CN111261197B - Real-time speech paragraph tracking method under complex noise scene - Google Patents
Real-time speech paragraph tracking method under complex noise scene Download PDFInfo
- Publication number
- CN111261197B CN111261197B CN202010029721.0A CN202010029721A CN111261197B CN 111261197 B CN111261197 B CN 111261197B CN 202010029721 A CN202010029721 A CN 202010029721A CN 111261197 B CN111261197 B CN 111261197B
- Authority
- CN
- China
- Prior art keywords
- noise
- calculating
- frame
- signal
- voice
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 20
- 238000001228 spectrum Methods 0.000 claims abstract description 28
- 238000001514 detection method Methods 0.000 claims abstract description 12
- 230000004913 activation Effects 0.000 claims abstract description 6
- 230000005236 sound signal Effects 0.000 claims description 7
- 230000000873 masking effect Effects 0.000 claims description 5
- 238000005311 autocorrelation function Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000005314 correlation function Methods 0.000 claims description 3
- 230000003595 spectral effect Effects 0.000 claims description 3
- 230000003213 activating effect Effects 0.000 claims description 2
- 238000009432 framing Methods 0.000 claims description 2
- 101150050759 outI gene Proteins 0.000 claims description 2
- 230000006870 function Effects 0.000 description 7
- 230000000694 effects Effects 0.000 description 4
- 230000001052 transient effect Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000001629 suppression Effects 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
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/78—Detection of presence or absence of voice signals
- G10L25/84—Detection of presence or absence of voice signals for discriminating voice from noise
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L21/00—Speech or voice signal processing techniques to produce another audible or non-audible signal, e.g. visual or tactile, in order to modify its quality or its intelligibility
- G10L21/02—Speech enhancement, e.g. noise reduction or echo cancellation
- G10L21/0208—Noise filtering
- G10L21/0216—Noise filtering characterised by the method used for estimating noise
-
- 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
- G10L25/21—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters the extracted parameters being power information
-
- 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
- G10L25/27—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the analysis technique
-
- 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
-
- 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
- G10L25/93—Discriminating between voiced and unvoiced parts of speech signals
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)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
The invention discloses a real-time voice paragraph tracking method under a complex noise scene, which comprises the following steps: A. pre-treating; B. calculating a discrete Fourier transform coefficient of an input audio frame, and calculating the power of initial noise, namely calculating the arithmetic mean value of a Fourier transform amplitude spectrum, assuming that a previous frame is a noise frame; assuming that the data after the frame is a signal with noise, calculating the power of the signal with noise; D. calculating the posterior signal-to-noise ratio; E. calculating a priori signal-to-noise ratio; F. voice activation detection; G. updating a noise spectrum; H. and calculating a gain coefficient, estimating the spectrum attribute of stationary noise in a scene by using paragraph noise between the speech segments, and designing a gain function to enhance the voice and inhibit the stationary noise. And performing voiced sound detection on the basis, tracking the speech paragraphs, and shielding various noises among the speech paragraphs. Therefore, the accuracy of voice detection can be improved, the noise of voice segment superposition can be inhibited, and the noise between the voice segments influencing the listening feeling can be thoroughly shielded.
Description
Technical Field
The invention relates to the technical field of voice processing, in particular to a real-time voice paragraph tracking method in a complex noise scene.
Background
Engineering in the field of speech signal processing is to be faced with complex noise scenarios including stationary noise, transient noise, time-varying noise, and strong noise, etc., which have different statistical properties. When the near-talking sound pickup equipment is used for voice collection, voice communication and voice recognition, background noise is easily picked up by the microphone, direct influence is caused to the voice communication from the listening aspect, and the performance of processing modules such as rear-end voice recognition and the like can be further influenced. In a complex noise scene, steady-state noise mixed in voice is inhibited, other types of noise mixed among voice paragraphs are shielded, and pure voice paragraphs are obtained by tracking, so that the hearing of voice communication can be effectively improved, and the performance of a back-end processing module such as voice recognition and the like is improved. The speech tracking under the single noise scene with the statistical characteristics is relatively easy to process, and the speech paragraph tracking under the complex noise scene is a difficult problem.
Disclosure of Invention
The present invention is directed to provide a real-time speech paragraph tracking method in a complex noise scene, so as to solve the problems mentioned in the background art.
In order to achieve the purpose, the invention provides the following technical scheme:
a real-time speech paragraph tracking method under a complex noise scene is characterized by comprising the following steps:
A. pretreatment: framing and windowing an input audio signal; taking 16ms data as a frame x i (n), wherein i is a frame number;
B. computing input audio frames
Discrete fourier transform coefficient Y of i (ω k ) Where k is the index of the spectral component;
C. assuming the previous L frames as noise frames, calculating the power of the initial noise, i.e. calculating
An arithmetic mean of the fourier transform magnitude spectrum; assuming the data after L frames as a noise signal, calculating the power of the noise signal
D. Calculating the posterior signal-to-noise ratio
E. Calculating the prior signal-to-noise ratio
F. Voice activation detection;
G. updating a noise spectrum;
H. calculating a gain coefficient;
I. signal reconstruction: calculating the amplitude spectrum and the power spectrum of the enhanced voice of the current frame, and performing inverse Fourier transform on the spectrum of the enhanced voice to obtain a reconstructed signal;
J. calculating out
Is self-correlation function of
Wherein r is t (tau) is an autocorrelation function with a delay of tau, N is a window length and N is greater than or equal to 1 and less than or equal to N;
K. calculating a difference function:
and (3) calculating:
l, judging the voiced sound according to the following conditions: p =1-d' (τ) is calculated, p characterizing the probability that a fundamental frequency component is clearly contained in a frame of speech. Since d' (τ) has a value in the range of [0,1 ]]Then p is in the value range of [0,1 ]](ii) a With p th As a threshold, is larger than p th The speech frame of (1) is reserved as voiced;
m, unvoiced sound compensation and noise masking.
As a further scheme of the invention: in the step A, the input audio signal is framed and windowed, and the window function is a Hamming window:
as a further scheme of the invention: and the step F is specifically to carry out voice activation detection on the input frame and select out the noise frame. According to the posterior signal-to-noise ratio gamma k And a priori signal-to-noise ratio
And solving a judgment parameter v for activating voice detection, judging as voice if v is greater than a judgment threshold eta, and judging as noise if v is less than eta, so as to update a noise spectrum. The calculation method of the decision parameter v is as follows.
As a further scheme of the invention: the step G is specifically as follows: after the noise frame is selected, the noise spectrum is updated according to the following formula:
as a further scheme of the invention: the step H is specifically as follows: calculating the weighting coefficient of the amplitude spectrum of the current frame according to the posterior signal-to-noise ratio and the prior signal-to-noise ratio:
as a further scheme of the invention: the function established in the step I is as follows:
as a further scheme of the invention: in the step M, if a certain frame is determined to be voiced and a signal frame within 400 milliseconds after the certain frame is determined to be not voiced, compensation is performed, that is, the signal frame is directly output without being processed; and (4) masking the non-voiced sound frame which does not meet the compensation condition, namely performing amplitude limiting processing and outputting.
Compared with the prior art, the invention has the beneficial effects that: the invention completely tracks the voice paragraph, shields the noise outside the voice paragraph, inhibits the noise superposed on the voice, and enhances the listening effect of the voice.
Drawings
FIG. 1 is a time domain waveform of an audio signal with stationary noise and transient noise superimposed on the speech and a noise peak exceeding 60 dB;
FIG. 2 is a time domain waveform of the signal of FIG. 1 after being processed by the present embodiment;
FIG. 3 is a time domain waveform of an audio signal with stationary noise and transient noise superimposed on the speech and a noise peak exceeding 110 dB;
FIG. 4 is a time domain waveform of the signal of FIG. 3 after being processed by the present invention;
fig. 5 is a flowchart of the method of the present embodiment.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Referring to fig. 1-5, example 1: in an embodiment of the present invention, a real-time speech paragraph tracking method in a complex noise scene includes the following steps:
A. and (4) preprocessing. The input audio signal is framed and windowed. Taking 16ms (256 samples) of data as a frame x i (n), where i is the frame number. Windowing is performed on the data, and the window function is a Hamming window:
B. computing input audio frames
Discrete fourier transform coefficient Y of i (ω k ) Where k is the index of the spectral component:
Y i (ω k )＝Y k exp(jθ y (k))
C. assuming the first L frames as noise frames, calculating the power of the initial noise, i.e. calculating
Arithmetic mean of fourier transform magnitude spectrum:
|Y i (ω k )| 2 ；
D. Calculating the posterior signal-to-noise ratio gamma k ＝|Y i (ω k )| 2 /λ d (k)；
E. Calculating the prior signal-to-noise ratio
F. Voice activity detection. Since the noise may be stationary for a short time, the noise spectrum needs to be updated in real time to ensure the effect of noise suppression. And carrying out voice activation detection on the input frame, and selecting out a noise frame. According to the posterior signal-to-noise ratio gamma k And a priori signal to noise ratio
A decision parameter v is derived that activates speech detection. If v is larger than the decision threshold eta, the voice is judged, and if v is smaller than eta, the voice is judged as noise, and the noise spectrum is updated. The calculation method of the decision parameter v comprises the following steps:
G. and updating the noise spectrum. After the noise frame is selected, the noise spectrum is updated according to the following formula:
H. a gain factor is calculated. Calculating a weighting coefficient of the magnitude spectrum of the current frame according to the posterior signal-to-noise ratio and the prior signal-to-noise ratio:
wherein exp (-) is an exponential function with a natural constant e as a base, and expint (-) is an exponential integration function with the natural constant e as a base.
I. The signal is reconstructed. Calculating the amplitude spectrum and the power spectrum of the enhanced voice of the current frame, and performing inverse Fourier transform on the frequency spectrum of the enhanced voice to obtain a reconstructed signal:
J. computing
Is self-correlation function of
Wherein r is t (tau) is an autocorrelation function with a delay of tau, N is a window length and N is greater than or equal to 1 and less than or equal to N;
K. calculating a difference function:
and (3) calculating:
l, judging the voiced sound according to the following conditions:
p =1-d' (τ) is calculated, p characterizing the probability that a fundamental frequency component is clearly contained in a frame of speech. Due to the range of d' (τ)Is enclosed as [0,1 ]]Then p has a value in the range of [0,1 ]]. With p th As a threshold, is larger than p th The speech frame of (1) is reserved as voiced;
m, unvoiced sound compensation and noise masking. If a certain frame is judged to be voiced and a signal frame within 400 milliseconds later is not voiced, compensation is carried out, namely the signal frame is directly output without being processed; and (4) masking the non-voiced sound frame which does not meet the compensation condition, namely performing amplitude limiting processing and outputting.
Fig. 3 and fig. 5 are audio time domain waveforms processed by the method of the present invention, and it can be seen from comparing with the original waveforms that, under the complex noise background, the method completely tracks the speech paragraphs, masks the noise outside the speech paragraphs, and also plays a role in suppressing the noise superimposed on the speech, thereby enhancing the listening effect of the speech itself.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential attributes thereof. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned.
Furthermore, it should be understood that although the present description refers to embodiments, not every embodiment may contain only a single embodiment, and such description is for clarity only, and those skilled in the art should integrate the description, and the embodiments may be combined as appropriate to form other embodiments understood by those skilled in the art.
Claims (5)
1. A real-time speech paragraph tracking method under a complex noise scene is characterized by comprising the following steps:
A. pretreatment: framing and windowing an input audio signal; taking 16ms data as a frame xi (n), wherein i is a frame number;
B. computing input audio frames
Discrete fourier transform coefficient Yi (ω) k ) Where k is the index of the spectral component;
C. assuming the previous L frames as noise frames, calculating the power of the initial noise, i.e. calculating
An arithmetic mean of the fourier transform magnitude spectrum; assuming that the data after the L frame is a signal with noise, the power | Yi (ω) of the signal with noise is calculated k )| 2 ；
D. Calculating the posterior signal-to-noise ratio gamma k ＝|Yi(ω k )| 2 /λ d (k)；
E. Calculating a priori signal-to-noise ratio
F. Voice activation detection; the step F specifically comprises the following steps: carrying out voice activation detection on an input frame, and selecting a noise frame; according to the posterior signal-to-noise ratio gamma k And a priori signal to noise ratio
Solving a judgment parameter v for activating voice detection, judging voice if v is greater than a judgment threshold eta, and judging noise if v is less than eta, wherein the judgment parameter v is used for updating a noise spectrum; the calculation method of the decision parameter v comprises the following steps:
G. updating a noise spectrum; the step G is specifically as follows: after the noise frame is selected, the noise spectrum is updated according to the following formula:
H. calculating a gain coefficient;
I. signal reconstruction: calculating the amplitude spectrum and the power spectrum of the enhanced voice of the current frame, and performing inverse Fourier transform on the spectrum of the enhanced voice to obtain a reconstructed signal;
J. calculating out
Is self-correlation function of
Wherein r is t (tau) is an autocorrelation function with a delay of tau, N is a window length and N is greater than or equal to 1 and less than or equal to N;
K. calculating a difference function:
and (3) calculating:
l, judging the voiced sound according to the following conditions: calculating p =1-d' (τ), wherein p represents the probability that a certain fundamental frequency component is obviously contained in a frame of voice; since d' (τ) has a value in the range of [0,1 ]]Then p has a value in the range of [0,1 ]](ii) a With p th As a threshold, is larger than p th The speech frame of (1) is reserved as voiced;
m, unvoiced sound compensation and noise masking.
2. The method according to claim 1, wherein the real-time speech paragraph tracking method under complex noise scene,
in the step A, the input audio signal is framed and windowed, and the window function is a Hamming window:
3. the method according to claim 1, wherein the real-time speech paragraph tracking method under the complex noise scene,
the step H is specifically as follows: calculating the weighting coefficient of the amplitude spectrum of the current frame according to the posterior signal-to-noise ratio and the prior signal-to-noise ratio:
4. the method according to claim 1, wherein the real-time speech paragraph tracking method under the complex noise scene,
the function established in the step I is as follows:
5. the method according to claim 1, wherein the real-time speech paragraph tracking method under the complex noise scene,
in the step M, if a certain frame is judged to be voiced and a signal frame within 400 milliseconds after the certain frame is judged to be not voiced, compensation is carried out, namely the signal frame is directly output without being processed; and (4) shielding the non-voiced sound frame which does not meet the compensation condition, namely performing amplitude limiting processing and outputting the non-voiced sound frame.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010029721.0A CN111261197B (en) | 2020-01-13 | 2020-01-13 | Real-time speech paragraph tracking method under complex noise scene |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010029721.0A CN111261197B (en) | 2020-01-13 | 2020-01-13 | Real-time speech paragraph tracking method under complex noise scene |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111261197A CN111261197A (en) | 2020-06-09 |
| CN111261197B true CN111261197B (en) | 2022-11-25 |
Family
ID=70950451
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010029721.0A Active CN111261197B (en) | 2020-01-13 | 2020-01-13 | Real-time speech paragraph tracking method under complex noise scene |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111261197B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1130952A (en) * | 1993-09-14 | 1996-09-11 | 英国电讯公司 | Voice activity detector |
| CN105845150A (en) * | 2016-03-21 | 2016-08-10 | 福州瑞芯微电子股份有限公司 | Voice enhancement method and system adopting cepstrum to correct |
| CN107452363A (en) * | 2017-07-03 | 2017-12-08 | 福建天泉教育科技有限公司 | Musical instrument tuner method and system |
| CN108831504A (en) * | 2018-06-13 | 2018-11-16 | 西安蜂语信息科技有限公司 | Determination method, apparatus, computer equipment and the storage medium of pitch period |
| 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 |
| CN110322898A (en) * | 2019-05-28 | 2019-10-11 | 平安科技（深圳）有限公司 | Vagitus detection method, device and computer readable storage medium |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101136199B (en) * | 2006-08-30 | 2011-09-07 | 纽昂斯通讯公司 | Voice data processing method and equipment |
| FR3014237B1 (en) * | 2013-12-02 | 2016-01-08 | Adeunis R F | METHOD OF DETECTING THE VOICE |
-
2020
- 2020-01-13 CN CN202010029721.0A patent/CN111261197B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1130952A (en) * | 1993-09-14 | 1996-09-11 | 英国电讯公司 | Voice activity detector |
| CN105845150A (en) * | 2016-03-21 | 2016-08-10 | 福州瑞芯微电子股份有限公司 | Voice enhancement method and system adopting cepstrum to correct |
| CN107452363A (en) * | 2017-07-03 | 2017-12-08 | 福建天泉教育科技有限公司 | Musical instrument tuner method and system |
| 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 |
| CN108831504A (en) * | 2018-06-13 | 2018-11-16 | 西安蜂语信息科技有限公司 | Determination method, apparatus, computer equipment and the storage medium of pitch period |
| CN110322898A (en) * | 2019-05-28 | 2019-10-11 | 平安科技（深圳）有限公司 | Vagitus detection method, device and computer readable storage medium |
Non-Patent Citations (2)
| Title |
|---|
| 一种基于Hilbert-Huang变换的基音周期检测新方法;杨志华等;《计算机学报》;20060112(第01期);全文 * |
| 基于浊音语音谐波谱子带加权重建的抗噪声说话人识别;曾毓敏等;《东南大学学报(自然科学版)》;20081120(第06期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111261197A (en) | 2020-06-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP2360685B1 (en) | Noise suppression | |
| Nakatani et al. | Robust and accurate fundamental frequency estimation based on dominant harmonic components | |
| CN108831499A (en) | Utilize the sound enhancement method of voice existing probability | |
| EP1065656B1 (en) | Method for reducing noise in an input speech signal | |
| US20070255535A1 (en) | Method of Processing a Noisy Sound Signal and Device for Implementing Said Method | |
| Verteletskaya et al. | Noise reduction based on modified spectral subtraction method | |
| CN113539285B (en) | Audio signal noise reduction method, electronic device and storage medium | |
| Wolfe et al. | Towards a perceptually optimal spectral amplitude estimator for audio signal enhancement | |
| CN114694670A (en) | Multi-task network-based microphone array speech enhancement system and method | |
| CN103295580A (en) | Method and device for suppressing noise of voice signals | |
| CN112185405B (en) | Bone conduction voice enhancement method based on differential operation and combined dictionary learning | |
| Ambikairajah et al. | Wavelet transform-based speech enhancement | |
| CN111261197B (en) | Real-time speech paragraph tracking method under complex noise scene | |
| Cao et al. | Research on noise reduction algorithm based on combination of LMS filter and spectral subtraction | |
| Bahadur et al. | Performance measurement of a hybrid speech enhancement technique | |
| Hamid et al. | Speech enhancement using EMD based adaptive soft-thresholding (EMD-ADT) | |
| Rao et al. | Speech enhancement using sub-band cross-correlation compensated Wiener filter combined with harmonic regeneration | |
| Graupe et al. | Blind adaptive filtering of speech from noise of unknown spectrum using a virtual feedback configuration | |
| Srinivas et al. | A classification-based non-local means adaptive filtering for speech enhancement and its FPGA prototype | |
| Islam et al. | Speech enhancement in adverse environments based on non-stationary noise-driven spectral subtraction and snr-dependent phase compensation | |
| CN112750451A (en) | Noise reduction method for improving voice listening feeling | |
| Upadhyay et al. | Recursive noise estimation-based Wiener filtering for monaural speech enhancement | |
| Zengyuan et al. | A speech denoising algorithm based on harmonic regeneration | |
| CN117995215B (en) | Voice signal processing method and device, computer equipment and storage medium | |
| Gbadamosi et al. | Development of non-parametric noise reduction algorithm for GSM voice signal |
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 |