CN103839552A - Environmental noise identification method based on Kurt - Google Patents
Environmental noise identification method based on Kurt Download PDFInfo
- Publication number
- CN103839552A CN103839552A CN201410107116.5A CN201410107116A CN103839552A CN 103839552 A CN103839552 A CN 103839552A CN 201410107116 A CN201410107116 A CN 201410107116A CN 103839552 A CN103839552 A CN 103839552A
- Authority
- CN
- China
- Prior art keywords
- kurt
- identification method
- environmental noise
- noise
- threshold value
- 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
Abstract
The invention discloses an environmental noise identification method based on Kurt. The environmental noise identification method is insensitive to changes in the amplitude ratio of a noise signal to an effective acoustical signal, an identification result only relates to statistical properties of the signals, and accurate identification can still be conducted under the circumstance that frequency band of the noise signal and the frequency band of the effective acoustical signal overlap. The environmental noise identification method based on kurtosis is implemented in the following steps that a threshold value alpha and a threshold value beta are set in advance, wherein the threshold value alpha and the threshold value beta are larger than zero; then, an on-site acoustical signal is acquired, and the Kurt (x) of the on-site acoustical signal is calculated; finally, judgment is made according to the threshold values which are set in advance. The environmental noise identification method based on the Kurt can be used for volume adjustment of voice/music playing devices; the environmental noise identification method can also be applied to voice recording equipment and provide a basis for voice enhancement strategy selection.
Description
Technical field
The present invention relates to field of voice signal, specifically a kind of environmental noise recognition methods based on kurtosis.
Background technology
Along with the development of multimedia technology, the application of recording and playback is more and more extensive.In tone playing equipment, people wish accurately to detect whether have noise, thereby carry out volume adjusting, such as in quiet office, only need to, compared with amount of bass, in more noisy occasion, need to improve volume, guarantee that volume is than more than the large 20dB of noise.In sound pick-up outfit, people also wish can according to different environment self-adaptions take different sound enhancement methods.Therefore the accurate recognition methods of environmental noise has stronger using value.
Some early stage noise recognition methodss are classified to sound by amplitude or the intensity analyzed in each frequency range, and those sound with minimum amplitude are assumed to be noise.Clearly, this method is very responsive to the amplitude proportional of noise signal and effective voice signal, under the occasion becoming, often obtains wrong identification result in the time of some noise intensity the unknowns or noise intensity; And in the time of the frequency overlap of effective voice signal and psophometer noise signal, this method can lose efficacy.Chinese patent ZL200610160644.2 has proposed a kind of environment noise test method, difference between the signal of playing by analysis and the signal of collection, judges whether to exist noise, and this method need to be predicted play signal, to with the equipment such as music, be suitable for completely; And for sound pick-up outfit, effective sound signal waveform of recording is unknown, this method cannot be suitable for.
Summary of the invention
The object of this invention is to provide a kind of automatic identifying method of environmental noise.
The technical solution used in the present invention is: the method based on kurtosis is identified environmental noise, and its specific implementation step is as follows:
Step 1, default two are greater than 0 threshold alpha and β;
Step 2, collection site voice signal, and calculate its kurtosis Kurt (x);
Step 3, adjudicate according to default threshold value:
If Kurt (x)≤α Then noise intensity is high
During If α <Kurt (x)≤β Then noise intensity is
If Kurt (x) > β Then noise intensity is low
Kurtosis in described step 2 is a kind of fourth order cumulant of stochastic variable, and computing formula is as follows:
Kurt(x)=E{x
4}-3(E{x
2})
2 (1)
For Gaussian distribution stochastic variable x (t), its Fourth-order moment E{x
4equal 3 (E{x
2)
2, kurtosis is 0; For non-gaussian variable, Gauss is stronger, and the absolute value of kurtosis just more approaches 0.
Compared with prior art, the invention has the beneficial effects as follows:
1, kurtosis method of the present invention belongs to the category of higher order statistical method, insensitive to the amplitude proportional variation of noise signal and effective voice signal, and and frequency-independent, even still effective in the situation that of frequency noise and sound frequency coincidence.
2, described method does not need extra hardware supported in the time realizing, and only needs the equipment such as common microphone just can easily to realize, and in modern sound pick-up outfit, tone playing equipment, all has, and does not need to add other hardware, therefore realizes simply, and cost is lower.
Accompanying drawing explanation
Fig. 1 is workflow diagram of the present invention;
Fig. 2 is signal waveforms in the embodiment of the present invention.
Embodiment
Main thought of the present invention is as follows: the voice signal that single sound source is sent is generally non-Gaussian distribution, and being the sound being sent by multiple separate sound sources, noise is formed by stacking, known according to the central limit theorem in theory of probability, under certain conditions, independently the distribution of stochastic variable sum trends towards Gaussian distribution.That is to say, the distribution that independent random variable sum forms than any one in original stochastic variable closer to Gaussian distribution.Therefore compared with effective sound, noise presents stronger Gauss, can identify noise according to the Gauss of acoustic scene tone signal.And the fourth order cumulant kurtosis of stochastic variable can be used for describing quantitatively the Gauss of a stochastic variable, in the present invention for describing the intensity of sound noise.
Describe the specific embodiment of the present invention in detail below in conjunction with accompanying drawing:
As shown in Figure 2, be the oscillogram of original signal in the present embodiment, the sampling rate of recording is 16KHz, duration is 2.5 seconds.Fig. 2-a is one section of sound recording on motorbus, and scene is comparatively noisy, and noise is larger; Fig. 2-b is one section of speech sound recording at Conference Hall, and scene is quieter, there is no noise.
As shown in Figure 1, the present embodiment comprises the steps:
Step 1, default two are greater than 0 threshold alpha and β: through great many of experiments, threshold value value is α=4.6, β=6.0 o'clock, discrimination is better;
Step 2, collection site voice signal x (t), and calculate its kurtosis Kurt (x): on bus, record two signals shown in Fig. 2 with Conference Hall respectively, according to formula Kurt (x)=E{x
4}-3 (E{x
2)
2calculate respectively its kurtosis, value is respectively 3.2992 and 8.2757;
Step 3, adjudicates according to default threshold value: because the kurtosis value of signal is 3.2992 shown in Fig. 2-a, be less than α, therefore noise signal can be judged to be stronger, the kurtosis value of signal shown in Fig. 2-b is 8.2757, is greater than β, therefore can be judged to be noise signal lower.
Claims (3)
1. the neighbourhood noise recognition methods based on kurtosis, is characterized in that, the method comprises the following steps:
Step 1, default two are greater than 0 threshold alpha and β;
Step 2, collection site voice signal, and calculate its kurtosis Kurt (x);
Step 3, adjudicates according to default threshold value.
2. the method for claim 1, is characterized in that, in described step 2, the computing formula of kurtosis is Kurt (x)=E{x
4}-3 (E{x
2)
2.
3. the method for claim 1, is characterized in that, in described step 3, decision method is:
If Kurt (x)≤α Then noise intensity is high
During If α <Kurt (x)≤β Then noise intensity is
If Kurt (x) > β Then noise intensity is low.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410107116.5A CN103839552A (en) | 2014-03-21 | 2014-03-21 | Environmental noise identification method based on Kurt |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410107116.5A CN103839552A (en) | 2014-03-21 | 2014-03-21 | Environmental noise identification method based on Kurt |
Publications (1)
Publication Number | Publication Date |
---|---|
CN103839552A true CN103839552A (en) | 2014-06-04 |
Family
ID=50802982
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410107116.5A Pending CN103839552A (en) | 2014-03-21 | 2014-03-21 | Environmental noise identification method based on Kurt |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103839552A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107176123A (en) * | 2016-03-10 | 2017-09-19 | 现代自动车株式会社 | Sound detection information providing method, vehicle periphery sound detection device and vehicle |
CN108922565A (en) * | 2018-07-30 | 2018-11-30 | 四川大学 | Cleft palate speech based on FTSL spectral line swallows fricative automatic testing method |
CN110487546A (en) * | 2018-05-10 | 2019-11-22 | 上汽通用汽车有限公司 | Gearbox beat noise test method, test device and evaluation method |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2002023776A (en) * | 2000-07-13 | 2002-01-25 | Univ Kinki | Method for identifying speaker voice and non-voice noise in blind separation, and method for specifying speaker voice channel |
CN1350747A (en) * | 2000-01-13 | 2002-05-22 | 皇家菲利浦电子有限公司 | Noise reduction |
CN1815550A (en) * | 2005-02-01 | 2006-08-09 | 松下电器产业株式会社 | Method and system for identifying voice and non-voice in envivonment |
CN1972120A (en) * | 2006-11-29 | 2007-05-30 | 北京中星微电子有限公司 | A method and device for implementing adjustment of volume based on environmental noise detection |
CN102519726A (en) * | 2011-12-28 | 2012-06-27 | 昆明理工大学 | Acoustic-based diagnosis (ABD) method for compound fault of rolling bearing |
CN103083012A (en) * | 2012-12-24 | 2013-05-08 | 太原理工大学 | Atrial fibrillation signal extraction method based on blind source separation |
-
2014
- 2014-03-21 CN CN201410107116.5A patent/CN103839552A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1350747A (en) * | 2000-01-13 | 2002-05-22 | 皇家菲利浦电子有限公司 | Noise reduction |
JP2002023776A (en) * | 2000-07-13 | 2002-01-25 | Univ Kinki | Method for identifying speaker voice and non-voice noise in blind separation, and method for specifying speaker voice channel |
CN1815550A (en) * | 2005-02-01 | 2006-08-09 | 松下电器产业株式会社 | Method and system for identifying voice and non-voice in envivonment |
CN1972120A (en) * | 2006-11-29 | 2007-05-30 | 北京中星微电子有限公司 | A method and device for implementing adjustment of volume based on environmental noise detection |
CN102519726A (en) * | 2011-12-28 | 2012-06-27 | 昆明理工大学 | Acoustic-based diagnosis (ABD) method for compound fault of rolling bearing |
CN103083012A (en) * | 2012-12-24 | 2013-05-08 | 太原理工大学 | Atrial fibrillation signal extraction method based on blind source separation |
Non-Patent Citations (2)
Title |
---|
董治强: "独立分量分析及其在语音特征提取中的应用", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
赵永健: "独立分量分析算法及其在信号处理中的应用研究", 《中国博士学位论文全文数据库 医药卫生科技辑》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107176123A (en) * | 2016-03-10 | 2017-09-19 | 现代自动车株式会社 | Sound detection information providing method, vehicle periphery sound detection device and vehicle |
CN107176123B (en) * | 2016-03-10 | 2021-04-16 | 现代自动车株式会社 | Sound detection information providing method, vehicle surrounding sound detection device, and vehicle |
CN110487546A (en) * | 2018-05-10 | 2019-11-22 | 上汽通用汽车有限公司 | Gearbox beat noise test method, test device and evaluation method |
CN110487546B (en) * | 2018-05-10 | 2021-12-14 | 上汽通用汽车有限公司 | Gearbox knocking noise testing method, testing device and evaluation method |
CN108922565A (en) * | 2018-07-30 | 2018-11-30 | 四川大学 | Cleft palate speech based on FTSL spectral line swallows fricative automatic testing method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10602267B2 (en) | Sound signal processing apparatus and method for enhancing a sound signal | |
US10242677B2 (en) | Speaker dependent voiced sound pattern detection thresholds | |
US9959886B2 (en) | Spectral comb voice activity detection | |
KR20210038871A (en) | Detection of replay attacks | |
US10535361B2 (en) | Speech enhancement using clustering of cues | |
CN104464722A (en) | Voice activity detection method and equipment based on time domain and frequency domain | |
US9792898B2 (en) | Concurrent segmentation of multiple similar vocalizations | |
CN104992712A (en) | Music reorganization-based music score automatic formation method | |
US20160295320A1 (en) | Audio capturing enhancement method and audio capturing system using the same | |
CN103839552A (en) | Environmental noise identification method based on Kurt | |
US10142730B1 (en) | Temporal and spatial detection of acoustic sources | |
Nakadai et al. | Footstep detection and classification using distributed microphones | |
US20210157838A1 (en) | Methods and apparatus to fingerprint an audio signal via exponential normalization | |
US20220150624A1 (en) | Method, Apparatus and Computer Program for Processing Audio Signals | |
Araki et al. | Speaker indexing and speech enhancement in real meetings/conversations | |
Kindt et al. | Exploiting speaker embeddings for improved microphone clustering and speech separation in ad-hoc microphone arrays | |
JP5970985B2 (en) | Audio signal processing apparatus, method and program | |
KR101732399B1 (en) | Sound Detection Method Using Stereo Channel | |
Ma et al. | Binaural cues for fragment-based speech recognition in reverberant multisource environments | |
WO2022188712A1 (en) | Method and apparatus for switching main microphone, voice detection method and apparatus for microphone, microphone-loudspeaker integrated device, and readable storage medium | |
Dang et al. | An Iterative Steered Response Power Algorithm for Multi-Source Localization and Counting Using Distributed Microphone Networks | |
Marin-Hurtado et al. | Robust non-vad implementation of Multichannel Wiener filter for binaural noise reduction | |
Cheng et al. | The comparative study on separation of sound sources in various cases | |
JP2015535616A (en) | How to determine whether a measurement signal matches a model signal | |
Abdipour et al. | Two‐Microphone Binary Mask Speech Enhancement in Diffuse and Directional Noise Fields |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20140604 |