CN109273002A - Vehicle configuration method, system, vehicle device and vehicle - Google Patents
Vehicle configuration method, system, vehicle device and vehicle Download PDFInfo
- Publication number
- CN109273002A CN109273002A CN201811260569.6A CN201811260569A CN109273002A CN 109273002 A CN109273002 A CN 109273002A CN 201811260569 A CN201811260569 A CN 201811260569A CN 109273002 A CN109273002 A CN 109273002A
- Authority
- CN
- China
- Prior art keywords
- user
- voiceprint
- vehicle
- voice messaging
- feature
- 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 68
- 238000012549 training Methods 0.000 claims description 85
- 238000005315 distribution function Methods 0.000 claims description 39
- 239000011159 matrix material Substances 0.000 claims description 27
- 238000001514 detection method Methods 0.000 claims description 13
- 239000012141 concentrate Substances 0.000 claims description 8
- 230000006870 function Effects 0.000 description 8
- 230000002452 interceptive effect Effects 0.000 description 6
- 230000000694 effects Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000007476 Maximum Likelihood Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000001149 cognitive effect Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 230000001755 vocal effect Effects 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 241001269238 Data Species 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000008447 perception Effects 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
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
-
- 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
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)
- Fittings On The Vehicle Exterior For Carrying Loads, And Devices For Holding Or Mounting Articles (AREA)
Abstract
The invention belongs to technical field of voice recognition, a kind of vehicle configuration method, system, vehicle device and vehicle are specifically provided, it is intended to solve the problems, such as to need manual setting individual cultivation again when the prior art replaces driver and passenger.For this purpose, the present invention provides a kind of vehicle configuration method, including obtaining the voice messaging of user and identifying the corresponding user's voiceprint of voice messaging；According to preset multiple voiceprints User ID corresponding with the one-to-one relationship of multiple User ID acquisition user's voiceprint；Execute the associated vehicle configuration operation of User ID corresponding with user's voiceprint in advance.Based on above-mentioned steps, the method provided according to the present invention, vehicle can execute automatically according to the individual cultivation before user to be operated with the associated vehicle configuration of the User ID, improve the car experience of user, corresponding vehicle configuration operation can also be executed according to the ID of different user, improve the intelligence of vehicle.
Description
Technical field
The invention belongs to technical field of voice recognition, and in particular to a kind of vehicle configuration method, system, vehicle device and vehicle
?.
Background technique
With the development of vehicle technology, vehicle has become essential a part in people's life, interior scene
At an important place in for people's lives.Requirement with people to vehicle is higher and higher, and people are more than and want to
Safe driving, under the premise of guaranteeing safe driving, it is desirable also to possess good driving experience by people.
When driving vehicle, different driver and passenger can carry out vehicle according to personalized preference and level of comfort personalized
Configuration, such as adjustment height of seat, door mirror angle and seat back angle, but when replacement driver and passenger, new driving
Personnel need manual setting individual cultivation again, cumbersome reduction user experience.
With the development of voice technology, user drive when can to vehicle carry out interactive voice to realize relevant operation,
Therefore, how vehicle to be carried out individual cultivation to improve user's car experience being that those skilled in the art need at present by voice
It solves the problems, such as.
Summary of the invention
In order to solve the above problem in the prior art, in order to be needed again when solving prior art replacement driver and passenger
The problem of manual setting vehicle personalization configures, the first aspect of the present invention provides a kind of vehicle configuration method, comprising:
It obtains the voice messaging of user and identifies the corresponding user's voiceprint of the voice messaging；
User's voiceprint is obtained according to the one-to-one relationship of preset multiple voiceprints and multiple User ID
Corresponding User ID；
Execute the associated vehicle configuration operation of User ID corresponding with user's voiceprint in advance.
Optionally, in provided vehicle configuration method, " identifying the corresponding user's voiceprint of the voice messaging "
The step of include:
The mean value of each gauss of distribution function in the universal background model constructed in advance is obtained, and according to the mean value meter
Calculate fixed reference feature；
The phonetic feature of the voice messaging is extracted, and the gauss of distribution function is carried out according to the phonetic feature
Interpolation obtains the corresponding gauss of distribution function of the phonetic feature；
According to the mean value computation target signature of the corresponding gauss of distribution function of the phonetic feature；
Global variable space matrix is obtained according to expectation-maximization algorithm and the voice messaging；
Method shown according to the fixed reference feature, target signature and global variable space matrix and according to the following formula obtains
The acoustic feature of the voice messaging:
W=(s-m)/T
Wherein, w indicates the acoustic feature, and s indicates the target signature, and m indicates the fixed reference feature, described in T expression
Global variable space matrix；
Channel compensation is carried out to the acoustic feature, obtains the corresponding voiceprint of the voice messaging；
Wherein, the universal background model is constructed based on preset gauss hybrid models and preset voice training collection
Model.
Optionally, it in provided vehicle configuration method, " is obtaining each high in the universal background model constructed in advance
Before the step of mean value of this distribution function, and according to the mean value computation fixed reference feature ", the method also includes:
The phonetic feature that the voice training concentrates voice messaging is extracted, and utilizes speech detection algorithms by the voice
The quiet data of feature is rejected, and universal background model training characteristics are obtained；
According to the universal background model training characteristics and utilize the expectation-maximization algorithm training general back
Scape model.
Optionally, in provided vehicle configuration method, the voice training collection includes multiple preset sub- training sets；
The quantity of the universal background model is multiple and each universal background model and each son training
Collection corresponds.
Optionally, in provided vehicle configuration method, " according to preset multiple voiceprints and multiple User ID
One-to-one relationship obtains the corresponding User ID of user's voiceprint " the step of include:
Calculate separately the similarity between the multiple voiceprint and user's voiceprint；
Judge whether the multiple similarities being calculated are more than or equal to preset similarity threshold respectively:
A similarity more than or equal to the similarity threshold if it exists, the then voiceprint for obtaining the similarity are corresponding
User ID；
Multiple similarities more than or equal to the similarity threshold if it exists then obtain maximum phase in the multiple similarity
Like the corresponding User ID of voiceprint of degree.
Optionally, in provided vehicle configuration method, the vehicle configuration operation includes body configurations operation, amusement
Configuration operation and driving configuration operation.
Optionally, in provided vehicle configuration method, the body configurations operation includes adjusting vehicle rearview pitch-angle
Degree, seat back angle and height of seat, the entertainment configuration operation includes logging in multimedia corresponding with the User ID
Account, the driving configuration operation include obtaining the User ID often to go to drive place.
The second aspect of the present invention provides a kind of vehicle configuration system, comprising:
User's voiceprint acquiring unit is configured to obtain the voice messaging of user and identifies that the voice messaging is corresponding
User's voiceprint；
User ID acquiring unit is configured to the one-to-one relationship according to preset multiple voiceprints and multiple User ID
Obtain the corresponding User ID of user's voiceprint；
Vehicle configuration operation execution unit is configured to execute User ID association corresponding with user's voiceprint in advance
Vehicle configuration operation.
Optionally, in provided vehicle configuration system, user's voiceprint acquiring unit is further configured to
Execute following operation:
The mean value of each gauss of distribution function in the universal background model constructed in advance is obtained, and according to the mean value meter
Calculate fixed reference feature；
The phonetic feature of the voice messaging is extracted, and the gauss of distribution function is carried out according to the phonetic feature
Interpolation obtains the corresponding gauss of distribution function of the phonetic feature；
According to the mean value computation target signature of the corresponding gauss of distribution function of the phonetic feature；
Global variable space matrix is obtained according to expectation-maximization algorithm and the voice messaging；
Method shown according to the fixed reference feature, target signature and global variable space matrix and according to the following formula obtains
The acoustic feature of the voice messaging:
W=(s-m)/T
Wherein, w indicates the acoustic feature, and s indicates the target signature, and m indicates the fixed reference feature, described in T expression
Global variable space matrix；
Channel compensation is carried out to the acoustic feature, obtains the corresponding voiceprint of the voice messaging；
Wherein, the universal background model is constructed based on preset gauss hybrid models and preset voice training collection
Model.
Optionally, in provided vehicle configuration system, the vehicle configuration system further includes model training unit, institute
Model training unit is stated to be configured to execute following operation:
The phonetic feature that the voice training concentrates voice messaging is extracted, by speech detection algorithms by the phonetic feature
Quiet data reject, obtain universal background model training characteristics；
According to the universal background model training characteristics and utilize the expectation-maximization algorithm training common background mould
Type.
Optionally, in provided vehicle configuration system, the voice training collection includes multiple preset sub- training sets；
The quantity of the universal background model is multiple and each universal background model and each son training
Collection corresponds.
Optionally, in provided vehicle configuration system, the User ID acquiring unit be further configured to execute with
Lower operation:
Calculate separately the similarity between the multiple voiceprint and user's voiceprint；
Judge whether the multiple similarities being calculated are more than or equal to preset similarity threshold respectively:
A similarity more than or equal to the similarity threshold if it exists, the then voiceprint for obtaining the similarity are corresponding
User ID；
Multiple similarities more than or equal to the similarity threshold if it exists then obtain maximum phase in the multiple similarity
Like the corresponding User ID of voiceprint of degree.
Optionally, in provided vehicle configuration system, the vehicle configuration operation includes body configurations operation, amusement
Configuration operation and driving configuration operation.
Optionally, in provided vehicle configuration system, the body configurations operation includes adjusting vehicle rearview pitch-angle
Degree, seat back angle and height of seat, the entertainment configuration operation includes logging in and the associated multimedia of the User ID
Account, the driving configuration operation, which includes that the acquisition User ID is corresponding, often goes to drive place.
Third aspect present invention provides a kind of vehicle device, including vehicle configuration system described in any of the above embodiments.
The fourth aspect of the present invention provides a kind of vehicle, including vehicle device described above.
The fifth aspect of the present invention provides a kind of storage device, wherein be stored with a plurality of program, described program be suitable for by
Processor is loaded to execute vehicle configuration method described in any of the above embodiments.
The sixth aspect of the present invention provides a kind of control device, including processor and storage equipment；The storage equipment,
Suitable for storing a plurality of program；Described program is suitable for being loaded by the processor to execute vehicle configuration side described in any of the above embodiments
Method.
Compared with the immediate prior art, above-mentioned technical proposal is at least had the following beneficial effects:
1, vehicle configuration method provided by the invention, can be by obtaining the voice messaging of user and identifying voice messaging pair
The user's voiceprint answered obtains user's vocal print according to the one-to-one relationship of preset multiple voiceprints and multiple User ID
The corresponding User ID of information, executes User ID corresponding with voiceprint associated vehicle configuration operation in advance, user with vehicle
Carry out interactive voice during, vehicle can execute automatically associated with the User ID according to the individual cultivation before user
Vehicle configuration operation, improves the car experience of user.
2, vehicle configuration method provided by the invention can identify the corresponding voiceprint of voice messaging, according to multiple sound
The one-to-one relationship of line information and multiple User ID obtains the corresponding User ID of voiceprint, uses when there is multiple vehicles to use
When family, corresponding vehicle configuration operation can be executed according to the ID of different user, improves the intelligence of vehicle.
3, vehicle configuration method provided by the invention, can be according to the multiple common background moulds of the classification of voice training collection training
Type obtains the voiceprint of voice messaging based on trained universal background model, and the frame alignment of voice messaging can be improved
Accuracy, and then available and user's strong correlation voiceprint, improve the accuracy rate of identification.
Scheme 1, a kind of vehicle configuration method, characterized by comprising:
It obtains the voice messaging of user and identifies the corresponding user's voiceprint of the voice messaging；
It is corresponding with the one-to-one relationship of multiple User ID acquisition user's voiceprint according to preset multiple voiceprints
User ID；
Execute the associated vehicle configuration operation of User ID corresponding with user's voiceprint in advance.
Scheme 2, vehicle configuration method according to scheme 1, which is characterized in that " identify the corresponding user of the voice messaging
The step of voiceprint " includes:
The mean value of each gauss of distribution function in the universal background model constructed in advance is obtained, and is joined according to the mean value computation
Examine feature；
The phonetic feature of the voice messaging is extracted, and according to the phonetic feature gauss of distribution function is carried out slotting
Value, obtains the corresponding gauss of distribution function of the phonetic feature；
According to the mean value computation target signature of the corresponding gauss of distribution function of the phonetic feature；
Global variable space matrix is obtained according to expectation-maximization algorithm and the voice messaging；
Described in method shown according to the fixed reference feature, target signature and global variable space matrix and according to the following formula obtains
The acoustic feature of voice messaging:
W=(s-m)/T
Wherein, w indicates that the acoustic feature, s indicate the target signature, and m indicates that the fixed reference feature, T indicate the overall situation
Variable space matrix；
Channel compensation is carried out to the acoustic feature, obtains the corresponding voiceprint of the voice messaging；
Wherein, the universal background model is the mould constructed based on preset gauss hybrid models and preset voice training collection
Type.
Scheme 3, the vehicle configuration method according to scheme 2, which is characterized in that " obtaining the common background mould constructed in advance
Before the step of mean value of each gauss of distribution function in type, and according to the mean value computation fixed reference feature ", the method is also
Include:
The phonetic feature that the voice training concentrates voice messaging is extracted, and utilizes speech detection algorithms by the phonetic feature
Quiet data reject, obtain universal background model training characteristics；
According to the universal background model training characteristics and utilize the expectation-maximization algorithm training common background mould
Type.
Scheme 4, vehicle configuration method according to scheme 3, which is characterized in that the voice training collection includes multiple preset
Sub- training set；
The quantity of the universal background model is multiple and each universal background model and each sub- training set one
One is corresponding.
Scheme 5, vehicle configuration method according to scheme 1, which is characterized in that " according to preset multiple voiceprints and more
The one-to-one relationship of a User ID obtains the corresponding User ID of user's voiceprint " the step of include:
Calculate separately the similarity between the multiple voiceprint and user's voiceprint；
Judge whether the multiple similarities being calculated are more than or equal to preset similarity threshold respectively:
A similarity more than or equal to the similarity threshold if it exists, then obtain the corresponding use of voiceprint of the similarity
Family ID；
Multiple similarities more than or equal to the similarity threshold if it exists, then obtain maximum similarity in the multiple similarity
The corresponding User ID of voiceprint.
Scheme 6, vehicle configuration method according to scheme 1, which is characterized in that the vehicle configuration operation includes body configurations
Operation, entertainment configuration operation and driving configuration operation.
Scheme 7, vehicle configuration method according to scheme 6, which is characterized in that the body configurations operation includes adjusting vehicle
Door mirror angle, seat back angle and height of seat, the entertainment configuration operation include logging in be associated with the User ID
Multimedia account, driving configuration operation includes obtaining that the User ID is corresponding often to go to drive place.
Scheme 8, a kind of vehicle configuration system, characterized by comprising:
User's voiceprint acquiring unit is configured to obtain the voice messaging of user and identifies the corresponding user of the voice messaging
Voiceprint；
User ID acquiring unit is configured to be obtained according to the one-to-one relationship of preset multiple voiceprints and multiple User ID
The corresponding User ID of user's voiceprint；
Vehicle configuration operation execution unit is configured to execute the associated vehicle of User ID corresponding with user's voiceprint in advance
Configuration operation.
Scheme 9, the vehicle configuration system according to scheme 8, which is characterized in that user's voiceprint acquiring unit is into one
Step is configured to execute following operation:
The mean value of each gauss of distribution function in the universal background model constructed in advance is obtained, and is joined according to the mean value computation
Examine feature；
The phonetic feature of the voice messaging is extracted, and according to the phonetic feature gauss of distribution function is carried out slotting
Value, obtains the corresponding gauss of distribution function of the phonetic feature；
According to the mean value computation target signature of the corresponding gauss of distribution function of the phonetic feature；
Global variable space matrix is obtained according to expectation-maximization algorithm and the voice messaging；
Described in method shown according to the fixed reference feature, target signature and global variable space matrix and according to the following formula obtains
The acoustic feature of voice messaging:
W=(s-m)/T
Wherein, w indicates that the acoustic feature, s indicate the target signature, and m indicates that the fixed reference feature, T indicate the overall situation
Variable space matrix；
Channel compensation is carried out to the acoustic feature, obtains the corresponding voiceprint of the voice messaging；
Wherein, the universal background model is the mould constructed based on preset gauss hybrid models and preset voice training collection
Type.
Scheme 10, vehicle configuration system according to scheme 9, which is characterized in that the vehicle configuration system further includes model
Training unit, the model training unit are configured to execute following operation:
The phonetic feature that the voice training concentrates voice messaging is extracted, by speech detection algorithms by the quiet of the phonetic feature
Sound data are rejected, and universal background model training characteristics are obtained；
According to the universal background model training characteristics and utilize the expectation-maximization algorithm training universal background model.
Scheme 11, vehicle configuration system according to scheme 10, which is characterized in that the voice training collection includes multiple default
Sub- training set；
The quantity of the universal background model is multiple and each universal background model and each sub- training set one
One is corresponding.
Scheme 12, the vehicle configuration system according to scheme 8, which is characterized in that the User ID acquiring unit is further matched
It is set to the following operation of execution:
Calculate separately the similarity between the multiple voiceprint and user's voiceprint；
Judge whether the multiple similarities being calculated are more than or equal to preset similarity threshold respectively:
A similarity more than or equal to the similarity threshold if it exists, then obtain the corresponding use of voiceprint of the similarity
Family ID；
Multiple similarities more than or equal to the similarity threshold if it exists, then obtain maximum similarity in the multiple similarity
The corresponding User ID of voiceprint.
Scheme 13, the vehicle configuration system according to scheme 8, which is characterized in that the vehicle configuration operation is matched including car body
It sets operation, entertainment configuration operation and drives configuration operation.
Scheme 14, vehicle configuration system according to scheme 13, which is characterized in that the body configurations operation includes adjusting vehicle
Door mirror angle, seat back angle and height of seat, the entertainment configuration operation include logging in close with the User ID
The multimedia account of connection, the driving configuration operation, which includes that the acquisition User ID is corresponding, often goes to drive place.
Scheme 15, a kind of vehicle device, which is characterized in that including vehicle configuration system described in any one of scheme 8-14.
Scheme 16, a kind of vehicle, which is characterized in that including vehicle device described in scheme 15.
Scheme 17, a kind of storage device, wherein being stored with a plurality of program, which is characterized in that described program is suitable for being added by processor
It carries with vehicle configuration method described in any one of the 1-7 that carries into execution a plan.
Scheme 18, a kind of control device, including processor and storage equipment；The storage equipment is suitable for storing a plurality of program；Its
It is characterized in that, described program is suitable for being loaded as the processor with vehicle configuration method described in any one of the 1-7 that carries into execution a plan.
Detailed description of the invention
Fig. 1 is the key step schematic diagram of the vehicle configuration method of an embodiment of the present invention；
Fig. 2 is the key step schematic diagram of the corresponding voiceprint of identification voice messaging of an embodiment of the present invention；
Fig. 3 is the primary structure schematic diagram of the vehicle configuration system of an embodiment of the present invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art
Every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
The preferred embodiment of the present invention described with reference to the accompanying drawings.It will be apparent to a skilled person that this
A little embodiments are used only for explaining technical principle of the invention, it is not intended that limit the scope of the invention.
Refering to attached drawing 1, Fig. 1 illustratively gives the key step of vehicle configuration method in the present embodiment.Such as Fig. 1 institute
Show, vehicle configuration method includes the following steps: in the present embodiment
Step S101: the voice messaging of user is obtained.
User is during carrying out interactive voice with vehicle, the available user of vehicle voice messaging currently entered,
Wherein it is possible to obtain the voice messaging of user by the voice acquisition device of vehicle, the voice messaging of user may include user
With the interactive voice of vehicle and the registration voice of user.What the interactive voice of user and vehicle can send for user to vehicle
Voice, such as " playing music " are controlled, the registration voice of user can be user and carry out specified number according to the prompt of onboard system
The voice of specified duration typing is measured, onboard system can store the registration voice of multiple common users.
Step S102: the corresponding user's voiceprint of identification voice messaging.
The voice messaging of user may include the content and user's voiceprint of voice, can be with by user's voiceprint
Differentiate the identity of user.
The identity that user is differentiated by user's voiceprint may include registration phase and cognitive phase.In registration phase,
Several voice messagings can be collected or recorded for each user, and vehicle device is extracted from voice messaging embodies user personality spy
The parameters,acoustic of sign, the parameters,acoustic for embodying the user personality feature can be user's voiceprint, further according to certain model
Method for building up forms the template for being directed to the user, and is stored in user template library.In cognitive phase, system can with registration rank
The identical method of section extracts the voiceprint of user speech information to be identified, and the body of user is differentiated according to certain method of discrimination
Part.
Refering to attached drawing 2, Fig. 2, which illustratively gives, identifies the corresponding user's voiceprint of voice messaging in the present embodiment
Key step.As shown in Fig. 2, identifying that the corresponding user's voiceprint of voice messaging may include following step in the present embodiment:
Step S1021: the mean value of each gauss of distribution function in the universal background model that acquisition constructs in advance, and according to
Mean value computation fixed reference feature.
Universal background model (UBM, Universal Background Model) can pass through the language of a large amount of speaker
Sound training forms, and has very high degree of mixing and reference background meaning, and careful statistical nature point can be carried out to voice messaging
Cloth description.Wherein, universal background model can be based on preset gauss hybrid models (GMM, Gaussian Mixed
Model) and preset voice training collection building model.
Gauss hybrid models are the feature distribution situations that speaker is described with the linear combination of multiple Gaussian Profiles, when identification
The corresponding speaker of speaker model of tested speech feature can will most be generated as recognition result, same speaker is when different
Even the phase says that also it occur frequently that biggish variation, but the personal characteristics of the speaker has from the angle of statistics in short
There is consistency, the gauss hybrid models Speaker Identification effect unrelated for text is preferable.Gauss hybrid models can be high by M
The summation of this probability density function linear weighted function is constituted, and when the M of gauss hybrid models is sufficiently high, can subtly approach speaker
The spatial distribution of characteristic vector.Preset voice training collection can be the voice collection being made of the voice messaging of a large amount of speaker
It closes.
Specifically, in available universal background model each gauss of distribution function mean value, according to mean value computation voice
The fixed reference feature of information, fixed reference feature can be the Gaussian mean super vector of universal background model, the super vector with specifically
The channel for talking about people and speaker's voice is unrelated, has good reference background meaning.
Before the step S1021 the step of, the method for the present embodiment further includes trained universal background model, specifically, instruction
The method for practicing universal background model includes the following steps:
Step S10211: the phonetic feature that voice training concentrates voice messaging is extracted, and will using speech detection algorithms
The quiet data of phonetic feature is rejected, and universal background model training characteristics are obtained.
The training data of universal background model, therefore can not in training data from a large amount of different backgrounds and noise circumstance
Avoid there are quiet datas, and it is the characteristic of voice that Application on Voiceprint Recognition, which utilizes, unrelated with quiet data, subsequent in order to reduce
Calculation amount can use the quiet data that speech detection algorithms detect voice messaging, and it rejected from voice messaging, obtain
To universal background model training characteristics.Wherein, speech detection algorithms can be VAD (Voice Activity Detection, language
Sound activation detection) algorithm etc., universal background model training characteristics can be MFCC (Mel Frequency Cepstral
Coefficent, mel-frequency cepstrum coefficient) feature etc., MFCC feature is the perception in order to simulate human ear to different frequency voice
Ability and the nonlinear characteristic parameters similar with human hearing characteristic proposed.It should be noted that protection scope of the present invention
It is not limited to above-mentioned specific embodiment, other than vad algorithm and MFCC feature, in the premise without departing from the principle of the present invention
Under, speech detection algorithms can also can reject the algorithm of quiet data for other, and the training characteristics of universal background model can also
Think that other can train the feature of universal background model.
Step S10212: according to universal background model training characteristics and expectation-maximization algorithm training common background is utilized
Model.
After obtaining universal background model training characteristics, expectation-maximization algorithm training universal background model can use.Tool
Body, training universal background model can use unsupervised e-learning, and the acoustic feature by counting each speaker is had
Some probability density functions construct the model of speaker, and voice training collection may include multiple preset sub- training sets, training
When, a universal background model can be trained with all data sets, but be the need to ensure that the equilibrium of training data, can also train
Multiple universal background models, then integrate them, for example, voice training collection can be divided into two son instructions according to gender
Practice collection, recycles the two sub- training sets that two universal background models are respectively trained, lack of balance data control can be effectively utilized
Make the universal background model generated.The quantity of universal background model can be multiple and each universal background model and every height
Training set can correspond.
Step S1022: extracting the phonetic feature of voice messaging, and is carried out to gauss of distribution function according to phonetic feature slotting
Value, obtains the corresponding gauss of distribution function of phonetic feature.
The phonetic feature for extracting voice messaging, according to phonetic feature and using maximum a posteriori probability adaptive algorithm to general
The Gaussian function of background model carries out linear interpolation, obtains the gauss hybrid models of user corresponding with the voice messaging, in turn
Obtain the corresponding gauss of distribution function of phonetic feature.On the basis of obtaining universal background model, in conjunction with height related to user
This mixed model can remove the influence in phonetic feature with background data similar portion, retain larger with background data difference
Part, increase to the robustness of noise, improve recognition effect.It is each general in the case where there is multiple universal background models
The Gaussian Mixture degree of background model can correspond, and facilitate the complete modification of model.
Step S1023: according to the mean value computation target signature of the corresponding gauss of distribution function of phonetic feature.
After obtaining the corresponding gauss of distribution function of phonetic feature, the mean value of gauss of distribution function is calculated as target spy
Sign.In practical applications, if universal background model has C Gaussian component, the dimension of feature is F, then the dimension of target signature is
C*F.Target signature contains the most individual information of user, can as the user characteristics with strong distinction, but
Be target signature dimension it is higher, a large amount of redundancy and scramble data related with channel are contained, in order to reduce the later period
Target signature can be decomposed using global space and carry out dimensionality reduction, obtain more compact feature by calculation amount.
Step S1024: global variable space matrix is obtained according to expectation-maximization algorithm and voice messaging.
In practical applications, it can use the Baum-Welch statistic that universal background model extracts every section of voice of user,
Pass through expectation-maximization algorithm (EM, Expectation Maximization Algorithm) iteration based on maximum-likelihood criterion
Solve global variable space matrix.Maximum likelihood requires the distribution of model description that can approach training dataset to the maximum extent
Distribution, therefore, training set is bigger and can more react the true distribution of user characteristics, and the performance of user's identifying system is also better.
The E step of expectation-maximization algorithm can be according to the mean value and variance of Baum-Welch normalized set acoustic feature, it is expected that maximum
The mean value and maximum variance for the acoustic feature that the M step for changing algorithm can be such that E step calculates, are walked by iterative solution E step and M, can
To obtain global variable space matrix.
Global variable space matrix can reduce the dimension of mean value super vector, reduce unwanted redundancy, reservation and user
The relevant low-dimensional vector of identity indicates.In embodiments of the present invention, preparatory trained DNN (Deep Neural can be passed through
Networks, deep neural network) model calculating Baum-Welch statistic, the accuracy of the frame alignment of voice messaging is improved,
And then the available voiceprint with user's strong correlation, improve the accuracy rate of identification.
Step S1025: according to fixed reference feature, target signature and global variable space matrix and according to shown in formula (1)
Method obtain voice messaging acoustic feature:
W=(s-m)/T (1)
Wherein, w indicates that acoustic feature, s indicate target signature, and m indicates that fixed reference feature, T indicate global variable space matrix.
After obtaining fixed reference feature, target signature and global variable space matrix, voice messaging can be further obtained
Acoustic feature, acoustic feature had both included the information of user, further included the channel information of voice, but the presence meeting of channel information
Interference is generated to user identity identification.
Step S1026: channel compensation is carried out to acoustic feature, obtains the corresponding voiceprint of voice messaging.
It, can be using general since the acoustic feature of voice messaging includes that channel information can generate interference to user identity identification
Rate linear discriminant analysis (PLDA, Probabilistic Linear Discriminant Analysis) method is to acoustic feature
Carry out channel compensation.Specifically, the voiceprint of user speech information, each user can be obtained according to expectation-maximization algorithm
Every words can generate a voiceprint, the voiceprint of the same user is averaging, it is unique that the user can be obtained
Voiceprint.
Step S103: user's vocal print is obtained according to the one-to-one relationship of preset multiple voiceprints and multiple User ID
The corresponding User ID of information.
When user specifies duration typing voice according to the prompt progress specified quantity of onboard system, onboard system can be deposited
It stores up the registration voice of multiple common users and obtains the corresponding voiceprint of user's registration voice according to step S101-S102,
And the one-to-one relationship of multiple voiceprints Yu multiple User ID is established, and the corresponding relationship is stored in onboard system.
After identifying user speech information corresponding user's voiceprint, from the one of preset multiple voiceprints and multiple User ID
User ID corresponding with user's voiceprint is obtained in one corresponding relationship, is associated with so as to subsequent according to User ID execution
Vehicle configuration operation.
Specifically, the similarity between multiple voiceprints and user's voiceprint can be calculated separately, respectively judgement meter
Whether obtained multiple similarities are more than or equal to preset similarity threshold:
A similarity more than or equal to similarity threshold if it exists, then obtain the corresponding use of voiceprint of the similarity
Family ID；
Multiple similarities more than or equal to similarity threshold if it exists, then obtain the sound of maximum similarity in multiple similarities
The corresponding User ID of line information.
In practical applications, it is understood that there may be the user for being not belonging to registration sound-groove model library interacts with vehicle, in order to mention
The accuracy of height identification, can calculate the similarity between multiple voiceprints and user's voiceprint, and judgement calculates respectively
Whether obtained multiple similarities are more than or equal to preset similarity threshold, if multiple similarities are respectively less than similarity threshold,
Illustrate the voiceprint of the user not in registration sound-groove model library, it is multiple similar more than or equal to similarity threshold if it exists
Degree, then using the corresponding User ID of the maximum voiceprint of similarity as the User ID of the user.In embodiments of the present invention, phase
It can be Likelihood Score like degree.
Step S104: the associated vehicle configuration operation of User ID corresponding with voice messaging in advance is executed.
In practical applications, in order to improve the driving experience of user, onboard system can obtain the traveling number of user automatically
According to driving data so that the later period provides intelligent service for user.Wherein, running data may include the driving that user often goes
Place and point-of-interest etc., driving data may include that user adjusts height of seat and backrest angle etc..When onboard system obtains
After taking the corresponding User ID of family voiceprint, the associated vehicle configuration behaviour of User ID corresponding with voice messaging in advance is executed
Make, wherein vehicle configuration operation may include body configurations operation, entertainment configuration operation and driving configuration operation.Car body is matched
Setting operation may include adjusting vehicle rearview mirror angle, seat back angle and height of seat, can also include the sky of vehicle
Adjust set temperature etc..After user gets on the bus, vehicle can be adjusted to above-mentioned hardware facility to meet users ' individualized requirement automatically,
In practical application scene, by taking height of seat as an example, if the height gap of a upper user and active user are larger, at upper one
After user has used vehicle, active user gets on the bus after determining identity, and onboard system can be automatically according to the height of active user
Height of seat is adjusted to suitable position by it.Entertainment configuration operation may include logging in and the associated multimedia account of User ID
Deng, it specifically, can be with automated log on and the associated multimedia account of the User ID after determining user identity, acquisition user's sense
The multimedia messages of interest, such as the music list of user's collection can obtain user when user wants to play music in time
Interested music.Driving configuration operation, which may include that acquisition User ID is corresponding, often goes to driving place and point-of-interest etc.,
After determining user identity, available User ID is corresponding often to go to drive place, such as CompanyAddress, the home address of user,
Further, it is also possible to obtain the point-of-interest of user, the power-up website gone as usual is rationally advised when vehicle needs to be powered on for user
Draw route.
Vehicle configuration method provided by the invention, the voice messaging of available user simultaneously identify the corresponding use of voice messaging
Family voiceprint obtains user's voiceprint pair according to the one-to-one relationship of preset multiple voiceprints and multiple User ID
The User ID answered, executes User ID corresponding with user's voiceprint associated vehicle configuration operation in advance, user with vehicle
During carrying out interactive voice, vehicle can execute automatically and the associated vehicle of the User ID according to the individual cultivation before user
Configuration operation, improve the car experience of user.
Although each step is described in the way of above-mentioned precedence in above-described embodiment, this field
Technical staff is appreciated that the effect in order to realize the present embodiment, executes between different steps not necessarily in such order,
It (parallel) execution simultaneously or can be executed with reverse order, these simple variations all protection scope of the present invention it
It is interior.
Refering to attached drawing 3, Fig. 3 illustratively gives the primary structure of vehicle configuration system in the present embodiment.Such as Fig. 3 institute
Show, vehicle configuration system includes that user's voiceprint acquiring unit 1, User ID acquiring unit 2 and vehicle configuration operation execute
Unit 3.
User's voiceprint acquiring unit 1 is configured to obtain the voice messaging of user and identifies the corresponding use of voice messaging
Family voiceprint；
User ID acquiring unit 2 is configured to be closed according to preset multiple voiceprints and the one-to-one correspondence of multiple User ID
System obtains the corresponding User ID of user's voiceprint；
It is associated to be configured to execute User ID corresponding with user's voiceprint in advance for vehicle configuration operation execution unit 3
Vehicle configuration operation.
Optionally, in provided vehicle configuration system, user's voiceprint acquiring unit 1 is further configured to execute
It operates below:
The mean value of each gauss of distribution function in the universal background model constructed in advance is obtained, and is joined according to mean value computation
Examine feature；
The phonetic feature of voice messaging is extracted, and interpolation is carried out to gauss of distribution function according to phonetic feature, obtains language
The corresponding gauss of distribution function of sound feature；
According to the mean value computation target signature of the corresponding gauss of distribution function of phonetic feature；
Global variable space matrix is obtained according to expectation-maximization algorithm and voice messaging；
It is obtained according to fixed reference feature, target signature and global variable space matrix and according to method shown in formula (1)
The acoustic feature of voice messaging:
Channel compensation is carried out to acoustic feature, obtains the corresponding voiceprint of voice messaging；
Wherein, universal background model is the mould constructed based on preset gauss hybrid models and preset voice training collection
Type.
Optionally, in provided vehicle configuration system, vehicle configuration system further includes model training unit, model instruction
Practice unit to be configured to execute following operation:
Firstly, the phonetic feature that voice training concentrates voice messaging is extracted, by speech detection algorithms by phonetic feature
Quiet data is rejected, and universal background model training characteristics are obtained.
Then, universal background model is trained according to universal background model training characteristics and using expectation-maximization algorithm.
Optionally, in provided vehicle configuration system, voice training collection includes multiple preset sub- training sets, accordingly
,
The quantity of universal background model is that multiple and each universal background model and every sub- training set correspond.
Optionally, in provided vehicle configuration system, User ID acquiring unit 2 can be further configured to execute with
Lower operation:
Firstly, calculating separately the similarity between multiple voiceprints and user's voiceprint.
Then, judge whether the multiple similarities being calculated are more than or equal to preset similarity threshold respectively:
A similarity more than or equal to similarity threshold if it exists, then obtain the corresponding use of voiceprint of the similarity
Family ID；
Multiple similarities more than or equal to similarity threshold if it exists, then obtain the sound of maximum similarity in multiple similarities
The corresponding User ID of line information.
Optionally, in provided vehicle configuration system, vehicle configuration operation may include body configurations operation, amusement
Configuration operation and driving configuration operation.
Optionally, in provided vehicle configuration system, body configurations operation may include adjusting vehicle rearview pitch-angle
Degree, seat back angle and height of seat, entertainment configuration operation may include log in the associated multimedia account of User ID,
Driving configuration operation, which may include that acquisition User ID is corresponding, often goes to driving place.
Further, it is based on the above system embodiment, it may include above-mentioned vehicle that the present invention also provides a kind of vehicle devices
Configuration system.In the present embodiment, vehicle device can be that by the control system of specific function, in some cases, may
Directly refer to the controller of the control system.It should be noted that protection scope of the present invention is not limited to this specific implementation
Mode.Under the premise of without departing from the principle of the present invention, it should be appreciated by the person skilled in the art that vehicle device can be vehicle-mounted joy
Happy information system can also be vehicle-mounted central control system and carried-on-vehicle computer system etc., the technology after these changes or replacement
Scheme will fall within the scope of protection of the present invention.
Still further, being based on above-mentioned vehicle device embodiment, it may include above-mentioned reality that the present invention also provides a kind of vehicles
Apply vehicle device described in example.
Still further, being based on above method embodiment, the present invention also provides a kind of storage devices, wherein being stored with more
Program, the program may be adapted to be loaded as processor to execute vehicle configuration method described in above method embodiment.
Still further, being based on above method embodiment, it may include place that the present invention also provides a kind of control devices
Manage device, storage equipment；Storage equipment may be adapted to store a plurality of program, which may be adapted to be loaded by processor on to execute
State vehicle configuration method described in embodiment of the method.
Person of ordinary skill in the field can be understood that, for convenience of description and succinctly, system of the present invention
The specific work process and related description of system, vehicle device and vehicle embodiments, can be with reference to the correspondence in preceding method embodiment
Process, and with above method beneficial effect having the same, details are not described herein.
Those skilled in the art should be able to recognize that, side described in conjunction with the examples disclosed in the embodiments of the present disclosure
Method step, can be realized with electronic hardware, computer software, or a combination of the two, in order to clearly demonstrate electronic hardware and
The interchangeability of software generally describes each exemplary composition and step according to function in the above description.These
Function is executed actually with electronic hardware or software mode, specific application and design constraint depending on technical solution.
Those skilled in the art can use different methods to achieve the described function each specific application, but this reality
Now it should not be considered as beyond the scope of the present invention.
So far, it has been combined preferred embodiment shown in the drawings and describes technical solution of the present invention, still, this field
Technical staff is it is easily understood that protection scope of the present invention is expressly not limited to these specific embodiments.Without departing from this
Under the premise of the principle of invention, those skilled in the art can make equivalent change or replacement to the relevant technologies feature, these
Technical solution after change or replacement will fall within the scope of protection of the present invention.
Claims (10)
1. a kind of vehicle configuration method, characterized by comprising:
It obtains the voice messaging of user and identifies the corresponding user's voiceprint of the voice messaging；
It is corresponding with the one-to-one relationship of multiple User ID acquisition user's voiceprint according to preset multiple voiceprints
User ID；
Execute the associated vehicle configuration operation of User ID corresponding with user's voiceprint in advance.
2. vehicle configuration method according to claim 1, which is characterized in that " identify the corresponding user of the voice messaging
The step of voiceprint " includes:
The mean value of each gauss of distribution function in the universal background model constructed in advance is obtained, and is joined according to the mean value computation
Examine feature；
The phonetic feature of the voice messaging is extracted, and according to the phonetic feature gauss of distribution function is carried out slotting
Value, obtains the corresponding gauss of distribution function of the phonetic feature；
According to the mean value computation target signature of the corresponding gauss of distribution function of the phonetic feature；
Global variable space matrix is obtained according to expectation-maximization algorithm and the voice messaging；
Described in method shown according to the fixed reference feature, target signature and global variable space matrix and according to the following formula obtains
The acoustic feature of voice messaging:
W=(s-m)/T
Wherein, w indicates that the acoustic feature, s indicate the target signature, and m indicates that the fixed reference feature, T indicate the overall situation
Variable space matrix；
Channel compensation is carried out to the acoustic feature, obtains the corresponding voiceprint of the voice messaging；
Wherein, the universal background model is the mould constructed based on preset gauss hybrid models and preset voice training collection
Type.
3. vehicle configuration method according to claim 2, which is characterized in that " obtaining the common background mould constructed in advance
Before the step of mean value of each gauss of distribution function in type, and according to the mean value computation fixed reference feature ", the method is also
Include:
The phonetic feature that the voice training concentrates voice messaging is extracted, and utilizes speech detection algorithms by the phonetic feature
Quiet data reject, obtain universal background model training characteristics；
According to the universal background model training characteristics and utilize the expectation-maximization algorithm training common background mould
Type.
4. vehicle configuration method according to claim 3, which is characterized in that the voice training collection includes multiple preset
Sub- training set；
The quantity of the universal background model is multiple and each universal background model and each sub- training set one
One is corresponding.
5. vehicle configuration method according to claim 1, which is characterized in that " according to preset multiple voiceprints and more
The one-to-one relationship of a User ID obtains the corresponding User ID of user's voiceprint " the step of include:
Calculate separately the similarity between the multiple voiceprint and user's voiceprint；
Judge whether the multiple similarities being calculated are more than or equal to preset similarity threshold respectively:
A similarity more than or equal to the similarity threshold if it exists, then obtain the corresponding use of voiceprint of the similarity
Family ID；
Multiple similarities more than or equal to the similarity threshold if it exists, then obtain maximum similarity in the multiple similarity
The corresponding User ID of voiceprint.
6. vehicle configuration method according to claim 1, which is characterized in that the vehicle configuration operation includes body configurations
Operation, entertainment configuration operation and driving configuration operation.
7. vehicle configuration method according to claim 6, which is characterized in that the body configurations operation includes adjusting vehicle
Door mirror angle, seat back angle and height of seat, the entertainment configuration operation include logging in be associated with the User ID
Multimedia account, driving configuration operation includes obtaining that the User ID is corresponding often to go to drive place.
8. a kind of vehicle configuration system, characterized by comprising:
User's voiceprint acquiring unit is configured to obtain the voice messaging of user and identifies the corresponding user of the voice messaging
Voiceprint；
User ID acquiring unit is configured to be obtained according to the one-to-one relationship of preset multiple voiceprints and multiple User ID
The corresponding User ID of user's voiceprint；
Vehicle configuration operation execution unit is configured to execute the associated vehicle of User ID corresponding with user's voiceprint in advance
Configuration operation.
9. vehicle configuration system according to claim 8, which is characterized in that user's voiceprint acquiring unit is into one
Step is configured to execute following operation:
The mean value of each gauss of distribution function in the universal background model constructed in advance is obtained, and is joined according to the mean value computation
Examine feature；
The phonetic feature of the voice messaging is extracted, and according to the phonetic feature gauss of distribution function is carried out slotting
Value, obtains the corresponding gauss of distribution function of the phonetic feature；
According to the mean value computation target signature of the corresponding gauss of distribution function of the phonetic feature；
Global variable space matrix is obtained according to expectation-maximization algorithm and the voice messaging；
Described in method shown according to the fixed reference feature, target signature and global variable space matrix and according to the following formula obtains
The acoustic feature of voice messaging:
W=(s-m)/T
Wherein, w indicates that the acoustic feature, s indicate the target signature, and m indicates that the fixed reference feature, T indicate the overall situation
Variable space matrix；
Channel compensation is carried out to the acoustic feature, obtains the corresponding voiceprint of the voice messaging；
Wherein, the universal background model is the mould constructed based on preset gauss hybrid models and preset voice training collection
Type.
10. vehicle configuration system according to claim 9, which is characterized in that the vehicle configuration system further includes model
Training unit, the model training unit are configured to execute following operation:
The phonetic feature that the voice training concentrates voice messaging is extracted, by speech detection algorithms by the quiet of the phonetic feature
Sound data are rejected, and universal background model training characteristics are obtained；
According to the universal background model training characteristics and utilize the expectation-maximization algorithm training universal background model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811260569.6A CN109273002B (en) | 2018-10-26 | 2018-10-26 | Vehicle configuration method and system, vehicle machine and vehicle |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811260569.6A CN109273002B (en) | 2018-10-26 | 2018-10-26 | Vehicle configuration method and system, vehicle machine and vehicle |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109273002A true CN109273002A (en) | 2019-01-25 |
| CN109273002B CN109273002B (en) | 2022-06-10 |
Family
ID=65193843
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811260569.6A Active CN109273002B (en) | 2018-10-26 | 2018-10-26 | Vehicle configuration method and system, vehicle machine and vehicle |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109273002B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110932862A (en) * | 2019-10-22 | 2020-03-27 | 安信通科技（澳门）有限公司 | Vehicle-mounted equipment login management method and device in vehicle |
| CN112185344A (en) * | 2020-09-27 | 2021-01-05 | 北京捷通华声科技股份有限公司 | Voice interaction method and device, computer readable storage medium and processor |
| CN112298104A (en) * | 2019-07-31 | 2021-02-02 | 比亚迪股份有限公司 | Vehicle control method and device, storage medium, electronic equipment and vehicle |
| CN113888777A (en) * | 2021-09-08 | 2022-01-04 | 南京金盾公共安全技术研究院有限公司 | Voiceprint unlocking method and device based on cloud machine learning |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103730114A (en) * | 2013-12-31 | 2014-04-16 | 上海交通大学无锡研究院 | Mobile equipment voiceprint recognition method based on joint factor analysis model |
| CN103809854A (en) * | 2014-02-20 | 2014-05-21 | 联想(北京)有限公司 | Data processing method and electronic equipment |
| CN203753082U (en) * | 2014-02-11 | 2014-08-06 | 北汽福田汽车股份有限公司 | Identity recognition based vehicle setting system and vehicle |
| CN105895096A (en) * | 2016-03-30 | 2016-08-24 | 乐视控股（北京）有限公司 | Identity identification and voice interaction operating method and device |
| US20160358610A1 (en) * | 2013-01-28 | 2016-12-08 | Tencent Technology (Shenzhen) Company Limited | Method and device for voiceprint recognition |
| CN106683673A (en) * | 2016-12-30 | 2017-05-17 | 智车优行科技（北京）有限公司 | Method, device and system for adjusting driving modes and vehicle |
| CN107886943A (en) * | 2017-11-21 | 2018-04-06 | 广州势必可赢网络科技有限公司 | Voiceprint recognition method and device |
-
2018
- 2018-10-26 CN CN201811260569.6A patent/CN109273002B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160358610A1 (en) * | 2013-01-28 | 2016-12-08 | Tencent Technology (Shenzhen) Company Limited | Method and device for voiceprint recognition |
| CN103730114A (en) * | 2013-12-31 | 2014-04-16 | 上海交通大学无锡研究院 | Mobile equipment voiceprint recognition method based on joint factor analysis model |
| CN203753082U (en) * | 2014-02-11 | 2014-08-06 | 北汽福田汽车股份有限公司 | Identity recognition based vehicle setting system and vehicle |
| CN103809854A (en) * | 2014-02-20 | 2014-05-21 | 联想(北京)有限公司 | Data processing method and electronic equipment |
| CN105895096A (en) * | 2016-03-30 | 2016-08-24 | 乐视控股（北京）有限公司 | Identity identification and voice interaction operating method and device |
| CN106683673A (en) * | 2016-12-30 | 2017-05-17 | 智车优行科技（北京）有限公司 | Method, device and system for adjusting driving modes and vehicle |
| CN107886943A (en) * | 2017-11-21 | 2018-04-06 | 广州势必可赢网络科技有限公司 | Voiceprint recognition method and device |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112298104A (en) * | 2019-07-31 | 2021-02-02 | 比亚迪股份有限公司 | Vehicle control method and device, storage medium, electronic equipment and vehicle |
| CN110932862A (en) * | 2019-10-22 | 2020-03-27 | 安信通科技（澳门）有限公司 | Vehicle-mounted equipment login management method and device in vehicle |
| CN112185344A (en) * | 2020-09-27 | 2021-01-05 | 北京捷通华声科技股份有限公司 | Voice interaction method and device, computer readable storage medium and processor |
| CN113888777A (en) * | 2021-09-08 | 2022-01-04 | 南京金盾公共安全技术研究院有限公司 | Voiceprint unlocking method and device based on cloud machine learning |
| CN113888777B (en) * | 2021-09-08 | 2023-08-18 | 南京金盾公共安全技术研究院有限公司 | Voiceprint unlocking method and device based on cloud machine learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109273002B (en) | 2022-06-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR102562227B1 (en) | Dialogue system, Vehicle and method for controlling the vehicle | |
| US20210249035A1 (en) | Emotion detection using speaker baseline | |
| CN109273002A (en) | Vehicle configuration method, system, vehicle device and vehicle | |
| CN111415677B (en) | Method, apparatus, device and medium for generating video | |
| JP3667332B2 (en) | Standard model creation apparatus and standard model creation method | |
| EP3090429B1 (en) | Modifying operations based on acoustic ambience classification | |
| Bisio et al. | Smart and robust speaker recognition for context-aware in-vehicle applications | |
| CN105390136B (en) | Vehicle arrangement control device and method for user's adaptive type service | |
| CN110178178A (en) | Microphone selection and multiple talkers segmentation with environment automatic speech recognition (ASR) | |
| CN106663446A (en) | User environment aware acoustic noise reduction | |
| CN108875682A (en) | Information-pushing method and device | |
| JP2017009826A (en) | Group state determination device and group state determination method | |
| US11393473B1 (en) | Device arbitration using audio characteristics | |
| Kashevnik et al. | Multimodal corpus design for audio-visual speech recognition in vehicle cabin | |
| Gajšek et al. | Speaker state recognition using an HMM-based feature extraction method | |
| JP2017009825A (en) | Conversation state analyzing device and conversation state analyzing method | |
| US20220360641A1 (en) | Dynamic time-based playback of content in a vehicle | |
| JP2019158975A (en) | Utterance system | |
| Chakraborty et al. | Literature Survey | |
| US11195534B1 (en) | Permissioning for natural language processing systems | |
| Loh et al. | Speech recognition interactive system for vehicle | |
| JP2005227794A (en) | Device and method for creating standard model | |
| CN111861666A (en) | Vehicle information interaction method and device | |
| WO2023159536A1 (en) | Human-computer interaction method and apparatus, and terminal device | |
| CN112863476A (en) | Method and device for constructing personalized speech synthesis model, method and device for speech synthesis and testing |
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 | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20200901 Address after: Susong Road West and Shenzhen Road North, Hefei Economic and Technological Development Zone, Anhui Province Applicant after: Weilai (Anhui) Holding Co., Ltd Address before: 30 Floor of Yihe Building, No. 1 Kangle Plaza, Central, Hong Kong, China Applicant before: NIO NEXTEV Ltd. |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |