CN102737646A - Real-time dynamic voice noise reduction method for single microphone - Google Patents
Real-time dynamic voice noise reduction method for single microphone Download PDFInfo
- Publication number
- CN102737646A CN102737646A CN2012102091699A CN201210209169A CN102737646A CN 102737646 A CN102737646 A CN 102737646A CN 2012102091699 A CN2012102091699 A CN 2012102091699A CN 201210209169 A CN201210209169 A CN 201210209169A CN 102737646 A CN102737646 A CN 102737646A
- Authority
- CN
- China
- Prior art keywords
- voice
- noise
- signals
- signal
- real
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Circuit For Audible Band Transducer (AREA)
Abstract
The invention discloses a real-time dynamic voice noise reduction method for a single microphone. The real-time dynamic voice noise reduction method comprises the following steps of: dividing an input signal of the microphone into sub-band signals with different frequencies by a filter bank in a filtering manner, and sampling the sub-band signals after adjusting the sub-band signals; statistically analyzing the sub-band signals, identifying voice signals and noise signals, separating the voice signals from the noise signals, modeling the voice signals and the noise signals, and extracting the feature of the voice signals and the feature of the noise signals in real time; dynamically designing an adaptive filter in real time according to the feature of the voice signals and the feature of the noise signals, enabling the adaptive filter to filter out noise to the great extent, and enabling voice to pass through the adaptive filter; filtering the input signal of the microphone by utilizing the adaptive filter in the design; and synthesizing and restoring the filtered sub-band signals into full-band signals by a sub-band synthesizing filter. Sub-band arithmetic is adopted, processing calculated quantity is reduced, the real-time dynamic voice noise reduction method is applicable to application from narrow bands to all bands, the stability of a system is increased, and processing efficiency is improved.
Description
Technical field
The present invention relates to the voice de-noising technical field, relating in particular to can be along with the variation of the voice of environment and noise and Real-time and Dynamic is carried out the method for voice de-noising to microphone.
Background technology
Noise problem is of long duration, all speech ciphering equipments of ground unrest problem and with giving birth to.In order to solve the problem that microphone picks up ground unrest in public address, recording and the communication, the most original method reduces input circuit (comprising microphone) gain, the intensity of enhanced speech signal itself (as speaking up, microphone is pressed close to lip etc.) exactly.Another kind method is to improve signal to noise ratio (S/N ratio) through the characteristic of microphone, as using directional microphone, reduces microphone pickup spatial noise, reaches the purpose of noise reduction.The index of above method noise reduction is limited, method and the environment that uses had restriction, therefore needs further method come noise reduction.Thus, simulation noise reduction and be the digital noise reduction of technical support with digital signal processing (DSP) and related algorithm thereof has been proposed.For example: CN101848288A discloses " a kind of simulation noise reduction system of microphone and method ", is to utilize the ground unrest microphone in the microphone array to pick up ground unrest, and the signal with the voice pickup microphone differs then, to offset ground unrest; The method and directional microphone are the acts that plays the same tune on different musical instruments; These class methods exist the problem that is difficult to overcome in realizations such as pairing of the spectrum gain of microphone and circuit conditioning.Present most method is to use the method for microphone array DSP; Utilize the correlativity of the ground unrest that different microphones pick up; The noise that picks up from the ground unrest microphone; Extrapolate the noise signal that speech microphone picks up, use difference algorithm to balance out the noise in the speech microphone then, reach the purpose of noise reduction.
The DSP noise-reduction method of delivering at present all uses microphone array, needs 2 microphones at least, and this makes troubles for the system mechanics structural design, makes production and processing complicated, and system cost rises.In addition, owing to adopting the method for calculating that noise is offset, the noise property of there are differences of diverse location can't obtain degree of depth noise reduction, and general noise reduction is not more than 6 decibels; Simultaneously, can in voice signal, add culture noise; Calculate noise signal owing to use least square method; Calculated amount is big; Existing algorithm can only be applied in arrowband (300Hz-3400Hz) voice communication system, is difficult to satisfy in wideband of new generation (50Hz-7000Hz) or full range (20Hz-20000Hz) communication system.
Summary of the invention
The technical matters that the present invention will solve provides a kind of and adopts single microphone; The method of voice de-noising along with the variation of the voice of environment and noise and real-time dynamicly; The algorithm computation amount of this voice de-noising method is few, applicable to from the arrowband to the application of full range band.
The technical scheme that adopts for solving the problems of the technologies described above:
A kind of Real-time and Dynamic voice de-noising method of single microphone is characterized in that may further comprise the steps:
One, the input signal of microphone is divided into the subband signal of different frequency by Methods of Subband Filter Banks filtering, and with its modulation back down-sampling subband signal;
Two, subband signal is carried out statistical study, identification and isolating speech signals and noise signal to voice and noise signal modeling, are extracted the characteristic of voice and noise signal in real time;
Three,, design sef-adapting filter real-time dynamicly and make its filtering noise and voice are passed through to greatest extent to voice signal and noise signal characteristic;
Four, use the sef-adapting filter of above-mentioned design that microphone input signal is carried out filtering;
Five, subband signal after the filtering is reduced to full band signal through subband synthesis filter is synthetic.
Adopt the beneficial effect that the present invention brought: the present invention only needs single microphone pickoff signals; Along with the variation of environment is analyzed recognizing voice and noise signal intelligently, extracted its characteristic and modeling, the design modification wave filter keeps voice simultaneously with filtering noise in real time therefrom, has simplified system mechanics structural design and production and processing, has reduced system cost; Can obtain degree of depth noise reduction (but the maximum noise reduction is greater than 15 decibels); Simultaneously, the noise reduction depth controlled does not have (or very low) culture noise.The present invention adopts the subband algorithm, has reduced computational processing, applicable to from the arrowband to the application of full range band, improved the stability of system, improve treatment effeciency.The present invention can be used for application such as voice public address, recording, communication, widespread use in systems such as mobile phone, computer, STB, local public address equipment, telecommunication equipment, voice recording equipment.
Description of drawings
Fig. 1 is the algorithm synoptic diagram of the Real-time and Dynamic voice de-noising method of the single microphone of the present invention.Wherein S_mic is the sampled signal of single microphone, and S ' _ mic is the output signal after noise reduction process, and steering order is used for controlling the degree of depth of whether making noise reduction process and noise reduction.
Fig. 2 is the sampled signal waveform of microphone.Be the ambient noise signal waveform than flat portions wherein, stronger bump is the mixed waveform signal of voice and noise.
Fig. 3 is Fig. 2 signal signal output waveform after noise reduction process of the present invention.Can see that from figure ground unrest is reduced significantly, and the wave amplitude of voice there is not significant change.
Embodiment
The present invention is different with the noise-reduction method of other use microphone array, and a kind of of motion used single microphone, utilized the method for sef-adapting filter to voice de-noising.Owing to only need a microphone, simplified the system mechanics structural design greatly and produced, reduced circuit cost simultaneously.The present invention can be used for application such as voice public address, recording, communication, widespread use in systems such as mobile phone, computer, STB, local public address equipment, telecommunication equipment, voice recording equipment.
In order to reach noise reduction, to improve the signal to noise ratio (S/N ratio) purpose; The present invention adopts unique voice and noise Intelligent Recognition isolation technics; To voice and noise modeling,, make it separate voice and noise to greatest extent according to its characteristic designing filter real-time dynamicly; Through using the sef-adapting filter that designs, reach the purpose of noise reduction to input signal filtering.The present invention constitutes (referring to Fig. 1) by following steps:
The input signal of microphone is obtained subband signal through the Methods of Subband Filter Banks separation.Number of sub-bands and subband bandwidth are looked application and are decided: generally speaking; The subband bandwidth is narrow more steep more; The degree of separation of subband signal is good more; Can design better wave filter, but the sub-filter that needs like this is long, long processing group delay is introduced in meeting, in local public address is used, long time-delay can not be arranged.The subband width such as can use at frequency range or proportional width, and number of sub-bands is relevant with the subband bandwidth with the frequency band (is the sampling trip mostly 8KHz, public address equipment are 48KHz like mobile phone) of application.For example, the sampling rate of Fig. 2, instance shown in 3 is 48KHz, and the subband bandwidth is the 187.5Hz equiband, has 129 subbands, and the group delay that entire process causes is about 6 to 8 milliseconds.
Each subband signal to reduce data volume, reaches the purpose that reduces calculated amount, improves efficiency of algorithm through modulation, LPF, down-sampling (Down-sampling).Down-sampling is bigger than more, and efficiency of algorithm is high more; Theoretically, under the prerequisite of drop-out not, down-sampling can reach N:1 than maximum, promptly is reduced to 1 point from the N point, and N is the number (or ratio of former bandwidth and subband bandwidth) of subband here; But consider the finiteness of practical application median filter performance and reduce the connection problem (Blocking-effect) between the data segment; The down-sampling that the present invention adopts is than being N:2; Make to have overlapping (Overlapping) between the data block, but the invention is not restricted to this.
According to the characteristic of voice and noise, in subband, signal is carried out analytic statistics, distinguish voice and noise signal, to voice and noise signal modeling and obtain its eigenwert.Ground unrest generally has more stable frequency spectrum and intensity, and voice then have tangible speech envelope, frequency spectrum and intensity to change with voice; Therefore, utilize the first order IIR filtering device can obtain the spectrum distribution of noise, thus can instantaneous ground recognizing voice or noise signal, and the characteristic of computing voice and noise signal more accurately.
According to voice and the noise signal model and the characteristic of above-mentioned acquisition, revise designing filter in real time, this wave filter make the noise characteristic signal be attenuated blocking-up and make the phonetic feature signal undampedly through, reach the purpose of filtering noise.Designing filter makes it separate voice and noise to greatest extent real-time dynamicly.
Change the noise reduction degree of depth of design sef-adapting filter according to steering order to noise.The noise filtering degree of depth can be continuous adjustable from 0 decibel to 20 decibels, and usually under the prerequisite that does not influence speech quality, but 10 to 15 decibels of noise reductions when noise reduction is more than 15 decibels, have certain influence to tonequality.
Use the sef-adapting filter of previous designs that microphone input signal is carried out filtering, it is constant to keep phonetic element to reduce noise contribution.This wave filter can use the FIR wave filter, but is not limited to above wave filter.
Subband signal behind the noise reduction filtering is restored wave filter through subband be reduced to former band speech signal.Subband restores wave filter must be corresponding with sub-filter; Restore (Perfect Reconstruction) wave filter to being ideal with perfection; In practical application, sub-filter and subband restore the caused distortion of wave filter should make distortion be difficult for being discovered less than-40 decibels.
The present invention need not second microphone and picks up ground unrest, only can realize the function of noise reduction with a microphone; Analyze recognizing voice and noise signal intelligently, extract its characteristic and modeling, the design modification wave filter keeps voice simultaneously with filtering noise in real time therefrom; Filtering comes noise reduction to input signal to use sef-adapting filter; Use the subband technology to reduce the signal Processing calculated amount, improve treatment effeciency.
The foregoing description does not limit the present invention in any way, and all employings are equal to the technical scheme that mode obtained of replacement or equivalent transformation, all drop in protection scope of the present invention.
Claims (1)
1. the Real-time and Dynamic voice de-noising method of a single microphone is characterized in that may further comprise the steps:
One, the input signal of microphone is divided into the subband signal of different frequency by Methods of Subband Filter Banks filtering, and with its modulation back down-sampling subband signal;
Two, subband signal is carried out statistical study, identification and isolating speech signals and noise signal to voice and noise signal modeling, are extracted the characteristic of voice and noise signal in real time;
Three,, design sef-adapting filter real-time dynamicly and make its filtering noise and voice are passed through to greatest extent to voice signal and noise signal characteristic;
Four, use the sef-adapting filter of above-mentioned design that microphone input signal is carried out filtering;
Five, subband signal after the filtering is reduced to full band signal through subband synthesis filter is synthetic.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012102091699A CN102737646A (en) | 2012-06-21 | 2012-06-21 | Real-time dynamic voice noise reduction method for single microphone |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012102091699A CN102737646A (en) | 2012-06-21 | 2012-06-21 | Real-time dynamic voice noise reduction method for single microphone |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102737646A true CN102737646A (en) | 2012-10-17 |
Family
ID=46993015
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2012102091699A Pending CN102737646A (en) | 2012-06-21 | 2012-06-21 | Real-time dynamic voice noise reduction method for single microphone |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102737646A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103594092A (en) * | 2013-11-25 | 2014-02-19 | 广东欧珀移动通信有限公司 | Single microphone voice noise reduction method and device |
| CN106228991A (en) * | 2014-06-26 | 2016-12-14 | 华为技术有限公司 | Decoding method, Apparatus and system |
| CN106454642A (en) * | 2016-09-23 | 2017-02-22 | 佛山科学技术学院 | Adaptive sub-band audio feedback suppression method |
| US9956661B2 (en) | 2014-11-20 | 2018-05-01 | Industrial Technology Research Institute | Feedback control numerical machine tool and method thereof |
| CN108540888A (en) * | 2018-05-24 | 2018-09-14 | 韩雪 | A kind of improved earphone noise reduction system and its noise-reduction method |
| WO2020051786A1 (en) * | 2018-09-12 | 2020-03-19 | Shenzhen Voxtech Co., Ltd. | Signal processing device having multiple acoustic-electric transducers |
| RU2771919C1 (en) * | 2018-09-12 | 2022-05-13 | Шэньчжэнь Шокз Ко., Лтд. | Signal processing apparatus with multiple acoustic-electrical converters |
| US11665482B2 (en) | 2011-12-23 | 2023-05-30 | Shenzhen Shokz Co., Ltd. | Bone conduction speaker and compound vibration device thereof |
| CN117350099A (en) * | 2023-09-11 | 2024-01-05 | 北京五瑞美阳医疗器械有限责任公司 | Finite element analysis-based respirator noise reduction structure optimization method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1201547A (en) * | 1995-09-14 | 1998-12-09 | 艾利森公司 | System for adaptively filtering audio signals to enhance speech intelligibility in noisy environmental conditions |
| CN1689072A (en) * | 2002-08-16 | 2005-10-26 | 数字信号处理工厂有限公司 | Method and system for processing subband signals using adaptive filters |
| CN1805011A (en) * | 2005-12-23 | 2006-07-19 | 北京中星微电子有限公司 | Adaptive filter method and apparatus for improving speech quality of mobile communication apparatus |
| CN101320996A (en) * | 2008-05-27 | 2008-12-10 | 中山大学 | Self-adapting noise elimination apparatus and method |
-
2012
- 2012-06-21 CN CN2012102091699A patent/CN102737646A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1201547A (en) * | 1995-09-14 | 1998-12-09 | 艾利森公司 | System for adaptively filtering audio signals to enhance speech intelligibility in noisy environmental conditions |
| CN1689072A (en) * | 2002-08-16 | 2005-10-26 | 数字信号处理工厂有限公司 | Method and system for processing subband signals using adaptive filters |
| CN1805011A (en) * | 2005-12-23 | 2006-07-19 | 北京中星微电子有限公司 | Adaptive filter method and apparatus for improving speech quality of mobile communication apparatus |
| CN101320996A (en) * | 2008-05-27 | 2008-12-10 | 中山大学 | Self-adapting noise elimination apparatus and method |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11665482B2 (en) | 2011-12-23 | 2023-05-30 | Shenzhen Shokz Co., Ltd. | Bone conduction speaker and compound vibration device thereof |
| CN103594092A (en) * | 2013-11-25 | 2014-02-19 | 广东欧珀移动通信有限公司 | Single microphone voice noise reduction method and device |
| US10614822B2 (en) | 2014-06-26 | 2020-04-07 | Huawei Technologies Co., Ltd. | Coding/decoding method, apparatus, and system for audio signal |
| CN106228991A (en) * | 2014-06-26 | 2016-12-14 | 华为技术有限公司 | Decoding method, Apparatus and system |
| US10339945B2 (en) | 2014-06-26 | 2019-07-02 | Huawei Technologies Co., Ltd. | Coding/decoding method, apparatus, and system for audio signal |
| CN106228991B (en) * | 2014-06-26 | 2019-08-20 | 华为技术有限公司 | Decoding method, apparatus and system |
| US9956661B2 (en) | 2014-11-20 | 2018-05-01 | Industrial Technology Research Institute | Feedback control numerical machine tool and method thereof |
| CN106454642A (en) * | 2016-09-23 | 2017-02-22 | 佛山科学技术学院 | Adaptive sub-band audio feedback suppression method |
| CN106454642B (en) * | 2016-09-23 | 2019-01-08 | 佛山科学技术学院 | Adaptive sub-band audio feedback suppression methods |
| CN108540888A (en) * | 2018-05-24 | 2018-09-14 | 韩雪 | A kind of improved earphone noise reduction system and its noise-reduction method |
| RU2771919C1 (en) * | 2018-09-12 | 2022-05-13 | Шэньчжэнь Шокз Ко., Лтд. | Signal processing apparatus with multiple acoustic-electrical converters |
| US11373671B2 (en) | 2018-09-12 | 2022-06-28 | Shenzhen Shokz Co., Ltd. | Signal processing device having multiple acoustic-electric transducers |
| WO2020051786A1 (en) * | 2018-09-12 | 2020-03-19 | Shenzhen Voxtech Co., Ltd. | Signal processing device having multiple acoustic-electric transducers |
| US11875815B2 (en) | 2018-09-12 | 2024-01-16 | Shenzhen Shokz Co., Ltd. | Signal processing device having multiple acoustic-electric transducers |
| CN117350099A (en) * | 2023-09-11 | 2024-01-05 | 北京五瑞美阳医疗器械有限责任公司 | Finite element analysis-based respirator noise reduction structure optimization method |
| CN117350099B (en) * | 2023-09-11 | 2024-04-16 | 北京五瑞美阳医疗器械有限责任公司 | Finite element analysis-based respirator noise reduction structure optimization method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102737646A (en) | Real-time dynamic voice noise reduction method for single microphone | |
| KR101422368B1 (en) | A method and an apparatus for processing an audio signal | |
| CN102074245B (en) | Dual-microphone-based speech enhancement device and speech enhancement method | |
| JP5127754B2 (en) | Signal processing device | |
| CN110536215A (en) | Method, apparatus, calculating and setting and the storage medium of Audio Signal Processing | |
| CN110537221A (en) | Two stages audio for space audio processing focuses | |
| CN106340303B (en) | A kind of voice de-noising method based on temporal frequency domain | |
| US20050004803A1 (en) | Audio signal bandwidth extension | |
| CN204836533U (en) | Acoustic control intelligence audio amplifier with echo elimination function | |
| CN101460999B (en) | blind signal extraction | |
| WO2010094710A3 (en) | Low delay modulated filter bank | |
| CN103514884A (en) | Communication voice denoising method and terminal | |
| CN102347027A (en) | Double-microphone speech enhancer and speech enhancement method thereof | |
| CN102576537B (en) | Method and apparatus for processing audio signals | |
| CN106463106A (en) | Wind noise reduction for audio reception | |
| CN101426058B (en) | System and method for improving quality of multichannel audio call | |
| CN102176312A (en) | System and method for reducing burst noise through wavelet trapped wave | |
| WO2019205796A1 (en) | Frequency-domain processing amount reduction method, apparatus and device | |
| CN109074813A (en) | Handle high definition audio data | |
| JP2012181561A (en) | Signal processing apparatus | |
| Alku et al. | Linear predictive method for improved spectral modeling of lower frequencies of speech with small prediction orders | |
| CN113611279A (en) | Active noise reduction system and noise reduction earphone | |
| Lee et al. | Analysis/synthesis of rolling sounds using a source-filter approach | |
| JPWO2020066542A1 (en) | Acoustic object extraction device and acoustic object extraction method | |
| Yoshida et al. | The Extremal Sampling Technique |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20121017 |