CN110211606A - A kind of Replay Attack detection method of voice authentication system - Google Patents
A kind of Replay Attack detection method of voice authentication system Download PDFInfo
- Publication number
- CN110211606A CN110211606A CN201910303649.3A CN201910303649A CN110211606A CN 110211606 A CN110211606 A CN 110211606A CN 201910303649 A CN201910303649 A CN 201910303649A CN 110211606 A CN110211606 A CN 110211606A
- Authority
- CN
- China
- Prior art keywords
- sequence
- voice
- signal
- sampled
- sampled point
- 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.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 20
- 239000000284 extract Substances 0.000 claims abstract description 8
- 238000000034 method Methods 0.000 claims description 11
- 238000005070 sampling Methods 0.000 claims description 8
- 230000000694 effects Effects 0.000 claims description 7
- 108010076504 Protein Sorting Signals Proteins 0.000 claims description 6
- 239000013074 reference sample Substances 0.000 claims description 6
- 230000005236 sound signal Effects 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 3
- 230000001755 vocal effect Effects 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 241000209140 Triticum Species 0.000 description 2
- 235000021307 Triticum Nutrition 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 239000012528 membrane Substances 0.000 description 2
- 230000037361 pathway Effects 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000011982 device technology Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000005674 electromagnetic induction Effects 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000000523 sample Substances 0.000 description 1
- 238000013456 study Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 210000001260 vocal cord Anatomy 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
- G10L25/00—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00
- G10L25/48—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use
- G10L25/51—Speech or voice analysis techniques not restricted to a single one of groups G10L15/00 - G10L21/00 specially adapted for particular use for comparison or discrimination
Landscapes
- Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Signal Processing (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
- Collating Specific Patterns (AREA)
- Lock And Its Accessories (AREA)
Abstract
The invention discloses a kind of Replay Attack detection methods based on the polar voice authentication system of voice signal time domain.Pass through voice authentication system acquisition and recording voice signal, extract the positive signal and minus polarity signal of voice signal, the proportionate relationship judgement for comparing positive signal and minus polarity signal obtains voice signal and belongs to Replay Attack or living body voice: if positive-negative polarity fraction gap is larger and positive signal ratio is higher than minus polarity signal ratio, then it is assumed that be Replay Attack；If positive-negative polarity fraction gap is larger and positive signal ratio is not higher than minus polarity signal ratio, then it is assumed that be living body voice.The present invention can accurately and effectively detect the Replay Attack in voice authentication system.
Description
Technical field
The invention belongs to voice authentication technology and security technology areas, and in particular to one kind is detectable to be directed to voice authentication system
The software processing method of the Replay Attack of system.
Background technique
Voice authentication system is a kind of using voice authentication technology extraction speaker's voice specific characteristics, passes through voice spy
Sign pattern match is to identify the security certification system of speaker's identity.Since it is low to hardware requirement, inexpensive, certification is simple
It is convenient, the characteristics of remote contactless certification can be carried out, be increasingly becoming a kind of mainstream user's certification and access control side
Formula.However, existing voice Verification System, is generally subject to Replay Attack.
Refer to that attacker prerecords the true legitimate user's speech samples of collection for the Replay Attack of voice authentication system
Segment, by it directly or through splicing after, broadcasted by loudspeaker, to cheat voice authentication system.Replay Attack does not need to attack
It hits promoter and grasps Speech processing knowledge, and with the development of electronic device technology, the loudspeaking of high quality and low cost
Device has become more common, these factors all make Replay Attack become but prestige most simple for voice authentication system
Coerce maximum attack；But meanwhile Replay Attack extremely difficult be found, defend again.
Detect and defend Replay Attack, it is to be understood that the sound-electric and electro-acoustic of microphone and loudspeaker transformation mechanism.Wheat
Gram wind, loudspeaker etc. are for sound wave-electromagnetic signal conversion converter.Microphone passes through sound wave bring vibration of thin membrane, benefit
With faraday's electromagnetic induction effect, vibration mechanical energy is converted to the electric energy of electric signal；Loudspeaker is then by this electric signal computer
It is reversely converted into the kinetic energy of film, so that film disturbance air is formed sound wave, and then restore the sound being converted into before electric signal.
Ideally, microphone and loudspeaker are converted to complete reciprocal process, i.e., in following Fig. 1, acoustical signal 1 should be with
Acoustical signal 2 is identical.But in the realistic case, both signals are often different.Lead to the main reason for distinguishing between the two
There is two o'clock: 1) in the pathway for electrical signals of microphone and loudspeaker, just like power amplifier, input and output filter, ad/da converter etc.
Circuit can introduce noise into electric signal；2) in vibrating membrane vibration realizing electricity-sound and sound-electricity conversion, a variety of mechanical resistances
Power will cause the variation of its motor pattern, cause conversion front and back signal inconsistent.
Since in Replay Attack, voice signal (being here the abstract summation of acoustical signal and electric signal) goes out from by human hair
To before being received by voice authentication system microphone, certification is directly carried out compared with living body user and has additionally gone through one group of wheat wind-loudspeaker
Hardware is attacked, therefore the voice signal of Replay Attack will change band comprising more noises and by vibrating diaphragm motor pattern compared with living body authentication
The distortion come.By detecting these distortions, it can theoretically detect, defend Replay Attack.
There are many correlative studys at present introduces noise by detection attack hardware to detect Replay Attack.This kind detection
The characteristics of method is usually had Detection accuracy lower and is affected by Replay Attack using microphone and loudspeaker quality.So
And do not have also on research concern attack device hardware access by the variation bring voice signal distortion of vibrating diaphragm motor pattern.
Summary of the invention
To solve technical problem present in above-mentioned background technique, the present invention provides one kind to be based on voice signal time domain pole
Property voice authentication system Replay Attack detection method, be collected into the time domain pole of voice signal by detecting voice authentication system
Property feature can accurately and effectively detect Replay Attack.
The present invention adopts the following technical scheme:
The present invention extracts the positive signal and cathode of voice signal by voice authentication system acquisition and recording voice signal
Property signal, the proportionate relationship judgement for comparing positive signal and minus polarity signal obtains voice signal and belongs to Replay Attack (recording
The sound that equipment issues) or living body voice (i.e. the sound of living body user sending):
If positive-negative polarity fraction gap is larger and positive signal ratio is higher than minus polarity signal ratio, then it is assumed that be
Replay Attack；
If positive-negative polarity fraction gap is larger and positive signal ratio is not higher than minus polarity signal ratio, then it is assumed that
It is living body voice.
The method is specific as follows:
1) speech activity inspection is carried out by the voice signal that the acquisition of certain sample frequency interval is collected into voice authentication system
It surveys, removes the noise in voice signal, extract a part in voice audio signals as pure vocal sections；
The voice activity detection that the method for the present invention uses passes through signal amplitude and duration mainly to judge specified section language
Sound signal is pure voice or noise.
2) polarity index calculating is carried out to the pure human voice signal of obtained time domain:
Pure human voice signal sequence S is the sequence comprising N number of sampled point, all sampled points that wherein sampled value is positive
Number is Npos, the absolute value of the sum of sampled value of all sampled points that sampled value is positive is | Sumpos|, sampled value is negative all
Number of sampling points is Nneg, the absolute value of the sum of sampled value of all sampled points that sampled value is negative is | Sumneg|, use is following
Formula manipulation obtains polarity number I:
I=(| Sumpos|/Npos)/(|Sumpos|/Npos+|Sumneg|/Nneg)
3) by obtained polarity number I and default polarity thresholds IthrCompare: when polarity number I is greater than polarity thresholds Ithr, sentence
Break as living body voice；Otherwise, it is judged as Replay Attack.
The step 1) specifically:
1.1) voice signal Sa is the sequence comprising Na sampled point, and the maximum value of all sampled point absolute values is | Amax
|, setting signal amplitude thresholds | Athr |=0.1 × | Amax |；
1.2) extract voice signal Sa in all sampled value absolute values be greater than signal amplitude threshold value | Athr | groups of samples
At First ray (Sai1,Sai2,Sai3,...Saix), and have 1≤i1<i2<i3<...<ix≤ N, i are that sampled point is believed in voice
Index numerical sequence in number Sa sequence, N indicate the sum of sampled point in voice signal Sa sequence；
1.3) to First ray (Sai1,Sai2,Sai3,...Saix) in, initially with i-th1A sampled point is as reference sample
Point, first from i-th1The index numerical sequence of a sampled point starts to traverse the index numerical sequence for finding each sampled point backward: if i-thpIt is a
The index numerical sequence of sampled point and i-th(p-1)The difference of the index numerical sequence of a sampled point is greater than default ordinal number threshold value D1, then by
ip-1A sampled point and i-th1First ray (Sa between a sampled pointi1,Sai2,Sai3,...Saix) in all groups of samples
At the 1st sequence of subsets Ssub1；
1.4) then from i-thpA sampled point is constantly repeated the above steps backward as beginning 1.3), by i-thq(q >=p) is a
Sampled point and its before closest to reference sample point between First ray (Sai1,Sai2,Sai3,...Saix) in all adopt
Sampling point forms next sequence of subsets, until traversal arrives last SaixA sampled point finally obtains y-th of sequence of subsets
Ssuby；
1.5) for the 1st sequence of subsets Ssub1 to y-th sequence of subsets Ssuby (y >=1), judge each sequence of subsets
The difference of largest index numerical sequence and minimum index numerical sequence that wherein whether each sampled point meets sampled point is greater than default index
Threshold value D2, the difference of all largest index numerical sequences for meeting sampled point and minimum index numerical sequence finally will be greater than default index
Threshold value D2Sequence of subsets merge become pure human voice signal sequence S.
Present invention discover that, since human vocal cord vibration beep pattern is relatively fixed, Verification System is direct in living body authentication
The living body voice recorded is presented that signal positive-negative polarity fraction gap is larger and positive signal ratio is higher than negative polarity substantially
The characteristics of signal proportion.
And in Replay Attack, due to attacking device hardware access bring diaphragm oscillations patterns of change, voice signal
The characteristics of basic positive-negative polarity fraction that presents is suitable, and even minus polarity signal ratio is higher than positive signal.
The present invention is the ratio for passing through the positive-negative polarity signal of voice signal collected by detection voice authentication system hardware
Compared with (time domain polarity), it simply but can effectively judge this voice signal from living body speaker or Replay Attack loudspeaking
Device.
The beneficial effects of the present invention are:
The present invention under conditions of only handling voice authentication time-domain signal, realize detection to Replay Attack with
Defence.Since method is very simple effective, processing step is few, and algorithm complexity is low, and the present invention has colleges and universities and is delayed low
Advantage；Simultaneously as object detected is unrelated with mixed noise in microphone and loudspeaker pathway for electrical signals, this method inspection
Survey success rate is not influenced using microphone with loudspeaker tonequality by Replay Attack institute, i.e. the loudspeaker to different quality class and wheat
The attack that gram wind is initiated has same protection effect.
The present invention can accurately and effectively detect the Replay Attack in voice authentication system.
Detailed description of the invention
Fig. 1 is the conversion process schematic diagram of ideally microphone and loudspeaker.
Fig. 2 is detection method flow chart of the invention.
Fig. 3 is the Speech signal detection figure of embodiment.
Specific embodiment
Present invention will be further explained below with reference to the attached drawings and examples.
Specific implementation process of the present invention is as follows:
1) voice signal being collected into the acquisition of voice authentication system interval carries out voice activity detection, removes voice signal
In noise, extract voice audio signals in a part as pure vocal sections；
1.1) voice signal Sa is the sequence comprising Na sampled point, and the maximum value of all sampled point absolute values is | Amax
|, setting signal amplitude thresholds | Athr |=0.1 × | Amax |；
1.2) extract voice signal Sa in all sampled value absolute values be greater than signal amplitude threshold value | Athr | groups of samples
At First ray (Sai1,Sai2,Sai3,...Saix), Sai1,Sai2,Sai3,...SaixRespectively indicate i-th1A sampled point is to i-thx
The sampled value of a sampled point, and have 1≤i1<i2<i3<...<ix≤ N, i are index of the sampled point in voice signal Sa sequence
Numerical sequence, N indicate the sum of sampled point in voice signal Sa sequence；
1.3) to First ray (Sai1,Sai2,Sai3,...Saix) in, initially with i-th1A sampled point is as reference sample
Point, first from i-th1The index numerical sequence of a sampled point starts to traverse the index numerical sequence for finding each sampled point backward: if i-thpIt is a
The index numerical sequence of sampled point and i-th(p-1)The difference of the index numerical sequence of a sampled point is greater than default ordinal number threshold value D1, then by
ip-1A sampled point and i-th1First ray (Sa between a sampled pointi1,Sai2,Sai3,...Saix) in all groups of samples
At the 1st sequence of subsets Ssub1；
1.4) then from i-thpA sampled point is constantly repeated the above steps backward as beginning 1.3), by i-thq(q >=p) is a
Sampled point and its before closest to reference sample point between First ray (Sai1,Sai2,Sai3,...Saix) in all adopt
Sampling point forms next sequence of subsets, until traversal arrives last SaixA sampled point finally obtains y-th of sequence of subsets
Ssuby；
1.5) for the 1st sequence of subsets Ssub1 to y-th sequence of subsets Ssuby (y >=1), judge each sequence of subsets
The difference of largest index numerical sequence and minimum index numerical sequence that wherein whether each sampled point meets sampled point is greater than default index
Threshold value D2, the difference of all largest index numerical sequences for meeting sampled point and minimum index numerical sequence finally will be greater than default index
Threshold value D2Sequence of subsets merge become pure human voice signal sequence S.
2) polarity index calculating is carried out to the pure human voice signal of obtained time domain:
Pure human voice signal sequence S is the sequence comprising N number of sampled point, all sampled points that wherein sampled value is positive
Number is Npos, the absolute value of the sum of sampled value of all sampled points that sampled value is positive is | Sumpos|, sampled value is negative all
Number of sampling points is Nneg, the absolute value of the sum of sampled value of all sampled points that sampled value is negative is | Sumneg|, use is following
Formula manipulation obtains polarity number I:
I=(| Sumpos|/Npos)/(|Sumpos|/Npos+|Sumneg|/Nneg)
3) by obtained polarity number I and default polarity thresholds IthrCompare: when polarity number I is greater than polarity thresholds Ithr, i.e. I
>IthrWhen, it is believed that voice signal meets living body user voice signal polarity feature, is judged as living body voice；Otherwise, judgement is attached most importance to
Put attack.
Embodiment one:
In Fig. 3, upper channel is the living body authentication voice signal that voice authentication system obtains, and lower channel is with HiVi sound equipment
The voice signal that Replay Attack obtains.It is obvious that the positive sex ratio of living body voice signal is much higher than negative polarity ratio
Example, and Replay Attack signal is then just the opposite.At this detection method first two steps (voice activity detection, polarity index calculate)
After reason, it is 0.583 that living body authentication voice signal polarity index, which can be calculated, hence it is evident that greater than the polarity of Replay Attack voice signal
Index is 0.494.
Embodiment two:
The present embodiment acquires the living body authentication voice of totally 20 people (14 male 6 female), and with including aforementioned HiVi sound equipment
8 kinds of quality loudspeaker distributed more widely carry out Replay Attack.Decision threshold is set to be 0.52, i.e., is greater than polarity index
0.52 voice is determined as living body voice, anyway be determined as Replay Attack, obtains to living body speech detection accuracy rate 93.2%,
To playback attack detecting accuracy rate 96.5%.
Claims (3)
1. a kind of Replay Attack detection method of voice authentication system, it is characterised in that: pass through voice authentication system acquisition and recording
Voice signal extracts the positive signal and minus polarity signal of voice signal, compares the ratio of positive signal and minus polarity signal
Example relationship judgement obtains voice signal and belongs to Replay Attack or living body voice: if positive-negative polarity fraction gap is larger and just
Polar signal ratio is higher than minus polarity signal ratio, then it is assumed that is Replay Attack；If positive-negative polarity fraction gap it is larger and
Positive signal ratio is not higher than minus polarity signal ratio, then it is assumed that is living body voice.
2. a kind of Replay Attack detection method of voice authentication system according to claim 1, it is characterised in that: method tool
Body is as follows:
1) voice signal being collected into the acquisition of voice authentication system interval carries out voice activity detection, removes in voice signal
Noise extracts a part in voice audio signals as pure vocal sections；
2) polarity index calculating is carried out to the pure human voice signal of obtained time domain:
Pure human voice signal sequence S is the sequence comprising N number of sampled point, and all number of sampling points that wherein sampled value is positive are
Npos, the absolute value of the sum of sampled value of all sampled points that sampled value is positive is | Sumpos|, all samplings that sampled value is negative
Point number is Nneg, the absolute value of the sum of sampled value of all sampled points that sampled value is negative is | Sumneg|, using following formula
Processing obtains polarity number I:
I=(| Sumpos|/Npos)/(|Sumpos|/Npos+|Sumneg|/Nneg)
3) by obtained polarity number I and default polarity thresholds IthrCompare: when polarity number I is greater than polarity thresholds Ithr, it is judged as
Living body voice；Otherwise, it is judged as Replay Attack.
3. a kind of Replay Attack detection method of voice authentication system according to claim 2, it is characterised in that:
The step 1) specifically:
1.1) voice signal Sa is the sequence comprising Na sampled point, and the maximum value of all sampled point absolute values is | Amax |, if
Confidence amplitude thresholds | Athr |=0.1 × | Amax |；
1.2) all sampled value absolute values in voice signal Sa are extracted and are greater than signal amplitude threshold value | Athr | groups of samples at the
One sequence (Sai1,Sai2,Sai3,...Saix), and have 1≤i1<i2<i3<...<ix≤ N, i are sampled point in voice signal Sa
Index numerical sequence in sequence, N indicate the sum of sampled point in voice signal Sa sequence；
1.3) to First ray (Sai1,Sai2,Sai3,...Saix) in, initially with i-th1A sampled point is as reference sample point, first
From i-th1The index numerical sequence of a sampled point starts to traverse the index numerical sequence for finding each sampled point backward: if i-thpA sampling
The index numerical sequence and i-th of point(p-1)The difference of the index numerical sequence of a sampled point is greater than default ordinal number threshold value D1, then by i-thp-1It is a
Sampled point and i-th1First ray (Sa between a sampled pointi1,Sai2,Sai3,...Saix) in all groups of samples at the 1st
A sequence of subsets Ssub1；
1.4) then from i-thpA sampled point is constantly repeated the above steps backward as beginning 1.3), by i-thq(q >=p) a sampling
Point and its before closest to reference sample point between First ray (Sai1,Sai2,Sai3,...Saix) in all sampled points
Next sequence of subsets is formed, until traversal arrives last SaixA sampled point finally obtains y-th of sequence of subsets Ssuby；
1.5) for the 1st sequence of subsets Ssub1 to y-th sequence of subsets Ssuby (y >=1), judge each sequence of subsets wherein
The difference of largest index numerical sequence and minimum index numerical sequence that whether each sampled point meets sampled point is greater than default index threshold value
D2, the difference of all largest index numerical sequences for meeting sampled point and minimum index numerical sequence finally will be greater than default index threshold value
D2Sequence of subsets merge become pure human voice signal sequence S.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910303649.3A CN110211606B (en) | 2019-04-12 | 2019-04-12 | Replay attack detection method of voice authentication system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910303649.3A CN110211606B (en) | 2019-04-12 | 2019-04-12 | Replay attack detection method of voice authentication system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110211606A true CN110211606A (en) | 2019-09-06 |
| CN110211606B CN110211606B (en) | 2021-04-06 |
Family
ID=67785410
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910303649.3A Active CN110211606B (en) | 2019-04-12 | 2019-04-12 | Replay attack detection method of voice authentication system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110211606B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111243600A (en) * | 2020-01-10 | 2020-06-05 | 浙江大学 | Voice spoofing attack detection method based on sound field and field pattern |
| WO2022052965A1 (en) * | 2020-09-10 | 2022-03-17 | 达闼机器人有限公司 | Voice replay attack detection method, apparatus, medium, device and program product |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1928991A (en) * | 2006-07-20 | 2007-03-14 | 中山大学 | Synchronous attack resistant audio frequency watermark handling method |
| JP2008058953A (en) * | 2006-07-26 | 2008-03-13 | Nec (China) Co Ltd | Media program identification method and apparatus based on audio watermarking |
| CN106297772A (en) * | 2016-08-24 | 2017-01-04 | 武汉大学 | Detection method is attacked in the playback of voice signal distorted characteristic based on speaker introducing |
| CN106531172A (en) * | 2016-11-23 | 2017-03-22 | 湖北大学 | Speaker voice playback identification method and system based on environmental noise change detection |
| CN109448759A (en) * | 2018-12-28 | 2019-03-08 | 武汉大学 | A kind of anti-voice authentication spoofing attack detection method based on gas explosion sound |
-
2019
- 2019-04-12 CN CN201910303649.3A patent/CN110211606B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1928991A (en) * | 2006-07-20 | 2007-03-14 | 中山大学 | Synchronous attack resistant audio frequency watermark handling method |
| JP2008058953A (en) * | 2006-07-26 | 2008-03-13 | Nec (China) Co Ltd | Media program identification method and apparatus based on audio watermarking |
| CN106297772A (en) * | 2016-08-24 | 2017-01-04 | 武汉大学 | Detection method is attacked in the playback of voice signal distorted characteristic based on speaker introducing |
| CN106531172A (en) * | 2016-11-23 | 2017-03-22 | 湖北大学 | Speaker voice playback identification method and system based on environmental noise change detection |
| CN109448759A (en) * | 2018-12-28 | 2019-03-08 | 武汉大学 | A kind of anti-voice authentication spoofing attack detection method based on gas explosion sound |
Non-Patent Citations (3)
| Title |
|---|
| CHEN YAN .ET AL: "The Feasibility of Injecting Inaudible Voice Commands to Voice Assistants", 《TRANSACTIONS ON DEPENDABLE AND SECURE COMPUTING》 * |
| 曹慧: "高保真录音回放攻击取证算法与仿真验证", 《安阳工学院学报》 * |
| 李亚: "语音信号幅值分布的统计分析", 《电脑知识与技术》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111243600A (en) * | 2020-01-10 | 2020-06-05 | 浙江大学 | Voice spoofing attack detection method based on sound field and field pattern |
| WO2022052965A1 (en) * | 2020-09-10 | 2022-03-17 | 达闼机器人有限公司 | Voice replay attack detection method, apparatus, medium, device and program product |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110211606B (en) | 2021-04-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110832580B (en) | Detection of replay attacks | |
| Nassi et al. | Lamphone: Real-time passive sound recovery from light bulb vibrations | |
| Wang et al. | Secure your voice: An oral airflow-based continuous liveness detection for voice assistants | |
| Wang et al. | Ghosttalk: Interactive attack on smartphone voice system through power line | |
| CN107274911A (en) | A kind of similarity analysis method based on sound characteristic | |
| CN108182418A (en) | A kind of thump recognition methods based on multidimensional acoustic characteristic | |
| CN110211606A (en) | A kind of Replay Attack detection method of voice authentication system | |
| Ganguly et al. | Real-time Smartphone implementation of noise-robust Speech source localization algorithm for hearing aid users | |
| WO2022052965A1 (en) | Voice replay attack detection method, apparatus, medium, device and program product | |
| CN111243600A (en) | Voice spoofing attack detection method based on sound field and field pattern | |
| CN110718229A (en) | Detection method for record playback attack and training method corresponding to detection model | |
| JP2000148184A (en) | Speech recognizing device | |
| Chen et al. | Push the limit of adversarial example attack on speaker recognition in physical domain | |
| Shabtai et al. | Room volume classification from room impulse response using statistical pattern recognition and feature selection | |
| Tian et al. | Spoofing detection under noisy conditions: a preliminary investigation and an initial database | |
| Shang et al. | Voice liveness detection for voice assistants through ear canal pressure monitoring | |
| CN109005023A (en) | A kind of smart phone pattern password guess method based on nearly ultrasonic wave | |
| CN104240705A (en) | Intelligent voice-recognition locking system for safe box | |
| CN103035237B (en) | Chinese speech signal processing method, device and hearing aid device | |
| Rumsey | Audio forensics: Keeping up in the age of smartphones and fakery | |
| Shi et al. | Authentication of voice commands by leveraging vibrations in wearables | |
| Wang et al. | SeVI: Boosting Secure Voice Interactions with Smart Devices | |
| CN115348049B (en) | User identity authentication method utilizing earphone inward microphone | |
| Nandyala et al. | Real time isolated word recognition using adaptive algorithm | |
| Wang et al. | Shift to Your Device: Data Augmentation for Device-Independent Speaker Verification Anti-Spoofing |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |