CN106448685A - System and method for identifying voice prints based on phoneme information - Google Patents
System and method for identifying voice prints based on phoneme information Download PDFInfo
- Publication number
- CN106448685A CN106448685A CN201610880776.6A CN201610880776A CN106448685A CN 106448685 A CN106448685 A CN 106448685A CN 201610880776 A CN201610880776 A CN 201610880776A CN 106448685 A CN106448685 A CN 106448685A
- Authority
- CN
- China
- Prior art keywords
- phoneme
- numeric string
- information
- module
- neural network
- 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
- 238000000034 method Methods 0.000 title claims abstract description 22
- 238000012549 training Methods 0.000 claims abstract description 21
- 238000013528 artificial neural network Methods 0.000 claims abstract description 13
- 230000011218 segmentation Effects 0.000 claims description 5
- 230000001755 vocal effect Effects 0.000 claims description 4
- 230000000295 complement effect Effects 0.000 claims description 2
- 208000031481 Pathologic Constriction Diseases 0.000 claims 1
- 210000001215 vagina Anatomy 0.000 claims 1
- 238000004458 analytical method Methods 0.000 abstract description 6
- 230000006870 function Effects 0.000 abstract description 5
- 230000008901 benefit Effects 0.000 abstract description 4
- 241001672694 Citrus reticulata Species 0.000 abstract 3
- 238000012795 verification Methods 0.000 description 7
- 230000008034 disappearance Effects 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 230000007935 neutral effect Effects 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 238000011161 development Methods 0.000 description 4
- 230000018109 developmental process Effects 0.000 description 4
- 238000013507 mapping Methods 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 230000004069 differentiation Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000009897 systematic effect Effects 0.000 description 3
- 238000012935 Averaging Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 2
- 230000001419 dependent effect Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000000556 factor analysis Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 238000005267 amalgamation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 239000000686 essence Substances 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000007477 logistic regression Methods 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 230000035479 physiological effects, processes and functions Effects 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 238000006467 substitution reaction 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
- G10L17/06—Decision making techniques; Pattern matching strategies
-
- 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
- G10L17/02—Preprocessing operations, e.g. segment selection; Pattern representation or modelling, e.g. based on linear discriminant analysis [LDA] or principal components; Feature selection or extraction
-
- 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
- G10L17/04—Training, enrolment or model building
-
- 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
- G10L17/18—Artificial neural networks; Connectionist approaches
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Game Theory and Decision Science (AREA)
- Business, Economics & Management (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Telephonic Communication Services (AREA)
- Machine Translation (AREA)
Abstract
The invention relates to a system and a method for identifying voice prints based on phoneme information. The system comprises a phoneme forced aligning module, a phoneme relevance model creating module and a neural network classifier module, wherein the phoneme forced aligning module is based on a Chinese mandarin voice identifier; the neural network classifier module is based on a dropout strategy. The method comprises the following steps of defining sixteen phoneme types of Chinese mandarin numeric string voice prints, and digitally utilizing each piece of pronunciation type information of a numeric string; according to the Chinese mandarin voice identifier, using a viterbi forced aligning algorithm to obtain the phoneme boundary of the text content of each numeric string; using a text irrelevance algorithm to establish a phoneme relevance model; calculating the phoneme relevance model, so as to obtain a fraction vector. The system and the method have the advantages that the functions of division of phoneme information, modeling of phonemes and analysis of distinguishing capability of the phoneme relevance model are realized, the neural network training method based on the dropout strategy is proposed, the problem of phoneme missing of the numeric string is solved, and the performance of a numeric string voice print identifying system is improved.
Description
Technical field
The present invention relates to voiceprint authentication system technical field, it particularly relates to a kind of vocal print based on phoneme information is recognized
Card system and method.
Background technology
Living things feature recognition is a kind of physiological feature intrinsic according to human body itself and behavior characteristicss recognizing identity
Technology, have the advantages that being difficult to forget, anti-counterfeiting performance is good, be difficult to forge or stolen, possess with oneself and can use whenever and wherever possible.With
The Internet fast development, traditional identity authentication techniques means cannot increasingly meet the need of user experience and security capabilities
Ask.Sound groove recognition technology in e easy to use, due to its wealthy application prospect, huge Social benefit and economic benefit, causes
The extensive concern of all trades and professions and great attention.
Application on Voiceprint Recognition, also known as Speaker Identification, is one kind of biological identification technology.The technology is by reflecting in speech waveform
Speak the speech parameter of human physiology and behavior characteristicss, and then tell speaker's identity.Convenient with safe, data acquisition
The features such as.
In recent years, the Speaker Identification of text correlation (Text-dependent) becomes the focus in user authentication field.By
In the major progress in unrelated (Text-independent) the Speaker Identification field of text, a lot of research worker are attempted by text no
Close speaker's recognizer and be applied to text association area, such as numeric string Application on Voiceprint Recognition.
Under numeric string authentication condition, have research worker using simultaneous factor analysises (Joint Factor Analysis,
JFA), gauss hybrid models-interference properties mapping (Gaussian Mixture Model-Nuisance Attribute
Projection, GMM-NAP) and HMM-interference properties mapping (Hidden Markov Model-Nuisance
Attribute Projection, HMM-NAP) it is compared.For comparing JFA, performed better than based on the algorithm of NAP, reason
It is to train JFA that substantial amounts of tagged data is needed, and exists between the training data of JFA matrix and numeric string test data and lose
Join.
In the unrelated Speaker Identification of text, JFA and be based on probability linear discriminant analysiss (Probabilistic Linear
Discriminant Analysis, PLDA) population variance modeling factors (iVector) algorithm all rely on substantial amounts of exploitation
Collection data.Increasing work is devoted to development set data in the limited field of process and asks to the migration of application data outside field
Topic, the self adaptation of such as lexical gap and backoff algorithm.
By Android system (Android) and the mobile phone of apple system (iOS), record and construct the number comprising 536 people
Word string voice set.It is divided into two kinds of scenes：Global condition and rand-n condition.Global condition represents that registration and checking are adopted
Identical numeric string content；Rand-n condition represents that each digit string is random number word string of the length for n, and this is at certain
More safer than global condition in the application system that anti-recording is attacked a bit.Be related in the present invention as shown in table 1 three kind registration/
Authentication condition：The whole numerical ciphers of fixation, dynamic 8 bit digital password and dynamic 6 bit digital password.Every kind of scene partitioning development set
With evaluation and test collection.Development set is used for training global context model (Universal Background Model, UBM), population variance
Modeling matrix (iVector T matrix) and linearly differentiation analysis matrix (Linear Discriminant Analysis,
LDA) etc..In three kinds of conditions of evaluation and test collection, everyone comprising a three registration voices and tested speech, per bar tested speech with
All speaker models are compared.
Table 1：Several form examples of password figure
Table 2 be GMM-NAP and using iVector voiceprint authentication system etc. error rate (Equal Error Rate,
EER) contrast.As a result show, with the increase of digital string length, the performance of voiceprint authentication system has obtained significantly as one man carrying
Rise.But GMM-NAP and iVector system does not all account for the utilization of phoneme (Phone/Phoneme) information, be based on text no
Close direct application of the Application on Voiceprint Recognition under text associated scenario.In numeric string vocal print application, ignore phoneme information or without sound
The effectively utilizes of prime information, it will limit the unrelated recognizer of text effect in actual applications.
Table 2：GMM-NAP and iVector system under different test conditions etc. error rate contrast
| The whole numerical ciphers of fixation | Dynamic 8 bit digital passwords | Dynamic 6 bit digital passwords | |
| GMM-NAP | 2.09% | 2.64% | 3.76% |
| iVector | 1.87% | 2.40% | 3.32% |
Content of the invention
It is an object of the invention to a kind of voiceprint authentication system based on phoneme information and method is proposed, sound can realized
While prime information cutting, phoneme model (Phone-dependent) model separating capacity analysis related with phoneme, number is solved
The problem of word string phoneme disappearance, and improve the performance of numeric string voiceprint authentication system.
For above-mentioned technical purpose is realized, the technical scheme is that and be achieved in that：
A kind of voiceprint authentication system based on phoneme information, forces including the phoneme based on Chinese putonghua speech evaluator
Alignment module, the model creation module of phoneme correlation and the neural network classifier module based on dropout strategy；
The phoneme based on Chinese putonghua speech evaluator forces alignment module to be used for 16 sounds to numeric string
Plain classification carries out cutting；
The model creation module of the phoneme correlation is used for setting up phoneme correlation model, and analyzes each phoneme correlation model
Separating capacity to voiceprint, features the differentiation feature of speaker, rather than difference between vocabulary；
The neural network classifier module based on dropout strategy is used for merging the complementary letter of phoneme correlation model
Breath.
A kind of voiceprint authentication method based on phoneme information, comprises the steps：
S01：Define 16 phoneme class of standard Chinese numeric string vocal print, explicit each pronunciation using numeric string
Classification information；
S02：Based on Chinese putonghua speech evaluator, force alignment algorithm to obtain each using Viterbi and correspond to numeric string
The phoneme boundary of content of text, completes the phone segmentation to voice content, i.e. speech feature vector to the mapping relations of phoneme, obtains
To the characteristic vector subclass for belonging to phoneme, each character subset conjunction is considered as independent data flow carries out subsequent treatment；
S03：Phoneme correlation model is set up using the unrelated algorithm of text, the model of phoneme correlation is set up process and reduces each
The parameter amount of phoneme correlation model, it is to avoid training crossed by model；
S04：Phoneme correlation model is calculated, obtains scores vector.
Further, using the dropout Strategies Training rear end integrated classification device in neural network algorithm in step S04.
Beneficial effects of the present invention：
(1) present invention forces alignment algorithm to obtain using typical Chinese putonghua speech evaluator is based on using Viterbi
Each phoneme boundary for corresponding to numeric string content of text is taken, the phone segmentation to voice content is completed, is based on compared to common
The cutting effect of dynamic time warping (Dynamic Time Warping, DTW) scheduling algorithm is advantageously；
(2) present invention defines 16 pronunciation classifications to the numeric string pronunciation of standard Chinese, it is to avoid affiliated phoneme class
Training problem crossed by the very few model for causing of characteristic vector, establishes phoneme correlation model, and analyzes each phoneme correlation model pair
The separating capacity of voiceprint, phoneme correlation model features the differentiation feature of speaker, rather than the difference between vocabulary；
(3) in order to improve the Information Pull effect of phoneme correlation model further, and certification language in practical application is considered
Sound only includes the partial content of set of phonemes, it is understood that there may be the problem of vector dimension disappearance, using dropout Strategies Training nerve
Network backend grader, realizes the amalgamation judging of phoneme associated score vector, and has been obviously improved the systematic function of voiceprint.
Description of the drawings
Fig. 1 is the rear end grader process chart of the scores vector in the present invention based on phoneme correlation；
Fig. 2 be in the present invention for different phoneme correlation models etc. error rate experimental result picture.
Specific embodiment
With reference to the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Ground description, it is clear that described embodiment is only a part of embodiment of the present invention, rather than whole embodiments.Based on this
Embodiment in invention, the every other embodiment obtained by those of ordinary skill in the art, belong to the model of present invention protection
Enclose.
The present invention is proposed phoneme information explicitly using the numeric string voiceprint authentication method for being combined with neural network classification,
For per bar digit string, alignment algorithm is forced to complete to voice content using the Viterbi of Chinese putonghua speech evaluator
Phone segmentation；Reduce the training parameter amount of phoneme correlation model, it is to avoid the training phonetic feature of each phoneme model is less can
Can caused crosses training problem, analyzes separating capacity of each phoneme model to Application on Voiceprint Recognition；Fraction to phoneme correlation model
Vector there may be the problem of dimension disappearance, using the dropout Strategies Training rear end integrated classification device in neural network algorithm,
The utilizing status of phoneme relevant information are improve, improves the systematic function of numeric string voiceprint further.
Table 3 gives the phonemic representation of ten standard Chinese numeric utterance.It is noted that digital " 1 " has " y i " and " y
Two kinds of pronunciations of ao ", therefore correspond to ten standard Chinese numeric utterance and have 16 phonemes.
Table 3：Ten digital standard Chinese pronunciation phonemes
In " the whole numerical ciphers of fixation " condition, phoneme content immobilizes." dynamic 8 bit digital password " and " dynamic 6
The phoneme content of numerical ciphers " is also known, because the random algorithm that digital text is typically based on background system is pushed or base
Generated according to special algorithm in OTP dynamic password (One-time Password).
Based on Chinese putonghua speech identifying system, force alignment algorithm to obtain each using Viterbi and correspond to content of text
Phoneme boundary, complete the phone segmentation to voice content, i.e. speech feature vector to the mapping of phoneme.
Therefore, the acoustic featuress sequence vector χ=x of a hop count word string voice is given1,...,xT, discrete son can be cut into
Set χ1,...,χ16.Wherein x ∈ χiRepresent the characteristic vector subclass for belonging to i-th phoneme.Each subclass is considered as solely
Vertical data flow carries out subsequent treatment.The voiceprint registration stage, the model of 16 phoneme correlations(i-th sound of speaker s
Sub-prime set) obtained by the unrelated Algorithm for Training of text.It should be noted that registration voice needs to cover ten numerals.This
In bright, registration phase registers voice using three numeric strings, it is ensured that in everyone registration voice, each numeral at least occurs one
Time.
During voiceprint, for " the whole numerical ciphers of fixation " condition, the scores vector ξ of ten 6 DOFs is obtained,Can be by averaging to scores vector ξ or the method such as logistic regression trains rear end grader to enter
Row judgement.But for the rand-n condition such as " dynamic 8 bit digital password " and " dynamic 6 bit digital password ", scores vector ξ may
There is disappearance, because tested speech only includes the partial content of set of phonemes.In order to solve the problem, using neural network algorithm
In dropout strategy, this be a kind of effective lifting generalization ability implementation method.
The dropout training algorithm of neutral net is stochastic gradient descent (the Standard Stochastic of standard
Gradient Descent), simply during forward calculation, some input blocks and hidden layer are ignored at random with certain probability γ
Unit.Only activation unit participates in back propagation (Back-propagation) and gradient calculation.Because dropout is not used to
Identification, in the training process, the output to per layer is readjusted：
Wherein δ (), WlAnd blIt is activation primitive, the weight of l layer and the biasing of l layer respectively.Bm is binary cover
(Binary mask) represents which dimension is disallowable, and * represents vector multiplication.
Said process can regard a kind of effective model averaging method as, i.e., by train the disappearance of substantial amounts of shared weight to
The average expression of the heterogeneous networks for measuring.As shown in figure 1, neural network classifier of the training package containing a hidden layer.Wherein defeated
It is scores vector to enter, and output includes two units, represents target authentication classification respectively and emits imitative certification classification.For " dynamic 8
Vector dimension disappearance problem under the conditions of the rand-n such as numerical ciphers " and " dynamic 6 bit digital password ", to input layer with probability γ
Application dropout strategy carries out network training.In Qualify Phase, the log-likelihood being calculated as follows is used for system output：
Wherein p (ξ | target verification class) and p (ξ | emit imitative checking class) is the likelihood score of scores vector ξ.By Bayes's public affairs
Formula, likelihood score can be exchanged into posteriority and represent,
P (ξ | target verification class)=p (target verification class | ξ) p (ξ)/p (target verification class)
P (ξ | emit imitative checking class)=p (emit imitative checking class | ξ) p (ξ)/p (emitting imitative checking class)
Wherein p (target verification class | ξ) and p (emit imitative checking class | ξ) it is that scores vector ξ is obtained by network forward calculation
Posteriority.P (target verification class) and p (emitting imitative checking class) is to estimate the priori of the target verification class for obtaining from training set and emit imitative
The priori of checking class.P (ξ) is unrelated with any model, can ignore during LLR is calculated.
Each phoneme model separating capacity to Application on Voiceprint Recognition is analyzed first.Training voice in view of each phoneme model
Feature is less, in order to avoid crossing training problem, reduces the training parameter amount of each phoneme correlation model.Fig. 2 gives each
Phoneme correlation model etc. error rate contrast.
From figure 2 it can be seen that first, in all phoneme correlation models, iVector is to be better than GMM- more by a small margin
NAP model.Secondly, the EER numerical value of the worst consonant of performance " w " is five times of the EER of the best vowel of performance " an " or so.This
Individual experimental result has directive function to practical application, and on-line system can be limited and push the bad numeral of performance, and such as " 5
[wu]”.
By dropout neutral net rear end grader is trained, fusion output is carried out to the scores vector of phoneme correlation.Table
4 give phoneme correlation model using different rear ends graders etc. error rate contrast.Compare for convenience, give also here
The authentication performance averaged by the phoneme associated score of GMM-NAP and iVector system.Fraction average formula is as follows：
Table 4：Phoneme correlation model using different rear ends graders etc. error rate contrast
From table 4, it can be seen that of the present invention based on phoneme information explicitly using the calculation with the fusion of neutral net rear end
Method can effectively lift the systematic function of numeric string voiceprint.The result average compared to fraction, neutral net rear end is divided
Class device etc. error rate lower, performance is more excellent.With GMM-NAP the and iVector Comparative result of table 2, register/recognize in three kinds of differences
Under the conditions of card, the algorithm of phoneme correlation model and neutral net rear end grader is all achieved under about 20% or so relative EER
Drop.
Presently preferred embodiments of the present invention is the foregoing is only, not in order to limit the present invention, all essences in the present invention
Within god and principle, any modification, equivalent substitution and improvement that is made etc., should be included within the scope of the present invention.
Claims (3)
1. a kind of voiceprint authentication system based on phoneme information, it is characterised in that include based on Chinese putonghua speech evaluator
Phoneme force alignment module, phoneme correlation model creation module and based on dropout strategy neural network classifier mould
Block；
The phoneme based on Chinese putonghua speech evaluator forces alignment module to be used for 16 phoneme classes to numeric string
Cutting is not carried out；
The model creation module of the phoneme correlation is used for setting up phoneme correlation model, and analyzes each phoneme correlation model to sound
The separating capacity of stricture of vagina certification；
The neural network classifier module based on dropout strategy is used for merging the complementary information of phoneme correlation model.
2. a kind of voiceprint authentication method based on phoneme information, it is characterised in that comprise the steps：
S01：Define 16 phoneme class of standard Chinese numeric string vocal print, explicit each pronunciation classification using numeric string
Information；
S02：Based on Chinese putonghua speech evaluator, force alignment algorithm to obtain each using Viterbi and correspond to numeric string text
The phoneme boundary of content, completes the phone segmentation to voice content, obtains belonging to the characteristic vector subclass of phoneme；
S03：Phoneme correlation model is set up using the unrelated algorithm of text；
S04：Phoneme correlation model is calculated, obtains scores vector.
3. the voiceprint authentication method based on phoneme information according to claim 2, it is characterised in that adopt in step S04
Dropout Strategies Training rear end integrated classification device in neural network algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610880776.6A CN106448685B (en) | 2016-10-09 | 2016-10-09 | A kind of voiceprint authentication system and method based on phoneme information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610880776.6A CN106448685B (en) | 2016-10-09 | 2016-10-09 | A kind of voiceprint authentication system and method based on phoneme information |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106448685A true CN106448685A (en) | 2017-02-22 |
| CN106448685B CN106448685B (en) | 2019-11-22 |
Family
ID=58172115
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610880776.6A Active CN106448685B (en) | 2016-10-09 | 2016-10-09 | A kind of voiceprint authentication system and method based on phoneme information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106448685B (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108198574A (en) * | 2017-12-29 | 2018-06-22 | 科大讯飞股份有限公司 | Change of voice detection method and device |
| CN108648760A (en) * | 2018-04-17 | 2018-10-12 | 四川长虹电器股份有限公司 | Real-time sound-groove identification System and method for |
| CN109065023A (en) * | 2018-08-23 | 2018-12-21 | 广州势必可赢网络科技有限公司 | A kind of voice identification method, device, equipment and computer readable storage medium |
| CN110111798A (en) * | 2019-04-29 | 2019-08-09 | 平安科技（深圳）有限公司 | A kind of method and terminal identifying speaker |
| CN110689895A (en) * | 2019-09-06 | 2020-01-14 | 北京捷通华声科技股份有限公司 | Voice verification method and device, electronic equipment and readable storage medium |
| CN110875044A (en) * | 2018-08-30 | 2020-03-10 | 中国科学院声学研究所 | Speaker identification method based on word correlation score calculation |
| CN111243603A (en) * | 2020-01-09 | 2020-06-05 | 厦门快商通科技股份有限公司 | Voiceprint recognition method, system, mobile terminal and storage medium |
| CN111341320A (en) * | 2020-02-28 | 2020-06-26 | 中国工商银行股份有限公司 | Phrase voice voiceprint recognition method and device |
| CN111785284A (en) * | 2020-08-19 | 2020-10-16 | 科大讯飞股份有限公司 | Method, device and equipment for recognizing text-independent voiceprint based on phoneme assistance |
| CN114299921A (en) * | 2021-12-07 | 2022-04-08 | 浙江大学 | Voiceprint security scoring method and system for voice command |
| CN115831120A (en) * | 2023-02-03 | 2023-03-21 | 北京探境科技有限公司 | Corpus data acquisition method and device, electronic equipment and readable storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070033041A1 (en) * | 2004-07-12 | 2007-02-08 | Norton Jeffrey W | Method of identifying a person based upon voice analysis |
| CN101467204A (en) * | 2005-05-27 | 2009-06-24 | 普提克斯科技股份有限公司 | Method and system for bio-metric voice print authentication |
| CN204465555U (en) * | 2015-04-14 | 2015-07-08 | 时代亿宝(北京)科技有限公司 | Based on the voiceprint authentication apparatus of time type dynamic password |
| CN104834849A (en) * | 2015-04-14 | 2015-08-12 | 时代亿宝(北京)科技有限公司 | Dual-factor identity authentication method and system based on voiceprint recognition and face recognition |
-
2016
- 2016-10-09 CN CN201610880776.6A patent/CN106448685B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070033041A1 (en) * | 2004-07-12 | 2007-02-08 | Norton Jeffrey W | Method of identifying a person based upon voice analysis |
| CN101467204A (en) * | 2005-05-27 | 2009-06-24 | 普提克斯科技股份有限公司 | Method and system for bio-metric voice print authentication |
| CN204465555U (en) * | 2015-04-14 | 2015-07-08 | 时代亿宝(北京)科技有限公司 | Based on the voiceprint authentication apparatus of time type dynamic password |
| CN104834849A (en) * | 2015-04-14 | 2015-08-12 | 时代亿宝(北京)科技有限公司 | Dual-factor identity authentication method and system based on voiceprint recognition and face recognition |
Non-Patent Citations (1)
| Title |
|---|
| 张涛涛: "语音声纹密码验证技术研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108198574A (en) * | 2017-12-29 | 2018-06-22 | 科大讯飞股份有限公司 | Change of voice detection method and device |
| CN108198574B (en) * | 2017-12-29 | 2020-12-08 | 科大讯飞股份有限公司 | Sound change detection method and device |
| CN108648760A (en) * | 2018-04-17 | 2018-10-12 | 四川长虹电器股份有限公司 | Real-time sound-groove identification System and method for |
| CN109065023A (en) * | 2018-08-23 | 2018-12-21 | 广州势必可赢网络科技有限公司 | A kind of voice identification method, device, equipment and computer readable storage medium |
| CN110875044A (en) * | 2018-08-30 | 2020-03-10 | 中国科学院声学研究所 | Speaker identification method based on word correlation score calculation |
| CN110875044B (en) * | 2018-08-30 | 2022-05-03 | 中国科学院声学研究所 | Speaker identification method based on word correlation score calculation |
| CN110111798A (en) * | 2019-04-29 | 2019-08-09 | 平安科技（深圳）有限公司 | A kind of method and terminal identifying speaker |
| CN110111798B (en) * | 2019-04-29 | 2023-05-05 | 平安科技（深圳）有限公司 | Method, terminal and computer readable storage medium for identifying speaker |
| CN110689895A (en) * | 2019-09-06 | 2020-01-14 | 北京捷通华声科技股份有限公司 | Voice verification method and device, electronic equipment and readable storage medium |
| CN111243603A (en) * | 2020-01-09 | 2020-06-05 | 厦门快商通科技股份有限公司 | Voiceprint recognition method, system, mobile terminal and storage medium |
| CN111341320B (en) * | 2020-02-28 | 2023-04-14 | 中国工商银行股份有限公司 | Phrase voice voiceprint recognition method and device |
| CN111341320A (en) * | 2020-02-28 | 2020-06-26 | 中国工商银行股份有限公司 | Phrase voice voiceprint recognition method and device |
| CN111785284A (en) * | 2020-08-19 | 2020-10-16 | 科大讯飞股份有限公司 | Method, device and equipment for recognizing text-independent voiceprint based on phoneme assistance |
| CN111785284B (en) * | 2020-08-19 | 2024-04-30 | 科大讯飞股份有限公司 | Text-independent voiceprint recognition method, device and equipment based on phoneme assistance |
| CN114299921B (en) * | 2021-12-07 | 2022-11-18 | 浙江大学 | Voiceprint security scoring method and system for voice command |
| CN114299921A (en) * | 2021-12-07 | 2022-04-08 | 浙江大学 | Voiceprint security scoring method and system for voice command |
| CN115831120A (en) * | 2023-02-03 | 2023-03-21 | 北京探境科技有限公司 | Corpus data acquisition method and device, electronic equipment and readable storage medium |
| CN115831120B (en) * | 2023-02-03 | 2023-06-16 | 北京探境科技有限公司 | Corpus data acquisition method and device, electronic equipment and readable storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106448685B (en) | 2019-11-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106448685A (en) | System and method for identifying voice prints based on phoneme information | |
| KR101995547B1 (en) | Neural Networks for Speaker Verification | |
| CN104575490B (en) | Spoken language pronunciation evaluating method based on deep neural network posterior probability algorithm | |
| US10013972B2 (en) | System and method for identifying speakers | |
| ES2605779T3 (en) | Speaker Recognition | |
| Gomez-Alanis et al. | On joint optimization of automatic speaker verification and anti-spoofing in the embedding space | |
| TWI527023B (en) | A voiceprint recognition method and apparatus | |
| Dileep et al. | GMM-based intermediate matching kernel for classification of varying length patterns of long duration speech using support vector machines | |
| CN107492382A (en) | Voiceprint extracting method and device based on neutral net | |
| CN107731233A (en) | A kind of method for recognizing sound-groove based on RNN | |
| CN104765996B (en) | Voiceprint password authentication method and system | |
| KR20060070603A (en) | Two stage utterance verification method and device of speech recognition system | |
| CN104240706B (en) | It is a kind of that the method for distinguishing speek person that similarity corrects score is matched based on GMM Token | |
| CN104462912B (en) | Improved biometric password security | |
| Lopez-Otero et al. | Analysis of gender and identity issues in depression detection on de-identified speech | |
| Wang et al. | A network model of speaker identification with new feature extraction methods and asymmetric BLSTM | |
| Wang et al. | Automatic detection of speaker state: Lexical, prosodic, and phonetic approaches to level-of-interest and intoxication classification | |
| CN105280181A (en) | Training method for language recognition model and language recognition method | |
| Safavi et al. | Fraud detection in voice-based identity authentication applications and services | |
| CN104464738B (en) | A kind of method for recognizing sound-groove towards Intelligent mobile equipment | |
| Folorunso et al. | A review of voice-base person identification: state-of-the-art | |
| Li et al. | Cost-sensitive learning for emotion robust speaker recognition | |
| Chen et al. | Speech emotion classification using acoustic features | |
| US6499012B1 (en) | Method and apparatus for hierarchical training of speech models for use in speaker verification | |
| Wang et al. | Capture interspeaker information with a neural network for speaker identification |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP01 | Change in the name or title of a patent holder | ||
| CP01 | Change in the name or title of a patent holder |
Address after: East Zone 9A, 9th Floor, Building 1, No. 158 West Fourth Ring North Road, Haidian District, Beijing, 100142 Patentee after: Beijing Yuan Jian Polytron Technologies Inc. Address before: East Zone 9A, 9th Floor, Building 1, No. 158 West Fourth Ring North Road, Haidian District, Beijing, 100142 Patentee before: Beijing Yuanjian Technologies Co.,Ltd. |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20231010 Address after: No. 016, Xiaocuigezhuang Village, Gaolou Town, Sanhe City, Langfang City, Hebei Province, 065200 Patentee after: Liu Xuefeng Address before: East Zone 9A, 9th Floor, Building 1, No. 158 West Fourth Ring North Road, Haidian District, Beijing, 100142 Patentee before: Beijing Yuan Jian Polytron Technologies Inc. |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20240218 Address after: Room 320, 3rd Floor, Building A, No. 119 West Fourth Ring North Road, Haidian District, Beijing, 100000 Patentee after: Beijing Yuanjian Information Technology Co.,Ltd. Country or region after: China Address before: No. 016, Xiaocuigezhuang Village, Gaolou Town, Sanhe City, Langfang City, Hebei Province, 065200 Patentee before: Liu Xuefeng Country or region before: China |