CN110459226A - A method of voice is detected by vocal print engine or machine sound carries out identity veritification - Google Patents
A method of voice is detected by vocal print engine or machine sound carries out identity veritification Download PDFInfo
- Publication number
- CN110459226A CN110459226A CN201910765307.3A CN201910765307A CN110459226A CN 110459226 A CN110459226 A CN 110459226A CN 201910765307 A CN201910765307 A CN 201910765307A CN 110459226 A CN110459226 A CN 110459226A
- Authority
- CN
- China
- Prior art keywords
- vocal print
- voice
- print engine
- sound
- audio
- 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
- 230000001755 vocal effect Effects 0.000 title claims abstract description 49
- 238000000034 method Methods 0.000 title claims abstract description 39
- 238000001514 detection method Methods 0.000 claims abstract description 19
- 238000009432 framing Methods 0.000 claims abstract description 5
- 239000013598 vector Substances 0.000 claims description 15
- 238000000605 extraction Methods 0.000 claims description 11
- 230000008569 process Effects 0.000 claims description 11
- 238000012795 verification Methods 0.000 claims description 10
- 239000000284 extract Substances 0.000 claims description 9
- 230000009466 transformation Effects 0.000 claims description 3
- 238000012546 transfer Methods 0.000 claims description 2
- 230000006870 function Effects 0.000 description 18
- 238000005516 engineering process Methods 0.000 description 12
- 210000000056 organ Anatomy 0.000 description 3
- 238000001228 spectrum Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 208000001613 Gambling Diseases 0.000 description 1
- HUTDUHSNJYTCAR-UHFFFAOYSA-N ancymidol Chemical compound C1=CC(OC)=CC=C1C(O)(C=1C=NC=NC=1)C1CC1 HUTDUHSNJYTCAR-UHFFFAOYSA-N 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000011840 criminal investigation Methods 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 229940079593 drug Drugs 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 210000000088 lip Anatomy 0.000 description 1
- 210000000214 mouth Anatomy 0.000 description 1
- 210000003205 muscle Anatomy 0.000 description 1
- 210000003928 nasal cavity Anatomy 0.000 description 1
- 210000003254 palate Anatomy 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 210000001584 soft palate Anatomy 0.000 description 1
- 230000005236 sound signal Effects 0.000 description 1
- 238000010183 spectrum analysis Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 210000002105 tongue Anatomy 0.000 description 1
- 210000000515 tooth Anatomy 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
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
- G10L17/00—Speaker identification or verification techniques
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/03—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of extracted parameters
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/45—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 characterised by the type of analysis window
-
- 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
-
- 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
- G10L2025/783—Detection of presence or absence of voice signals based on threshold decision
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Computational Linguistics (AREA)
- Signal Processing (AREA)
- Soundproofing, Sound Blocking, And Sound Damping (AREA)
Abstract
The method that voice or machine sound carry out identity veritification is detected by vocal print engine the invention discloses a kind of, comprising the following steps: A, receive audio to be verified；B, received audio-frequency information is pre-processed；Pretreatment comprises the steps of: a) duration detection；B) mute detection；C) preemphasis；D) framing adding window；C, vocal print comparison is carried out, judges identity information, the present invention detects voice by way of vocal print engine or machine sound carries out identity veritification.Wherein the vocal print engine can efficiently, fast and accurately recognition detection sound is that speaker expresses in real time, or the recording for passing through audio player plays, and judge whether the sound is me, so that having higher accuracy rate and efficiency in the application scenarios of Application on Voiceprint Recognition, there are wide application scenarios.
Description
Technical field
It is specifically a kind of that voice or the progress of machine sound are detected by vocal print engine the present invention relates to identity identification technical field
The method that identity is veritified.
Background technique
The theoretical basis of Application on Voiceprint Recognition is that each sound has unique feature, so as to by the sound of different people into
Row is effective to be distinguished.This unique feature mainly includes two aspect contents: the 1, size of the operatic tunes；Specifically include throat, nasal cavity
With oral cavity etc., shape, size and the position of these organs determine the size of vocal chord tension and the range of sound frequency.Therefore not
With people although if same, but the frequency distribution of sound is different, sound it is droning have it is loud and clear.Everyone
Sounding chamber be all different, just as fingerprint, everyone sound also just has unique feature.2, phonatory organ is manipulated
Mode, phonatory organ includes lip, tooth, tongue, soft palate and palate muscle etc., and interaction will generate clearly language between them
Sound.And the cooperation mode between them is people is arrived by incidental learning in the exchanging of the day after tomorrow and people around.People speaks in study
During, by simulating the tongue of surrounding different people, the vocal print feature of oneself will be gradually formed.
Sound groove recognition technology in e is developed by AT&T Labs earliest, is mainly used for military information field.With this technology
Gradually develop, the later period at the end of the sixties all employed this technology in fields such as forensic identification, the court evidences in the U.S., from 1967
Till now, the U.S. at least more than 5000 a case include murder, rape, blackmailing, drug smuggling, gambling, corrosion of politics etc. is all
Effective clue and strong evidence are provided by sound groove recognition technology in e.Special emphasis is that vocal print identification has been at present
The standard of the Ministry of Public Security can be used as evidence and be identified.
Application on Voiceprint Recognition is a wide in range concept, and technical aspect, which has, is divided into two classes: i.e. speaker verification's technology and speaker distinguish
Recognize technology, speaker verification's technology is for judging whether unknown speaker is some nominator；The latter is then for recognizing not
Know that speaker is to have recorded who in speaker.
It is all speaker's recognition techniques that we are normally understood, is usually applied to criminal investigation and case detection, criminal's tracking, national defence prison
It listens, personalized application etc., speaker verification's technology is usually applied to securities trading, bank transaction, public security evidence obtaining, PC
Acoustic control lock, automobile sound control lock, identity card, identification of credit card etc..
At present, the common method of Application on Voiceprint Recognition includes template matching method, arest neighbors method, Neural Networks Learning Algorithm,
VQ clustering procedure etc..Although these methods processing means are different, basic principle is similar, for example just started to show to everybody
Sound spectrograph.Sound spectrograph is a kind of representation of image conversion of voice signal, its horizontal axis represents the time, and the longitudinal axis represents frequency,
Voice is distinguished in the amplitude size of each Frequency point with color.The fundamental frequency and harmonics of the sound of speaker show on sound spectrograph
It for bright line one by one, then by different processing means can be obtained by the similarity between different sound spectrographs, finally reach
To the purpose of Application on Voiceprint Recognition.
Speaker verification's technology of Application on Voiceprint Recognition plays very important in the daily information of guarantee modern people, property safety
Effect, for example, can guarantee to log in using sound groove recognition technology in e during credit card, bank transaction, system login
The safety of the property of people, information.In addition, remote identity confirmation has specific application in company's routine attendance check.But exist at present
During these identity validations, all exists and carry out authentication using the mode of recording, so as to cause system vulnerability, serious shadow
The reliability of acoustic system.If Application on Voiceprint Recognition speaker verification function will also become a kind of ornaments without good control.This
Outside, for the identification of machine sound, in certain payment industries, by extracting background sound, sentenced by the way of the verifying of secondary auxiliary
Disconnected to whether there is machine sound, this kind of mode is to a certain extent, it is therefore prevented that user directly prevents from recording using the mode of recording
A possibility that.But in such a way that this kind dependence background sound is come secondary verifying, the complexity of user's checking is not only increased, and
Can not directly it judge with the presence or absence of recording, also, user context sound may change the previous second with next second generation conspicuousness,
To cause recognition failures.Moreover, usually requirement of real-time is relatively high in verification process, it is desirable that program has quick response
Speed.
Therefore, it is necessary to a kind of technologies for being able to detect voice or machine sound, so that Application on Voiceprint Recognition speaker verification's function
Speaker verification can more accurately be completed.
Summary of the invention
The purpose of the present invention is to provide a kind of sides that voice or the progress identity veritification of machine sound are detected by vocal print engine
Method, to solve the problems mentioned in the above background technology.
To achieve the above object, the invention provides the following technical scheme:
A method of voice is detected by vocal print engine or machine sound carries out identity veritification, comprising the following steps:
A, audio to be verified is received；
B, received audio-frequency information is pre-processed；Pretreatment comprises the steps of: a) duration detection；B) mute inspection
It surveys；C) preemphasis；D) framing adding window；
C, vocal print comparison is carried out, judges identity information.
As further technical solution of the present invention: the step a) is specifically: according to incoming voice, judging voice duration
Whether satisfaction setting detection duration requires, and is unsatisfactory for duration requirement, directly return error message.
As further technical solution of the present invention: the step b) is specifically: whether the decibel value of the incoming voice of detection
Greater than given threshold, when all audio point decibel values are both less than given threshold, error message is returned to.
As further technical solution of the present invention: the step c) is realized by a high-pass filter, is transmitted
Function are as follows: H (z)=1- μ z-1.Wherein, z indicates that timing node, H (z) are indicated to moment z aggravated consequence.
As further technical solution of the present invention: the step d) utilizes the window function of transportable specific length
Weighting realize, using Hamming window as window function, function expression are as follows:Wherein, N) table] show length of window, (n is wherein) takes
Break number, w (x) indicates the amplitude of the selected right break n of length of window.
As further technical solution of the present invention: when vocal print engine, which opens, whether detects machine sound function switch
Wait, vocal print engine after the completion of being pre-processed to the audio that receives can asynchronous comparison carry out vocal print and compare and whether be that machine sound is tested
Two steps are demonstrate,proved, wherein asynchronous comparison carries out above-mentioned two step simultaneously with regard to the mode using multithreading, when vocal print engine is opened
When whether detecting machine sound function switch, using asynchronous system, speech samples are obtained from sound-groove model database respectively
Characteristic model, and extract speech feature vector；And noise characteristic model is obtained, and extract the optimal classification surface of noise.
As further technical solution of the present invention: the voice print verification is specifically: it carries out in speaker's comparison process, it is first
First vocal print engine carries out vocal print feature extraction operation to by the pretreated audio received, then passes through the side of pattern match
Whether formula is come to verify audio and the sample voice model to be detected that vocal print engine receives be the same person, most by applied probability
Bigization matching algorithm carrys out implementation pattern matching, and recognition function is as follows:Wherein, n* is the speaker identified, and N indicates speaker
, λnFor the GMM model of n-th of speaker, z indicates observation vector.Pass through the feature extracted to voice signal to be identified
It is carried out characteristic probability matching with sound-groove model feature vector by parameter, and the maximum identity of matching probability is recognition result.
As further technical solution of the present invention: it carries out in machine sound detection process, it first can be to by pretreated
The audio received extracts channelling mode noise, then by using Legendre multinomial coefficient and extraction statistical nature
The mode of value extract audio to be verified it is long when characteristic vector, finally by svm classifier know method for distinguishing, judge acoustic to be checked
Whether frequency is machine sound, and Legendre fitting of a polynomial expression formula is as follows:Wherein LnFor
Legendre multinomial coefficient, PnFor Legnedre polynomial, it is defined as follows,By
In equipment generate interchannel noise with the presence of voice signal and presence and transformation it is very slow, using statistical characteristics come table
The statistics frame characteristic information for showing the interchannel noise that equipment generates, finally, by using the feature of both the above channelling mode noise
Value as the characteristic vector for judging whether it is voice playback, thus when completing long characteristic vector extraction, svm classifier identification side
Method is the noise optimal classification surface that will be extracted in sound-groove model database according to svm classifier function, the sound to be verified with extraction
Frequency it is long when characteristic parameter carry out Classification and Identification.
Compared with prior art, the invention has the following advantages: the present invention detects people by way of vocal print engine
Sound or machine sound carry out identity veritification.Wherein the vocal print engine can efficiently, fast and accurately recognition detection sound is speaker
Expression in real time, or by the recording of audio player plays, and judge whether the sound is me, so that vocal print
There is higher accuracy rate and efficiency in the application scenarios of identification, there are wide application scenarios.
Detailed description of the invention
Fig. 1 is vocal print engine identification process figure.
Fig. 2 is pretreatment process figure.
Fig. 3 is feature vector flow chart when extracting long.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
Fig. 1-3 is please referred to, a method of by vocal print engine detection voice or the progress identity veritification of machine sound, including with
Lower step:
A, audio to be verified is received；
B, received audio-frequency information is pre-processed；Pretreatment comprises the steps of: a) duration detection；B) mute inspection
It surveys；C) preemphasis；D) framing adding window；
C, vocal print comparison is carried out, judges identity information.
In the present invention, raw tone be exactly system directly utilize the acquisition equipment of signal to the pronunciation information of speaker into
Row acquisition, and require no intermediate other equipment.Voice playback refers to that the voice of speaker is acquired by sound pick-up outfit, is
What system acquired is speaker's pronunciation information in recording file.Since voice playback is undergone than raw tone more when acquisition
The channel of sound pick-up outfit, extra channel can increase the interchannel noise information of playback apparatus in voice segments.
In sound-groove model library, there is corresponding sound for speaker's speech model feature and environmental noise model is possessed
Line model.
When vocal print engine receives an audio to be verified, the audio can be pre-processed first, implementation process is such as
Fig. 2；
Duration detection is specifically: according to incoming voice, judging whether voice duration meets setting detection duration requirement, is discontented with
What sufficient duration required, directly return error message.
Mute detection is specifically: whether the decibel value for detecting incoming voice is greater than given threshold, when all audio point decibels
When value is both less than given threshold, error message is returned.
Preemphasis: to solve the problems, such as the reduction of the high fdrequency component due to caused by attenuation effect, in the frequency of analysis voice signal
High fdrequency component is promoted before spectrum information, keeps the variation between the high frequency spectrum of signal and low-frequency spectra flat, improves language
The resolution ratio of the high fdrequency component of sound signal is joined by preemphasis treated voice signal is more advantageous to spectrum analysis or channel
Several analyses.This process realized by a high-pass filter, transfer function are as follows: H (z)=1- μ z-1。
Framing adding window: mainly determining the size that frame length and frame move, and the voice in a frame length is referred to as a speech frame,
The unit that frame length and frame move is millisecond (ms).This patent utilizes the weighting of the window function of transportable specific length to realize,
Using Hamming window as window function, function expression are as follows:
When vocal print engine, which opens, whether detects machine sound function switch, vocal print engine is pre- to the audio received
It handles the asynchronous progress speaker comparison of meeting after the completion, whether be that machine sound verifies two steps.
It carries out in speaker's comparison process, vocal print engine is special to vocal print is carried out by the pretreated audio received first
Extraction operation is levied, audio and the sample voice mould to be detected that vocal print engine receives then is verified by way of pattern match
Whether type is the same person.
The problem of matching algorithm is come implementation pattern matching is maximized by applied probability, to realize the identification function of module
Can, recognition function is as follows:
By the characteristic parameter extracted to voice signal to be identified, it is subjected to feature with sound-groove model feature vector
Probability match, the maximum identity of matching probability is recognition result.
It carries out in machine sound detection process, channelling mode can be extracted to by the pretreated audio received first
Noise, then by using Legendre multinomial coefficient and extract statistical characteristics mode extract audio to be verified it is long when
Characteristic vector knows method for distinguishing finally by svm classifier, judges whether audio to be detected is machine sound, and Legendre is multinomial
The fitting expression of formula is as follows:Wherein LnFor Legendre multinomial coefficient, PnIt is multinomial for Legendre
Formula is defined as follows,Since the interchannel noise of equipment generation is with voice signal
In the presence of and presence and transformation it is very slow, indicated using statistical characteristics equipment generate interchannel noise statistics frame feature letter
Breath, finally, by using the characteristic value of both the above channelling mode noise as the characteristic vector for judging whether it is voice playback,
The extraction of characteristic vector when to complete long.
Judge that an input voice whether be voice playback is substantially two points of problems.And SVM is one good
The realization of two sorting algorithms, it is a kind of statistical method of structuring risk.
Embodiment 2, on the basis of embodiment 1, it is according to svm classifier function, by vocal print mould that svm classifier, which knows method for distinguishing,
The noise optimal classification surface extracted in type database, characteristic parameter carries out classification knowledge when long with the audio to be verified of extraction
Not.
It is obvious to a person skilled in the art that invention is not limited to the details of the above exemplary embodiments, Er Qie
In the case where without departing substantially from spirit or essential attributes of the invention, the present invention can be realized in other specific forms.Therefore, no matter
From the point of view of which point, the present embodiments are to be considered as illustrative and not restrictive, and the scope of the present invention is by appended power
Benefit requires rather than above description limits, it is intended that all by what is fallen within the meaning and scope of the equivalent elements of the claims
Variation is included within the present invention.Any reference signs in the claims should not be construed as limiting the involved claims.
In addition, it should be understood that although this specification is described in terms of embodiments, but not each embodiment is only wrapped
Containing an independent technical solution, this description of the specification is merely for the sake of clarity, and those skilled in the art should
It considers the specification as a whole, the technical solutions in the various embodiments may also be suitably combined, forms those skilled in the art
The other embodiments being understood that.
Claims (8)
1. a kind of detect the method that voice or machine sound carry out identity veritification by vocal print engine, which is characterized in that including following
Step:
A, audio to be verified is received；
B, received audio-frequency information is pre-processed；Pretreatment comprises the steps of: a) duration detection；B) mute detection；c)
Preemphasis；D) framing adding window；
C, vocal print comparison is carried out, judges identity information.
2. a kind of method that voice or the progress identity veritification of machine sound are detected by vocal print engine according to claim 1,
It is characterized in that, the step a) is specifically: according to incoming voice, judging whether voice duration meets setting detection duration and want
It asks, is unsatisfactory for duration requirement, directly return error message.
3. a kind of method that voice or the progress identity veritification of machine sound are detected by vocal print engine according to claim 1,
It is characterized in that, the step b) is specifically: whether the decibel value for detecting incoming voice is greater than given threshold, when all audio points
When decibel value is both less than given threshold, error message is returned.
4. a kind of method that voice or the progress identity veritification of machine sound are detected by vocal print engine according to claim 1,
It is characterized in that, the step c) is realized by a high-pass filter, transfer function are as follows:
H (z)=1- μ z-1
Wherein, z indicates that timing node, H (z) are indicated to moment z aggravated consequence.
5. a kind of method that voice or the progress identity veritification of machine sound are detected by vocal print engine according to claim 1,
It is characterized in that, the step d) is realized using the weighting of the window function of transportable specific length, with Hamming window work
For window function, function expression are as follows:
Wherein, N indicates length of window, and n is the break number wherein taken, and w (x) indicates the amplitude of the selected right break n of length of window.
6. a kind of method that voice or the progress identity veritification of machine sound are detected by vocal print engine according to claim 1,
It is characterized in that, vocal print engine is to the sound received when vocal print engine opens and whether detects machine sound function switch
Whether the asynchronous comparison progress vocal print of meeting compares and is machine sound two steps of verifying after the completion for frequency pretreatment, wherein asynchronous compare just
Above-mentioned two step is carried out simultaneously using the mode of multithreading, when vocal print engine, which opens, whether detects machine sound function switch
Wait, using asynchronous system, respectively from sound-groove model database obtain speech samples characteristic model, and extract phonetic feature to
Amount；And noise characteristic model is obtained, and extract the optimal classification surface of noise.
7. a kind of method that voice or the progress identity veritification of machine sound are detected by vocal print engine according to claim 6,
It is characterized in that, the voice print verification is specifically: carrying out in speaker's comparison process, vocal print engine is to by pretreated first
The audio received carries out vocal print feature extraction operation, then verifies what vocal print engine received by way of pattern match
Whether audio and sample voice model to be detected is the same person, maximizes matching algorithm by applied probability come implementation pattern
Match, recognition function is as follows:
Wherein, n* is the speaker identified, and N indicates speaker, λnFor the GMM model of n-th of speaker, z indicates observation arrow
Amount；
By the characteristic parameter extracted to voice signal to be identified, it is subjected to characteristic probability with sound-groove model feature vector
Matching, the maximum identity of matching probability is recognition result.
8. a kind of method that voice or the progress identity veritification of machine sound are detected by vocal print engine according to claim 6,
It is characterized in that, carrying out in machine sound detection process, channel can be extracted to by the pretreated audio received first
Then modal noise extracts audio to be verified by using the mode of Legendre multinomial coefficient and extraction statistical characteristics
Characteristic vector when long knows method for distinguishing finally by svm classifier, judges whether audio to be detected is machine sound, Legendre
Fitting of a polynomial expression formula is as follows:Wherein LnFor Legendre multinomial coefficient, PnIt is more for Legendre
Item formula, is defined as follows,Since the interchannel noise of equipment generation is with voice signal
Presence and presence and transformation it is very slow, indicated using statistical characteristics equipment generate interchannel noise statistics frame feature
Information, finally, by using the characteristic value of both the above channelling mode noise as the Characteristic Vectors for judging whether it is voice playback
Amount, thus when completing long characteristic vector extraction, svm classifier know method for distinguishing be according to svm classifier function, by sound-groove model number
According to the noise optimal classification surface extracted in library, characteristic parameter carries out Classification and Identification when long with the audio to be verified of extraction.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910765307.3A CN110459226A (en) | 2019-08-19 | 2019-08-19 | A method of voice is detected by vocal print engine or machine sound carries out identity veritification |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910765307.3A CN110459226A (en) | 2019-08-19 | 2019-08-19 | A method of voice is detected by vocal print engine or machine sound carries out identity veritification |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110459226A true CN110459226A (en) | 2019-11-15 |
Family
ID=68487746
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910765307.3A Pending CN110459226A (en) | 2019-08-19 | 2019-08-19 | A method of voice is detected by vocal print engine or machine sound carries out identity veritification |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110459226A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112309404A (en) * | 2020-10-28 | 2021-02-02 | 平安科技（深圳）有限公司 | Machine voice identification method, device, equipment and storage medium |
| CN113035230A (en) * | 2021-03-12 | 2021-06-25 | 北京百度网讯科技有限公司 | Authentication model training method and device and electronic equipment |
| CN113053397A (en) * | 2021-03-04 | 2021-06-29 | 常州分音塔科技有限公司 | Recording attack prevention identity authentication method, device and system |
| CN113515664A (en) * | 2021-09-14 | 2021-10-19 | 北京远鉴信息技术有限公司 | Abnormal audio determining method and device, electronic equipment and readable storage medium |
| CN113539298A (en) * | 2021-07-19 | 2021-10-22 | 中通服咨询设计研究院有限公司 | Sound big data analysis calculates imaging system based on cloud limit end |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102436810A (en) * | 2011-10-26 | 2012-05-02 | 华南理工大学 | Record replay attack detection method and system based on channel mode noise |
| CN106409298A (en) * | 2016-09-30 | 2017-02-15 | 广东技术师范学院 | Identification method of sound rerecording attack |
| CN108039176A (en) * | 2018-01-11 | 2018-05-15 | 广州势必可赢网络科技有限公司 | A kind of voiceprint authentication method, device and the access control system of anti-recording attack |
| CN108877813A (en) * | 2017-05-12 | 2018-11-23 | 阿里巴巴集团控股有限公司 | The methods, devices and systems of man-machine identification |
| CN109243487A (en) * | 2018-11-30 | 2019-01-18 | 宁波大学 | A kind of voice playback detection method normalizing normal Q cepstrum feature |
| CN109346089A (en) * | 2018-09-27 | 2019-02-15 | 深圳市声扬科技有限公司 | Living body identity identifying method, device, computer equipment and readable storage medium storing program for executing |
-
2019
- 2019-08-19 CN CN201910765307.3A patent/CN110459226A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102436810A (en) * | 2011-10-26 | 2012-05-02 | 华南理工大学 | Record replay attack detection method and system based on channel mode noise |
| CN106409298A (en) * | 2016-09-30 | 2017-02-15 | 广东技术师范学院 | Identification method of sound rerecording attack |
| CN108877813A (en) * | 2017-05-12 | 2018-11-23 | 阿里巴巴集团控股有限公司 | The methods, devices and systems of man-machine identification |
| CN108039176A (en) * | 2018-01-11 | 2018-05-15 | 广州势必可赢网络科技有限公司 | A kind of voiceprint authentication method, device and the access control system of anti-recording attack |
| CN109346089A (en) * | 2018-09-27 | 2019-02-15 | 深圳市声扬科技有限公司 | Living body identity identifying method, device, computer equipment and readable storage medium storing program for executing |
| CN109243487A (en) * | 2018-11-30 | 2019-01-18 | 宁波大学 | A kind of voice playback detection method normalizing normal Q cepstrum feature |
Non-Patent Citations (2)
| Title |
|---|
| 代亚丽: "防录音回放攻击的说话人认证算法及系统设计", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
| 陈亚楠: "语音回放攻击检测关键技术研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112309404A (en) * | 2020-10-28 | 2021-02-02 | 平安科技（深圳）有限公司 | Machine voice identification method, device, equipment and storage medium |
| CN112309404B (en) * | 2020-10-28 | 2024-01-19 | 平安科技（深圳）有限公司 | Machine voice authentication method, device, equipment and storage medium |
| CN113053397A (en) * | 2021-03-04 | 2021-06-29 | 常州分音塔科技有限公司 | Recording attack prevention identity authentication method, device and system |
| CN113035230A (en) * | 2021-03-12 | 2021-06-25 | 北京百度网讯科技有限公司 | Authentication model training method and device and electronic equipment |
| CN113539298A (en) * | 2021-07-19 | 2021-10-22 | 中通服咨询设计研究院有限公司 | Sound big data analysis calculates imaging system based on cloud limit end |
| CN113539298B (en) * | 2021-07-19 | 2023-11-14 | 中通服咨询设计研究院有限公司 | Sound big data analysis and calculation imaging system based on cloud edge end |
| CN113515664A (en) * | 2021-09-14 | 2021-10-19 | 北京远鉴信息技术有限公司 | Abnormal audio determining method and device, electronic equipment and readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kamble et al. | Advances in anti-spoofing: from the perspective of ASVspoof challenges | |
| Hansen et al. | Speaker recognition by machines and humans: A tutorial review | |
| CN110459226A (en) | A method of voice is detected by vocal print engine or machine sound carries out identity veritification | |
| RU2161336C2 (en) | System for verifying the speaking person identity | |
| Wu et al. | A study on spoofing attack in state-of-the-art speaker verification: the telephone speech case | |
| Singh et al. | Applications of speaker recognition | |
| Bigun et al. | Multimodal biometric authentication using quality signals in mobile communications | |
| CN105933323B (en) | Voiceprint registration, authentication method and device | |
| Huang et al. | Audio replay spoof attack detection by joint segment-based linear filter bank feature extraction and attention-enhanced DenseNet-BiLSTM network | |
| Patil et al. | A survey on replay attack detection for automatic speaker verification (ASV) system | |
| CN105933272A (en) | Voiceprint recognition method capable of preventing recording attack, server, terminal, and system | |
| EP1569200A1 (en) | Identification of the presence of speech in digital audio data | |
| CN110299142A (en) | A kind of method for recognizing sound-groove and device based on the network integration | |
| Tan et al. | A survey on presentation attack detection for automatic speaker verification systems: State-of-the-art, taxonomy, issues and future direction | |
| Paul et al. | Countermeasure to handle replay attacks in practical speaker verification systems | |
| CN108985776A (en) | Credit card security monitoring method based on multiple Information Authentication | |
| CN113823293B (en) | Speaker recognition method and system based on voice enhancement | |
| Orken et al. | Development of security systems using DNN and i & x-vector classifiers | |
| Charisma et al. | Speaker recognition using mel-frequency cepstrum coefficients and sum square error | |
| CN112349267B (en) | Synthesized voice detection method based on attention mechanism characteristics | |
| CN110415707B (en) | Speaker recognition method based on voice feature fusion and GMM | |
| Saleema et al. | Voice biometrics: the promising future of authentication in the internet of things | |
| Kounoudes et al. | Voice biometric authentication for enhancing Internet service security | |
| Sukor et al. | Speaker identification system using MFCC procedure and noise reduction method | |
| CN113012684B (en) | Synthesized voice detection method based on voice segmentation |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191115 |