CN100426378C - Dynamic noise eliminating method and digital filter - Google Patents
Dynamic noise eliminating method and digital filter Download PDFInfo
- Publication number
- CN100426378C CN100426378C CNB2005100891488A CN200510089148A CN100426378C CN 100426378 C CN100426378 C CN 100426378C CN B2005100891488 A CNB2005100891488 A CN B2005100891488A CN 200510089148 A CN200510089148 A CN 200510089148A CN 100426378 C CN100426378 C CN 100426378C
- Authority
- CN
- China
- Prior art keywords
- frequency
- signal
- domain
- energy
- time domain
- 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 38
- 238000006243 chemical reaction Methods 0.000 claims abstract description 16
- 230000009466 transformation Effects 0.000 claims abstract description 7
- 230000008569 process Effects 0.000 claims description 14
- 238000004088 simulation Methods 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 5
- 238000005070 sampling Methods 0.000 abstract description 10
- 230000005236 sound signal Effects 0.000 abstract description 3
- 230000006870 function Effects 0.000 description 16
- 238000004422 calculation algorithm Methods 0.000 description 5
- 238000013016 damping Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 239000012141 concentrate Substances 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 238000009413 insulation Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000007664 blowing Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 238000005562 fading Methods 0.000 description 1
- 210000004704 glottis Anatomy 0.000 description 1
- 210000004072 lung Anatomy 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000010183 spectrum analysis Methods 0.000 description 1
Images
Abstract
The present invention discloses a dynamic noise eliminating method and a digital filter. The method of the present invention comprises that: a, windows are added to time domain sampling sound signals, and Fourier transformation is used for converting scattered time domain signals in the window into frequency domain digital signals; B, converted frequency domain digital signals can carry out energy attenuating treatment; C, the Fourier transformation is used for converting frequency domain digital signals after energy attenuating treatment into time domain scattered signals, and then, sound signals are recovered. The digital filter provided by the present invention comprises a die / digital conversion unit, a windowing adding processing unit, a time domain and frequency domain conversion unit, a signal and noise processing unit, a frequency domain and time domain conversion unit and a digital/ die conversion unit. The present invention can realize noise treatment and can simultaneously ensure the quantity of sound and enhance users' feeling.
Description
Technical field
The present invention relates to the communications field, relate in particular to a kind of voice signal dynamic noise eliminating method and corresponding digital wave filter based on the IP agreement.
Background technology
Voice are that air-flow produces by the excitation of glottis from lung, frequency mainly at 300Hz between the 3400Hz, and 700Hz is to particularly concentrated between 2000Hz.The mechanism complexity that noise produces, various informative, the noise of introducing when mainly being divided into simulating signal/digital signal (A/D) conversion of background noise, noise that Device element produces and sound signal.Background noise is produced by surrounding environment, as the machine roar of machine room, and the automobile of the outdoor highway sound of blowing a whistle, raindrop sound of rainy day or the like.The noise that Device element produces is because device self is normally inevitable, is exactly very big noise source as 50Hz (or 60Hz) power supply, and in addition, sound card device, microphone all may produce noise.It is that continuous sound wave analog passband signal over-sampling is being become discrete digital signal that noise is introduced in the A/D conversion, can produce noise owing to the loss of precision when digital signal is quantized.Common speech sample rate has two kinds of 8KHz and 16KHz, and the preceding a kind of following noise of 4KHz is collected together then a kind ofly then gathered the noise below the 8KHz together.From spectral characteristic, the frequency of noise can run through the people audibility range (20Hz---20000Hz), and, mainly be distributed in low frequency part from the angle of energy.
Often occur when carrying out voice collecting mixing various noises, make aphthenxia clear with PC, even situation about being capped.In order to reduce the influence of noise, improve voice quality, can be by taking certain measure: as select better sound pick-up outfit for use, better sound insulation environment is provided.Yet, the higher cost input of sound pick-up outfit meaning of using instead, the sound insulation environment that daily life is difficult to find.
Means by digital signal processing reduce the influence of noise and show very big advantage.At present, the normal influence of adopting the technology of controlling microphone volume automatically to reduce noise in VOIP (Voice over IP, the IP-based audio call) field.Its ultimate principle is the short-time average energy of noise in the performance analysis audio stream, if energy is higher, then reduces microphone automatically and gathers volume.This technology can reduce the influence of noise to the sense of hearing really, yet it has reduced the volume of speech simultaneously, causes in some cases speech not hear.
Summary of the invention
The invention provides a kind of dynamic noise eliminating method and NEURAL DISCHARGE BY DIGITAL FILTER thereof, when eliminating dynamic noise, be subjected to the environmental baseline restriction maybe can not guarantee the impregnable problem of speech volume in the prior art in order to solve.
Dynamic noise eliminating method provided by the invention comprises:
A, to the windowing of time domain sample voice signal, and adopt Fourier transform that discrete time signal in the window is converted to the frequency domain digital signal;
B, the frequency domain digital signal after the conversion carried out energy attenuation handle:
Signal energy decay 0dB to the voice signal frequency range;
Low frequency signal and higher frequency signal energy near the voice signal frequency are decayed by linear relationship;
To the signal energy of all the other frequencies decay one set greater than 0 energy value.
After C, the frequency domain digital signal after will carrying out energy attenuation and handling adopt inverse fourier transform to be converted to the time domain discrete signal, restore out voice signal.
According to said method of the present invention, described Fourier transform is a fast fourier transform; Described inverse fourier transform is an invert fast fourier transformation.
According to said method of the present invention, also comprise the step of the frequency domain digital signal after the energy attenuation processing being carried out the window function calculation process before the described step C.
According to said method of the present invention, described window function is a Hamming window function.
According to said method of the present invention, first frequency, second frequency, the 3rd frequency and the 4th frequency are set from low to high by signal frequency; Described second frequency is a voice signal low frequency frequency, and described the 3rd frequency is a voice signal high frequency frequency;
With signal frequency greater than described second frequency and less than the signal energy of described the 3rd frequency decay 0dB;
With signal frequency more than or equal to described first frequency and smaller or equal to described second frequency, and signal frequency more than or equal to described the 3rd frequency and smaller or equal to the signal energy of described the 4th frequency by the linear relationship decay;
With signal frequency less than described first frequency and greater than the signal energy of described the 4th frequency decay one set greater than 0 energy value.
According to said method of the present invention, described first frequency is 300Hz, and second frequency is 700Hz, and the 3rd frequency is 2000Hz, and the 4th frequency is 3400Hz.
According to said method of the present invention, in the described steps A, the size of window is 256 or 512.
The present invention provides a kind of digital filter in addition, comprising:
A/D converter: the time domain simulation voice signal is converted to the time domain discrete digital signal;
Windowing process unit: the time domain discrete digital signal after the A/D converter conversion is carried out windowing process;
Time-domain and frequency-domain converting unit: adopt Fourier transform to be converted to the frequency domain digital signal discrete time-domain digital signal in the window;
Signal noise processing unit: at frequency domain digital signal is carried out energy attenuation and handle;
The time domain frequency domain converting unit: the frequency domain digital signal that will carry out after energy attenuation is handled adopts inverse fourier transform to be reduced to the time-domain digital signal;
The D/A switch unit: with the time-domain digital conversion of signals is time domain simulation signal and output.
According to said apparatus of the present invention, also include the window function arithmetic element, be connected between described signal noise processing unit and the time domain frequency domain converting unit, after the frequency domain digital signal after will carrying out energy attenuation and handling carries out the window function computing, export to described time domain frequency domain converting unit again.
According to said apparatus of the present invention, be provided with first frequency, second frequency, the 3rd frequency and the 4th frequency from low to high by signal frequency; Described second frequency is a voice signal low frequency frequency, and described the 3rd frequency is a voice signal high frequency frequency; Described signal noise processing unit comprises first module and Unit second;
Described first module to signal frequency less than first frequency and greater than the signal energy of the 4th frequency decay one set greater than 0 energy value;
Described Unit second to signal frequency more than or equal to described first frequency and smaller or equal to described second frequency, and signal frequency more than or equal to described the 3rd frequency and smaller or equal to the signal energy of described the 4th frequency by linear attenuation.
According to said apparatus of the present invention, described first frequency is 300Hz, and second frequency is 700Hz, and the 3rd frequency is 2000Hz, and the 4th frequency is 3400Hz.
According to said apparatus of the present invention, described window function is a Hamming window function.
Beneficial effect of the present invention is as follows:
(1) the present invention is when handling the dynamic noise signal, only the decay energy of low frequency signal and the energy of high-frequency signal, keep the signal energy in the voice signal frequency range as far as possible, thereby be implemented in when carrying out noise treatment, guarantee that the volume of voice is unaffected.
(2) the present invention carries out staging treating with voice signal from low to high by frequency, for directly decay a certain amount of signal energy or by the linear relationship decay of the lower signal of frequency and frequency higher signal, realizes simple.
(3) the present invention does not make the signal energy attenuation processing to the most concentrated intermediate bands of speech frequency, guarantees that speech volume is not lowered, and improves voice signal and handles quality.
Description of drawings
Fig. 1 is the process flow diagram of dynamic noise eliminating method of the present invention;
Fig. 2 is the synoptic diagram of signal frequency range in the embodiment of the invention;
Fig. 3 is the structural representation of digital filter of the present invention.
Embodiment
Referring to Fig. 1, the process flow diagram for dynamic noise eliminating method of the present invention specifically comprises:
1, the time domain discrete signal behind the speech simulation signal sampling is carried out windowing process, be about on the time domain continuous several division of signal in groups so that realize the conversion of voice signal from the time domain discrete signal to the frequency domain digital signal.The data that comprise in the window are many more, and spectrum analysis is just accurate more, simultaneously complexity, time-delay also can be big more.Usually, when being applied to the VOIP field, window size (data volume that comprises) is 256 or 512.
2, adopt fast fourier algorithm, with the time domain discrete signal S after the windowing
The n time domainBe converted to frequency domain digital signal S
The n frequency domainThe quantity of frequency domain digital signal is identical with the window size.For example, if the size of window is 256, then every window time domain discrete signal is converted into 256 frequency domain digital signals; If the size of window is 512, then every window time domain discrete signal is converted into 512 frequency domain digital signals.
The frequency range H of frequency domain digital signal
wRelevant with sampling rate with the window size, the computing formula of frequency range is:
Wherein, Hsr is a sampling rate, and N is the window size.
Common speech sample rate has two kinds of 8KHz and 16KHz.For example, when sampling rate is 8KHz, the size of window is 256 o'clock, frequency range H
wBe 15.625Hz; Same sampling rate, and the size of window is 512 o'clock, frequency range H
wBe 7.8125Hz.When sampling rate is 16KHz, the size of window is 256 o'clock, frequency range H
wBe 31.25Hz; Same sampling rate, and the size of window is 512 o'clock, frequency range H
wBe 15.625Hz.
3, the energy of decay low frequency and high-frequency signal.
As shown in Figure 2, frequency values f1, f2, f3, f4 are divided into 5 zones with signal frequency, and the signal that frequency is lower than f2 (containing f2) is a low frequency signal, and the signal that frequency is higher than f3 (containing f3) is a high-frequency signal.F1 is 2 zones with the low frequency signal area dividing, and f4 is divided into 2 zones with region of high frequency signals.In the zone of above-mentioned division, frequency is lower than the zone of f1 and zone that frequency is higher than f4 is tangible noise signal frequency range, to the energy of this regional a certain setting of frequency domain digital signal attenuation; To the f1 frequency domain digital signal regional to f2 (comprising f1 and f2) zone and f3 to f4 (comprising f3 and f4) by linear relationship its energy of decaying.F2 is the voice signal frequency range to the f3 zone, and is therefore to the frequency domain digital signal attenuation 0dB (promptly signal energy not being decayed) of this frequency range, unaffected with the volume that guarantees voice signal.
Because the frequency of voice signal mainly concentrates on 300Hz between the 3400Hz, and 700Hz is to particularly concentrated between 2000Hz, so the preferable value of f1, f2, f3, f4 is followed successively by 300Hz, 700Hz, 2000Hz, 3400Hz.
For example, for being lower than 300Hz and the frequency domain digital signal attenuation 30dB that is higher than 3400Hz, computing formula is:
S
n'
Frequency domain=dS
The n frequency domain(wherein,
NH
w＜=300 and nH
w＞=3400)
For example: when sample frequency is 8KHz, the windowing size is 256 o'clock, frequency range H
wBe 15.625Hz, frequency is nH
wThe frequency domain digital signal by above-mentioned formula damping capacity, obtain:
Work as n=12, i.e. nH
wDuring for 187.5Hz,
Work as n=256, i.e. nH
wDuring for 4000Hz,
Press linear attenuation energy at 300Hz to the frequency domain digital signal that (comprises 300Hz and 700Hz) between the 700Hz for frequency, computing formula is:
(wherein,
NH
w＞=300 and nH
w＜=700)
For example: when sample frequency is 8KHz, the windowing size is 256 o'clock, frequency range H
wBe 15.625Hz, frequency is nH
wThe frequency domain digital signal by above-mentioned formula damping capacity, obtain:
Work as n=20, i.e. nH
wDuring=312.5Hz,
Work as n=32, i.e. nH
wDuring=500Hz,
Work as n=44, i.e. nH
wDuring=687.5Hz,
For frequency at 2000Hz to the frequency domain digital signal wiring damping capacity that (comprises 2000Hz and 3400Hz) between the 3400Hz, computing formula is:
(wherein,
NH
w＞=2000 and nH
w＜=3400)
For example: when sample frequency is 8KHz, the windowing size is 256 o'clock, frequency range H
wBe 15.625Hz, frequency is nH
wThe frequency domain digital signal by above-mentioned formula damping capacity, obtain:
Work as n=128, i.e. nH
wDuring=2000Hz,
Work as n=172, i.e. nH
wDuring=2687.5Hz,
Work as n=217, i.e. nH
wDuring=3390.625Hz,
4, the frequency domain digital signal after the energy attenuation processing is carried out the window function calculation process.In order to make the data processing between window and the window can not produce excessive variation, select the Hamming window function that suppresses good for use, the formula of Hamming window function is:
(wherein, N is the window size, and n is the position of digital signal in the window)
The processing formula of Hamming window function is:
S
n"
Frequency domain=W (n) S '
The n frequency domain
5, will adopt invert fast fourier transformation to be converted to the time domain discrete signal through the frequency domain digital signal after the window function calculation process, be the time domain simulation signal with the time domain discrete conversion of signals again, restores voice signal.Be about to frequency domain digital signal S
n"
Frequency domainBe converted to time domain simulation signal s '
The n time domainTime domain simulation signal after handle this moment then is a signal of having eliminated noise effect.
According to said method, digital filter provided by the invention as shown in Figure 3, comprising:
A/D converter: dynamic time domain simulation voice signal is converted to the time domain discrete digital signal;
Windowing process unit: the time domain discrete digital signal after the A/D converter conversion is carried out windowing process; Usually, when being applied to the VOIP field, added window size is 256 or 512;
Time-domain and frequency-domain converting unit: adopt fast fourier transform to be converted to the frequency domain digital signal discrete time-domain digital signal in the window;
Signal noise processing unit: at frequency domain digital signal is carried out energy attenuation and handle;
Window function arithmetic element: the frequency domain digital signal that carries out after energy attenuation is handled is carried out the window function computing; In order to make the data processing between window and the window can not produce excessive variation, select the Hamming window function that suppresses good for use;
The time domain frequency domain converting unit: the frequency domain digital signal that will carry out after the window function calculation process adopts invert fast fourier transformation to be converted to the time-domain digital signal;
D/A switch unit: will be time domain simulation signal and output through the time-domain digital conversion of signals behind the invert fast fourier transformation.
Described digital filter is provided with first frequency, second frequency, the 3rd frequency and the 4th frequency from low to high by signal frequency, and described signal noise processing unit comprises first module and Unit second; Described first module to signal frequency less than first frequency and the energy value set greater than the signal energy of the 4th frequency decay one; Described Unit second to signal frequency more than or equal to described first frequency and smaller or equal to described second frequency, and signal frequency more than or equal to described the 3rd frequency and smaller or equal to the signal energy of described the 4th frequency by linear attenuation.
The frequency of considering voice signal mainly concentrates on 300Hz between the 3400Hz, and 700Hz is to particularly concentrated between 2000Hz, and therefore described first frequency is 300Hz, and second frequency is 700Hz, and the 3rd frequency is 2000Hz, and the 4th frequency is 3400Hz.
The first module of described noise treatment unit to frequency less than 300Hz and frequency greater than the signal attenuation 30dB of 3400Hz, adopt following algorithm:
S
n'
Frequency domain=dS
The n frequency domain(wherein,
NH
w＜=300 and nH
w＞=3400)
Unit second of described noise treatment unit is pressed the linear attenuation energy to signal frequency more than or equal to 300Hz and smaller or equal to the signal of 700Hz, adopts following algorithm:
(wherein,
NH
w＞=300 and nH
w＜=700)
Unit second of described noise treatment unit is pressed the linear attenuation energy to signal frequency more than or equal to 2000Hz and smaller or equal to the signal of 3400Hz, adopts following algorithm:
(wherein,
NH
w＞=2000 and nH
w＜=3400)
H in the above-mentioned formula
wBe frequency range, computing formula is as follows:
Description by above method and apparatus as can be known, for less than 300Hz or greater than the noise signal of the 3400Hz 30dB that decays without exception; Signal between 300Hz and the 700Hz is pressed linear attenuation, and more near 700Hz, attenuation degree is littler; Signal to 2000Hz and 3400Hz is pressed linear attenuation, and more near 2000Hz, attenuation degree is littler.This shows, when noise signal is handled, frequency of fadings is at the energy of the signal below the 700Hz or more than the 2000Hz, and keep the voice signal frequency ratio as far as possible than the energy of the 300Hz that concentrates to signal between the 3400Hz, thereby be implemented in when carrying out noise treatment, guarantee the volume of voice, improved user's impression.
In addition, noise signal is handled the employing linear attenuation, algorithm is simple, is easy to realize.
Obviously, those skilled in the art can carry out various changes and modification to the present invention and not break away from the spirit and scope of the present invention.Like this, if of the present invention these are revised and modification belongs within the scope of claim of the present invention and equivalent technologies thereof, then the present invention also is intended to comprise these changes and modification interior.
Claims (11)
1, a kind of dynamic noise eliminating method is characterized in that, comprising:
A, to the windowing of time domain sample voice signal, and adopt Fourier transform that discrete time signal in the window is converted to the frequency domain digital signal;
B, the frequency domain digital signal after the conversion is carried out energy attenuation handle: to the signal energy decay 0dB of voice signal frequency range; Low frequency signal and higher frequency signal energy near the voice signal frequency are decayed by linear relationship; To the signal energy of all the other frequencies decay one set greater than 0 energy value;
After C, the frequency domain digital signal after will carrying out energy attenuation and handling adopt inverse fourier transform to be converted to the time domain discrete signal, restore out voice signal.
2, the method for claim 1 is characterized in that, described Fourier transform is a fast fourier transform; Described inverse fourier transform is an invert fast fourier transformation.
3, the method for claim 1 is characterized in that, also comprises the step of the frequency domain digital signal after the energy attenuation processing being carried out the window function calculation process before the described step C.
4, method as claimed in claim 3 is characterized in that, described window function is a Hamming window function.
5, the method for claim 1 is characterized in that, by signal frequency first frequency, second frequency, the 3rd frequency and the 4th frequency is set from low to high; Described second frequency is a voice signal low frequency frequency, and described the 3rd frequency is a voice signal high frequency frequency;
With signal frequency greater than described second frequency and less than the signal energy of described the 3rd frequency decay 0dB;
With signal frequency more than or equal to described first frequency and smaller or equal to described second frequency, and signal frequency more than or equal to described the 3rd frequency and smaller or equal to the signal energy of described the 4th frequency by the linear relationship decay;
With signal frequency less than described first frequency and greater than the signal energy of described the 4th frequency decay one set greater than 0 energy value.
6, method as claimed in claim 5 is characterized in that, described first frequency is 300Hz, and second frequency is 700Hz, and the 3rd frequency is 2000Hz, and the 4th frequency is 3400Hz.
7, the method for claim 1 is characterized in that, in the described steps A, the size of window is 256 or 512.
8, a kind of digital filter comprises:
A/D converter: the time domain simulation voice signal is converted to the time domain discrete digital signal;
Windowing process unit: the time domain discrete digital signal after the A/D converter conversion is carried out windowing process;
Time-domain and frequency-domain converting unit: adopt Fourier transform to be converted to the frequency domain digital signal discrete time-domain digital signal in the window;
Signal noise processing unit: at frequency domain digital signal is carried out energy attenuation and handle;
The time domain frequency domain converting unit: the frequency domain digital signal that will carry out after energy attenuation is handled adopts inverse fourier transform to be reduced to the time-domain digital signal;
The D/A switch unit: with the time-domain digital conversion of signals is time domain simulation signal and output.
It is characterized in that described signal noise processing unit comprises:
First module, to signal frequency less than first frequency and greater than the signal energy of the 4th frequency decay one set greater than 0 energy value;
Unit second, to signal frequency more than or equal to described first frequency and smaller or equal to described second frequency, and signal frequency more than or equal to described the 3rd frequency and smaller or equal to the signal energy of described the 4th frequency by linear attenuation;
Described first frequency, second frequency, the 3rd frequency and the 4th frequency frequency for being provided with from low to high by signal frequency; Described second frequency is a voice signal low frequency frequency, and described the 3rd frequency is a voice signal high frequency frequency.
9, digital filter as claimed in claim 8, it is characterized in that, also include the window function arithmetic element, be connected between described signal noise processing unit and the time domain frequency domain converting unit, after frequency domain digital signal after will carrying out energy attenuation and handling carries out the window function computing, export to described time domain frequency domain converting unit again.
10, digital filter as claimed in claim 8 is characterized in that, described first frequency is 300Hz, and second frequency is 700Hz, and the 3rd frequency is 2000Hz, and the 4th frequency is 3400Hz.
11, digital filter as claimed in claim 8 is characterized in that, described window function is a Hamming window function.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100891488A CN100426378C (en) | 2005-08-04 | 2005-08-04 | Dynamic noise eliminating method and digital filter |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005100891488A CN100426378C (en) | 2005-08-04 | 2005-08-04 | Dynamic noise eliminating method and digital filter |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1719517A CN1719517A (en) | 2006-01-11 |
| CN100426378C true CN100426378C (en) | 2008-10-15 |
Family
ID=35931334
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2005100891488A Expired - Fee Related CN100426378C (en) | 2005-08-04 | 2005-08-04 | Dynamic noise eliminating method and digital filter |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100426378C (en) |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1822092B (en) * | 2006-03-28 | 2010-05-26 | 北京中星微电子有限公司 | Method and its device for elliminating background noise in speech input |
| DE602006018030D1 (en) * | 2006-11-24 | 2010-12-16 | Research In Motion Ltd | System and method for reducing uplink noise |
| US9058819B2 (en) | 2006-11-24 | 2015-06-16 | Blackberry Limited | System and method for reducing uplink noise |
| GB2466668A (en) * | 2009-01-06 | 2010-07-07 | Skype Ltd | Speech filtering |
| CN101567676B (en) * | 2009-05-08 | 2012-05-23 | 江苏大学 | Wave filtering method |
| CN101930425B (en) | 2009-06-24 | 2015-09-30 | 华为技术有限公司 | Signal processing method, data processing method and device |
| CN103187069B (en) * | 2011-12-29 | 2015-03-18 | 福建联拓科技有限公司 | Method and device for subaudio frequency last syllable detection |
| CN105469806B (en) * | 2014-09-12 | 2020-02-21 | 联想(北京)有限公司 | Sound processing method, device and system |
| CN105427868A (en) * | 2015-10-30 | 2016-03-23 | 杭州乐哈思智能科技有限公司 | Method for eliminating noise of VOIP system bidirectional duplex hand-free voice |
| CN106887235A (en) * | 2015-12-16 | 2017-06-23 | 芋头科技（杭州）有限公司 | A kind of voice de-noising method and intelligent electronic device |
| CN106816155B (en) * | 2016-12-23 | 2020-04-24 | 维沃移动通信有限公司 | Method and device for improving signal-to-noise ratio of voice transmission |
| CN109951174B (en) * | 2019-04-16 | 2021-07-13 | 上海交通大学 | Fast Fourier transform filtering method |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1529974A (en) * | 2001-05-07 | 2004-09-15 | 西门子信息及通讯网络公司 | Enhancement of sound quality for computer telephony system |
-
2005
- 2005-08-04 CN CNB2005100891488A patent/CN100426378C/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1529974A (en) * | 2001-05-07 | 2004-09-15 | 西门子信息及通讯网络公司 | Enhancement of sound quality for computer telephony system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1719517A (en) | 2006-01-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100426378C (en) | Dynamic noise eliminating method and digital filter | |
| Schroeder | Vocoders: Analysis and synthesis of speech | |
| WO2001033550A1 (en) | Speech parameter compression | |
| AU8227798A (en) | Method and apparatus for speech enhancement in a speech communication system | |
| MY121575A (en) | Method for noise reduction | |
| CN108564965B (en) | Anti-noise voice recognition system | |
| EP3113183B1 (en) | Speech intelligibility improving apparatus and computer program therefor | |
| CN1263426A (en) | Voice frequency processing equipment, receiver, filtering and method for recovering useful signal | |
| CN101176149A (en) | Signal processing system for tonal noise robustness | |
| DE60212617D1 (en) | DEVICE FOR LANGUAGE IMPROVEMENT | |
| DE60124079D1 (en) | language processing | |
| CN102314883B (en) | Music noise judgment method and voice noise elimination method | |
| EP1343146A3 (en) | Audio signal processing based on a perceptual model | |
| WO1998006090A1 (en) | Speech/audio coding with non-linear spectral-amplitude transformation | |
| CN202650529U (en) | Simple and easy voice gender detection apparatus | |
| EP2063420A1 (en) | Method and assembly to enhance the intelligibility of speech | |
| Lefebvre et al. | Shaping coding noise with frequency-domain companding | |
| GB2336978A (en) | Improving speech intelligibility in presence of noise | |
| Singh et al. | A real-time DSP-based system for voice activity detection and background noise reduction | |
| Lefebvre et al. | Spectral amplitude warping (SAW) for noise spectrum shaping in audio coding | |
| Zhen et al. | An Enhanced RASTA processing for speaker identification | |
| EP1395982A1 (en) | Adpcm speech coding system with phase-smearing and phase-desmearing filters | |
| Fulop et al. | Signal Processing in Speech and Hearing Technology | |
| d'Alessandro | Auditory-based wavelet representation | |
| WO1998005031A3 (en) | A method and a device for the reduction impulse noise from a speech signal |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20081015 Termination date: 20120804 |