. However, the performance of these algorithms is very limited when at low signal-to-noise ratios, especially in non-stationary noise environments. Because the noise in life is complex, for example, when a car or a train passes through, various noises are generated by speaking and chatting of pedestrians, and it is necessary to develop a speech enhancement algorithm which can work well under the condition of non-stationary noise.
Disclosure of Invention
The invention provides a target speech signal enhancement method based on continuous noise tracking, which comprises the following steps:
step 1: receiving a voice signal with noise, performing frame division and windowing processing on the voice signal with noise, and then obtaining a time-frequency domain relation by using short-time Fourier transform;
step 2: estimating a noise power spectrum;
and step 3: estimating a clean speech power spectrum;
and 4, step 4: the pure voice signal is estimated through a voice estimator, and the prior signal-to-noise ratio of the voice estimator is realized by utilizing a decision-directed algorithm estimator;
and 5: inverse fourier transform, windowing and using overlap-add techniques to achieve clean speech recovery.
As a further improvement of the present invention, in said step 2, a minimum mean square error estimator of the noise power is used to estimate the noise power spectrum.
As a further improvement of the present invention, in said step 3, a minimum mean square error estimator based on the existence probability of the voice is used to realize the calculation of the voice power spectrum.
As a further improvement of the present invention, in the step 4, a prior probability model based on generalized gamma is applied to obtain a minimum mean square error speech enhancement algorithm.
The invention also provides a target speech signal enhancement system based on continuous noise tracking, which comprises: memory, a processor and a computer program stored on the memory, the computer program being configured to carry out the steps of the method of the invention when called by the processor.
The invention also provides a computer-readable storage medium having stored thereon a computer program configured to, when invoked by a processor, perform the steps of the method of the invention.
The invention has the beneficial effects that: the invention effectively separates the target voice signal, greatly reduces the noise residual amount in the voice signal and greatly improves the quality of the target signal. This is very important for applications such as automatic speech recognition, speaker recognition, man-machine interface and hearing aids.
Drawings
Fig. 1 is a framework diagram of the present invention.
FIG. 2(a) is a time domain waveform of noisy speech contaminated by traffic noise under the condition of a signal-to-noise ratio of-5 dB.
Fig. 2(b) is a graph showing a comparison of tracking performance of the respective noise tracking methods against a rapidly changing noise level.
FIG. 3 is a speech waveform diagram, in which (a) is a clean speech diagram, (b) is a noisy speech diagram, and (c) is an enhanced speech diagram.
Detailed Description
The invention discloses a target voice signal enhancement method based on continuous noise tracking, which can realize effective separation of a target source signal and background noise aiming at noise in life.
As shown in fig. 1, the frame of the present invention comprises two main parts: a speech estimator and a noise tracker.
And (3) signal model: we consider an additive signal model, y (n) ═ x (n) + d (n), where y (n) is the noisy speech signal, and x (n) and d (n) represent the clean speech signals, respectivelySign and noise signal. The relationship of the time-frequency domain is obtained by using a short-time fourier transform, Y (l, k) ═ X (l, k) + D (l, k), where l and k represent the frame number and the index of the frequency point, respectively. The polar coordinate representation is as follows: y is Rejα,X＝AejβAnd D ═ Nejθ。E{|X(l,k)|2}＝λxAnd E { | D (l, k) > ceiling2}＝λdThe variance of the speech and noise signals, respectively. From fig. 1 we see the main flow of the process: 1. the noisy speech signal is subjected to framing and windowing, followed by a short-time Fourier transform → 2. noise power spectrum estimation → 3. Prior Signal-to-noise ratio estimation → 4. speech signal estimation → 5. synthesis (inverse Fourier transform, windowing and using overlap-and-add techniques to achieve speech recovery).
The target speech signal enhancement method based on continuous noise tracking comprises the following steps:
in step 1: and receiving the voice signal with noise, performing frame division and windowing processing on the voice signal with noise, and then obtaining the relation of a time-frequency domain by using short-time Fourier transform.
A noise tracker:
to estimate the noise power spectrum, we propose to estimate the noise power spectrum using a minimum mean square error estimator of the noise power. Therefore, in step 2, a minimum mean square error estimator of the noise power is used to estimate the noise power spectrum.
Using Bayesian criteria, we can derive a minimum mean square error estimator, as follows
In formula (1), n (N) represents a noise spectrum variable, θ represents an angle (a real part and an imaginary part have an angle) of a noise short-time Fourier transform coefficient,
since the Fourier coefficients of clean speech and noise are assumed to follow a Gaussian distribution, we can obtain
In the formula (2), n (N) represents a noise spectrum variable, λdRepresentative of the power spectral density of the noise,
wherein λxRepresentative of the power spectral density of speech, which we derive by derivation
Where xi ═ λx/λdAnd γ ═ R2/λdRepresenting the prior and posterior signal-to-noise ratios, respectively, and R ═ Y (l, k) | is the magnitude of the short-time fourier transform coefficient of the noisy speech. From equation (3) we can see the computation of the prior snr from the noise estimator, which requires information of the speech power spectrum. The next step is the estimation of the clean speech power spectrum.
And step 3: the estimation of the pure speech power spectrum is realized by using a minimum mean square error estimator based on the existence probability of speech, and the expression is
In formula (4), a (a) represents the speech spectrum amplitude, the upper case represents the variable, the lower case represents the value of the variable, and the former noise is the same as that in this case. H1And H0Representing a binary hypothesis, which refers to the presence and absence of speech, respectively.
Since the second part is zero, we only need to compute the first part. Can be calculated by using Bayesian formula
In equation (5), β denotes the angle of the speech short-time fourier transform coefficient.
Further by deriving and integrating with a special integration function Bessel, we obtain
For the existing probability of the voice, a simple and effective probability estimation method is obtained by using a fixed prior signal-to-noise ratio. The probability is calculated by the formula
In the formula (7), the first and second groups,
representing an estimate of the a priori signal-to-noise ratio.
A voice estimator:
the algorithm obtains a minimum mean square error speech enhancement algorithm by using a prior probability model based on generalized gamma. The prior generalized gamma probability model is
In equation (8), the variable represented by a is the speech spectral amplitude,
the other parameters are the shape parameters of the gamma model.
Our parameters were chosen as μ ═ 1 and ν ═ 6. For the a priori signal-to-noise ratio of the speech estimator, we use a decision directed algorithm estimator to implement.
In order to evaluate the performance of the method, a large number of experiments are carried out, and the method is proved to be capable of effectively inhibiting non-stationary noise so as to enhance the target speech signal. From fig. 2 and fig. 3, it can be seen more intuitively that the method can achieve efficient noise tracking and suppression of non-stationary noise.
Fig. 2(a) and 2(b) show the experimental results of tracking noise, and we can clearly see that the proposed noise tracker can track the rapid changes of the noise level quickly and accurately. Fig. 3 shows waveforms of clean speech, noisy speech and enhanced speech, and we can very intuitively see that non-stationary noise is well suppressed. Overall, this method works very well for suppression of non-stationary noise.
The invention has the following beneficial effects:
1. the target speech signal enhancement method of the present invention eliminates the need for a speech activity detector to detect speech segments and speech segments.
2. Noise can be continuously tracked even in a speech segment, and accurate tracking and estimation of the rapidly changing noise level are completed.
3. The target voice signal is effectively separated, the noise residual quantity in the voice signal is greatly reduced, and the quality of the target signal is greatly improved. This is very important for applications such as automatic speech recognition, speaker recognition, man-machine interface and hearing aids.
The foregoing is a more detailed description of the invention in connection with specific preferred embodiments and it is not intended that the invention be limited to these specific details. For those skilled in the art to which the invention pertains, several simple deductions or substitutions can be made without departing from the spirit of the invention, and all shall be considered as belonging to the protection scope of the invention.
Claims (3)
1. A target speech signal enhancement method based on continuous noise tracking is characterized by comprising the following steps:
step 1: receiving a voice signal with noise, performing frame division and windowing processing on the voice signal with noise, and then obtaining a time-frequency domain relation by using short-time Fourier transform;
step 2: estimating a noise power spectrum;
and step 3: estimating a clean speech power spectrum;
and 4, step 4: the pure voice signal is estimated through a voice estimator, and the prior signal-to-noise ratio of the voice estimator is realized by utilizing a decision-directed algorithm estimator;
and 5: performing inverse Fourier transform, windowing and using an overlap-add technique to achieve clean speech recovery;
in the step 2, estimating a noise power spectrum by using a minimum mean square error estimator of the noise power;
in the step 2, a minimum mean square error estimator can be obtained by using a Bayes criterion, and the formula is as follows
In the formula (1), N (N) represents a noise spectrum variable, theta represents an angle of a noise short-time Fourier transform coefficient, N and N respectively represent the amplitude of a variable noise spectrum and a value thereof, Y represents a short-time Fourier transform coefficient of noisy speech, E {. cndot } represents an expected operator, d has no meaning, and d is only a general expression mode in an integral formula;
since the Fourier coefficients of clean speech and noise are assumed to follow a Gaussian distribution, the result is
In the formula (2), n (N) represents a noise spectrum variable, λdRepresentative of the power spectral density of the noise,
wherein λxRepresenting the power spectral density of speech, R representing the amplitude of the noisy speech spectrum, I0(. cndot.) A first class of Bezier functions representing a zero order correction;
by deducing we obtain
Where xi ═ λx/λdAnd γ ═ R2/λdRespectively representing a priori signal-to-noise ratio and a posteriori signal-to-noise ratio, wherein R ═ Y (l, k) | is the amplitude of a short-time Fourier transform coefficient of the voice with noise;
in the step 3, a minimum mean square error estimator based on the existence probability of the voice is used for realizing the calculation of a pure voice power spectrum;
in the step 4, a least mean square error speech enhancement algorithm is obtained by applying a prior probability model based on generalized gamma;
the prior generalized gamma probability model is
The parameters μ ═ 1 and ν ═ 6;
in equation (8), the variable represented by a is the speech spectral amplitude,
Γ (·) represents the gamma function, and the other parameters are the shape parameters of the gamma model.
2. A target speech signal enhancement system based on continuous noise tracking, comprising: memory, a processor and a computer program stored on the memory, the computer program being configured to carry out the steps of the method of claim 1 when invoked by the processor.
3. A computer-readable storage medium characterized by: the computer-readable storage medium stores a computer program configured to implement the steps of the method of claim 1 when invoked by a processor.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910168105.0A CN109817234B (en) | 2019-03-06 | 2019-03-06 | Target speech signal enhancement method, system and storage medium based on continuous noise tracking |
| PCT/CN2019/117078 WO2020177374A1 (en) | 2019-03-06 | 2019-11-11 | Continuous noise tracking-based target speech signal enhancement method and system, and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910168105.0A CN109817234B (en) | 2019-03-06 | 2019-03-06 | Target speech signal enhancement method, system and storage medium based on continuous noise tracking |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109817234A CN109817234A (en) | 2019-05-28 |
| CN109817234B true CN109817234B (en) | 2021-01-26 |
Family
ID=66608268
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910168105.0A Expired - Fee Related CN109817234B (en) | 2019-03-06 | 2019-03-06 | Target speech signal enhancement method, system and storage medium based on continuous noise tracking |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN109817234B (en) |
| WO (1) | WO2020177374A1 (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109817234B (en) * | 2019-03-06 | 2021-01-26 | 哈尔滨工业大学（深圳） | Target speech signal enhancement method, system and storage medium based on continuous noise tracking |
| CN112216303A (en) * | 2019-07-11 | 2021-01-12 | 北京声智科技有限公司 | Voice processing method and device and electronic equipment |
| CN111564154B (en) * | 2020-03-23 | 2023-08-08 | 北京邮电大学 | Method and device for defending against sample attack based on voice enhancement algorithm |
| CN112002339B (en) * | 2020-07-22 | 2024-01-26 | 海尔优家智能科技（北京）有限公司 | Speech noise reduction method and device, computer-readable storage medium and electronic device |
| CN112289337B (en) * | 2020-11-03 | 2023-09-01 | 北京声加科技有限公司 | Method and device for filtering residual noise after machine learning voice enhancement |
| CN113472629A (en) * | 2021-04-19 | 2021-10-01 | 重庆移通学院 | Instant voice pushing system based on Internet of things technology |
| CN113129922B (en) * | 2021-04-21 | 2022-11-08 | 维沃移动通信有限公司 | Voice signal processing method and device |
| CN114242103A (en) * | 2021-12-21 | 2022-03-25 | 游密科技(深圳)有限公司 | Voice noise reduction method and device, computer equipment and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5721694A (en) * | 1994-05-10 | 1998-02-24 | Aura System, Inc. | Non-linear deterministic stochastic filtering method and system |
| CN106971740A (en) * | 2017-03-28 | 2017-07-21 | 吉林大学 | Probability and the sound enhancement method of phase estimation are had based on voice |
| CN107610712A (en) * | 2017-10-18 | 2018-01-19 | 会听声学科技（北京）有限公司 | The improved MMSE of combination and spectrum-subtraction a kind of sound enhancement method |
| CN108711432A (en) * | 2017-04-10 | 2018-10-26 | 中山大学 | A kind of sound enhancement method of the perception gain function of single microphone |
| CN108831499A (en) * | 2018-05-25 | 2018-11-16 | 西南电子技术研究所（中国电子科技集团公司第十研究所） | Utilize the sound enhancement method of voice existing probability |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6289309B1 (en) * | 1998-12-16 | 2001-09-11 | Sarnoff Corporation | Noise spectrum tracking for speech enhancement |
| CN106448692A (en) * | 2016-07-04 | 2017-02-22 | Tcl集团股份有限公司 | RETF reverberation elimination method and system optimized by use of voice existence probability |
| CN108074582B (en) * | 2016-11-10 | 2021-08-06 | 电信科学技术研究院 | Noise suppression signal-to-noise ratio estimation method and user terminal |
| CN108735225A (en) * | 2018-04-28 | 2018-11-02 | 南京邮电大学 | It is a kind of based on human ear masking effect and Bayesian Estimation improvement spectrum subtract method |
| CN109308904A (en) * | 2018-10-22 | 2019-02-05 | 上海声瀚信息科技有限公司 | A kind of array voice enhancement algorithm |
| CN109817234B (en) * | 2019-03-06 | 2021-01-26 | 哈尔滨工业大学（深圳） | Target speech signal enhancement method, system and storage medium based on continuous noise tracking |
-
2019
- 2019-03-06 CN CN201910168105.0A patent/CN109817234B/en not_active Expired - Fee Related
- 2019-11-11 WO PCT/CN2019/117078 patent/WO2020177374A1/en active Application Filing
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5721694A (en) * | 1994-05-10 | 1998-02-24 | Aura System, Inc. | Non-linear deterministic stochastic filtering method and system |
| CN106971740A (en) * | 2017-03-28 | 2017-07-21 | 吉林大学 | Probability and the sound enhancement method of phase estimation are had based on voice |
| CN108711432A (en) * | 2017-04-10 | 2018-10-26 | 中山大学 | A kind of sound enhancement method of the perception gain function of single microphone |
| CN107610712A (en) * | 2017-10-18 | 2018-01-19 | 会听声学科技（北京）有限公司 | The improved MMSE of combination and spectrum-subtraction a kind of sound enhancement method |
| CN108831499A (en) * | 2018-05-25 | 2018-11-16 | 西南电子技术研究所（中国电子科技集团公司第十研究所） | Utilize the sound enhancement method of voice existing probability |
Non-Patent Citations (2)
| Title |
|---|
| 《Minimum mean-square error estimation of discrete Fourier coefficients with generalized gamma priors》;Jan S.Erkelens et al.;《IEEE Transactions on Audio, Speech,and Language Processing》;20070831;第15卷(第6期);第1741-1752页 * |
| 《基于短时谱估计的MMSE语音增强算法研究》;于文慧;《中国优秀硕士学位论文全文数据库 信息科技辑》;20120915(第09期);第29-38页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2020177374A1 (en) | 2020-09-10 |
| CN109817234A (en) | 2019-05-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109817234B (en) | Target speech signal enhancement method, system and storage medium based on continuous noise tracking | |
| Gerkmann et al. | Noise power estimation based on the probability of speech presence | |
| Erkelens et al. | Tracking of nonstationary noise based on data-driven recursive noise power estimation | |
| CN106875938B (en) | Improved nonlinear self-adaptive voice endpoint detection method | |
| US20040230428A1 (en) | Method and apparatus for blind source separation using two sensors | |
| Fischer et al. | Combined single-microphone Wiener and MVDR filtering based on speech interframe correlations and speech presence probability | |
| JP2007047427A (en) | Sound processor | |
| US9875748B2 (en) | Audio signal noise attenuation | |
| Chinaev et al. | Noise-presence-probability-based noise PSD estimation by using DNNs | |
| KR20160116440A (en) | SNR Extimation Apparatus and Method of Voice Recognition System | |
| Hashemgeloogerdi et al. | Joint beamforming and reverberation cancellation using a constrained Kalman filter with multichannel linear prediction | |
| JP2014194437A (en) | Voice processing device, voice processing method and voice processing program | |
| Chinaev et al. | A priori SNR Estimation Using a Generalized Decision Directed Approach. | |
| Tashev et al. | Unified framework for single channel speech enhancement | |
| Yang et al. | Deep learning-based speech presence probability estimation for noise PSD estimation in single-channel speech enhancement | |
| Krawczyk-Becker et al. | Utilizing spectro-temporal correlations for an improved speech presence probability based noise power estimation | |
| Ji et al. | A priori SAP estimator based on the magnitude square coherence for dual-channel microphone system | |
| Chehresa et al. | MMSE speech enhancement based on GMM and solving an over-determined system of equations | |
| Parchami et al. | Model-based estimation of late reverberant spectral variance using modified weighted prediction error method | |
| Nakai et al. | Theoretical analysis of biased MMSE short-time spectral amplitude estimator and its extension to musical-noise-free speech enhancement | |
| CN105788606A (en) | Noise estimation method based on recursive least tracking for sound pickup devices | |
| Fu et al. | Speech presence probability estimation based on integrated time-frequency minimum tracking for speech enhancement in adverse environments | |
| Shen et al. | A priori SNR estimator based on a convex combination of two DD approaches for speech enhancement | |
| Yong et al. | Real time noise suppression in social settings comprising a mixture of non-stationary anc transient noise | |
| Choi et al. | A two-channel minimum mean-square error log-spectral amplitude estimator 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20210126 |