CN104464733B - A kind of more scene management method and devices of voice dialogue - Google Patents
A kind of more scene management method and devices of voice dialogue Download PDFInfo
- Publication number
- CN104464733B CN104464733B CN201410590076.4A CN201410590076A CN104464733B CN 104464733 B CN104464733 B CN 104464733B CN 201410590076 A CN201410590076 A CN 201410590076A CN 104464733 B CN104464733 B CN 104464733B
- Authority
- CN
- China
- Prior art keywords
- scene
- demand information
- characteristic
- feature
- maximum value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Telephonic Communication Services (AREA)
- Machine Translation (AREA)
Abstract
The present invention provides a kind of more scene management method and devices of voice dialogue, this method comprises: obtaining the demand information of user's input from text information, wherein the text information carries out text identification from the voice messaging of the user and obtains；At least one corresponding score value of at least one scene in scene is obtained according to the demand information；The scene switching to be executed movement is determined according at least one described score value, and shows voice content corresponding with the scene after switching.The embodiment of the present invention can well solve more scene switching problems in user and conversational system during voice dialogue.
Description
Technical field
The present invention relates to technical field of voice recognition more particularly to the more scene management methods and dress of a kind of voice dialogue
It sets.
Background technique
With the continuous development of speech recognition technology and development of Mobile Internet technology, voice inputs the Heterosis in mobile terminal
It obtains more obvious.As major Internet company issues speech dialogue system respectively, by oneself however the voice of low cost is inputted,
To understand the demand of user and be solved the problems, such as user.
In speech recognition process, it is understood that there may be the multi-field dialogue of more scenes, and need to solve in multi-field more wheels pair
Decision problem during words.More scene managements in the prior art, first is that rule-based (rule-based), passes through formulation
A series of rule realizes the management switched between scene；First is that being based on disaggregated model, is used and divided by current system conditions
Class model prediction is following to execute movement.
Rule-based method needs rule author to have good background knowledge, as the factor that rule is related to becomes more,
Processing logic becomes complicated, and effect cannot be optimal state；Rule-based more scene managements are not bound with the feedback of user
Information, therefore the true service condition of user is not known about, the decision movement ultimately generated is not necessarily most reasonable.
Summary of the invention
The embodiment of the present invention provides a kind of more scene management method and devices of voice dialogue, and realization is effectively performed more
Scape handover management.
In order to achieve the above objectives, the embodiment of the present invention adopts the following technical scheme that
A kind of more scene management methods of voice dialogue, this method comprises:
The demand information of user's input is obtained from text information, wherein voice of the text information from the user
Text identification is carried out in information to obtain；
At least one corresponding score value of at least one scene in scene is obtained according to the demand information；
The scene switching to be executed movement is determined according at least one described score value, and is shown and the scene phase after switching
Corresponding voice content.
A kind of more scene management devices of voice dialogue, the device include:
First obtains module, for obtaining the demand information of user's input from text information, wherein the text information
Text identification is carried out from the voice messaging of the user to obtain；
Second obtains module, corresponding extremely for obtaining at least one scene in scene according to the demand information
A few score value；
Switching module, for determining the scene switching to be executed movement according at least one described score value, and show with
The corresponding voice content of scene after switching.
More scene management method and devices of voice dialogue provided in an embodiment of the present invention, by being obtained from Textual information
User input demand information, according to user input demand information obtain scene at least one scene, thus for for
Voice content that is being executed and being suitable for user demand is provided, can be well solved in user and conversational system in voice pair
More scene switching problems during words.
Detailed description of the invention
The relation schematic diagram for more scene managements that Fig. 1 is applicable in by the embodiment of the present invention.
Fig. 2 is the flow diagram of the more scene management methods for the voice dialogue that the embodiment of the present invention one provides.
Fig. 3 is the flow diagram of more scene management methods of voice dialogue provided by Embodiment 2 of the present invention.
Fig. 4 is the schematic diagram of scene switching action in the embodiment of the present invention two.
Fig. 5 is the flow diagram of the more scene management methods for the voice dialogue that the embodiment of the present invention three provides.
Fig. 6 is the structural schematic diagram of the more scene management devices for the voice dialogue that the embodiment of the present invention four provides.
Fig. 7 is the structural schematic diagram of the more scene management devices for the voice dialogue that the embodiment of the present invention five provides.
Specific embodiment
More scene management method and devices of voice dialogue provided in an embodiment of the present invention are carried out with reference to the accompanying drawing detailed
Thin description.
More wheel dialogues described in the embodiment of the present invention, refer to and help user complete with interacting for more wheels of terminal by user
At specific demand or task.In embodiments of the present invention, current more wheels dialogue is divided according to field (alternatively, scene), can
Think more wheels dialogue of Task.The relation schematic diagram for more scene managements that Fig. 1 is applicable in by the embodiment of the present invention, such as Fig. 1 institute
Show, may include multiple scenes in embodiments of the present invention, for example, travelling scene, food and drink scene, weather scene and stock
Scene, for another example travelling scene can be divided into multiple sub-scenes: air ticket scene, train ticket scene and hotel's scene.
Specifically, the embodiment of the present invention is illustrated for talking with as follows:
User: the air ticket in Qingdao is removed.
Mobile terminal: it is good, to the plane ticket in Qingdao.It may I ask when you leave?
User: tomorrow sets out.
Mobile terminal: good, the plane ticket of tomorrow to Qingdao.You it is estimated by where?
User: Beijing.
Talk at this point, mobile terminal by the relative information displaying of air ticket to user.
User: does is there nice Sichuan cuisine in Qingdao?
(" having the Sichuan cuisine of nice acid? " has been arrived in speech recognition)
Will mobile terminal: you be after renewal air tickets?
User: it is not.
Mobile terminal: the Sichuan cuisine for having nice acid is heard?
User: tomorrow goes to Qingdao.
Talk at this point, the Sichuan cuisine shop accordingly in Qingdao is showed user by mobile terminal.
User: the dining room near Qingdao Airport.
Mobile terminal: following dining room (and carrying out corresponding result displaying) is found for you
Can user: Qingdao rain tomorrow?
Mobile terminal: Qingdao shower tomorrow, 16 degrees Celsius to 20 degrees Celsius of temperature, 5-6 grades of east wind.
Embodiment one:
Fig. 2 is the flow diagram of the more scene management methods for the voice dialogue that the embodiment of the present invention one provides, such as Fig. 1 institute
Show, the embodiment of the present invention includes following steps:
Step 101, the demand information of user's input is obtained from text information, wherein voice of the text information from user
Text identification is carried out in information to obtain.
Step 102, at least one scene in acquisition of information scene, and acquisition and at least one scene difference according to demand
At least one corresponding feature vector.
Step 103, the inner product for obtaining at least one feature vector Yu corresponding weight vectors, obtains at least one
Product.
Step 104, it determines that scene switching to be executed acts according at least one inner product, and shows and the scene phase after switching
Corresponding voice content.
In a step 101, the voice messaging of user is converted to by text information by speech recognition, according to the present invention
One embodiment obtains the demand information of user from the text information that identification obtains, for example, user, which inputs voice, " goes to Qingdao
Air ticket ", after which is identified as text information, the demand information for getting user's input is " air ticket ".
In a step 102, at least one scene in scene is obtained according to the demand information obtained in step 101, one
In a embodiment, at least one scene in scene can be judged according in the contextual information of voice dialogue.Wherein, field
In scape for multiple scenes preset in conversational system (for example, travelling scene, food and drink scene, weather scene shown in Fig. 1 with
And stock scene), specifically, the demand information " air ticket " of user's input is got in a step 101, it can be according to the demand information
(scene of wherein, travelling can also include air ticket scene, train ticket scene and hotel's scene etc. to the tourism scene for getting in scene
Multiple sub-scenes), further, which has corresponded to air ticket scene this sub-scene in the tourism scene.At one
In embodiment, at least one feature vector corresponding with the tourism scene can be obtained from voice messaging, for example, believing in voice
It ceases in " air ticket for going to Qingdao ", the feature that " going, Qingdao, air ticket " forms the voice messaging quantifies features described above
Feature vector is formed to specifically include in this feature vector: destination (Qingdao), air ticket (demand information), in addition, of the invention
It can also be including but not limited to following information in feature vector in embodiment: departure place, date, type of seat, departure time
Etc. information.In one embodiment, departure place, destination and date are essential information, and type of seat, departure time are optional letter
Breath；By features described above vector, it can make the embodiment of the present invention that there is good generalization ability, avoid every increasing in the prior art
Add a new scene will corresponding labeled data, and model corresponding to Training scene again.
In step 103, obtain at least one feature vector for obtaining in a step 102 and with its corresponding weight
At least one inner product of vector is (for example, obtain inner product are as follows: A1, A2, A3..., An, n is the number of inner product), wherein weight vectors
It is the corresponding weight vectors of scene characteristic obtained according to the training of the corpus of collection, it will be appreciated by persons skilled in the art that
The embodiment of the present invention is specially that inner product illustrates with score value, and the specific calculation of inner product can not be formed to this
The limitation of inventive embodiments.
At step 104, determine that the scene switching to be executed is dynamic according at least one inner product obtained in step 103
Make, and shows voice content corresponding with the scene after switching.An embodiment according to the present invention at step 104 will at least
One inner product is ranked up, and obtains the maximum value at least one inner product, by the corresponding scene switching movement conduct pair of the inner product
It answers the decision of scene to act, and it is fed back into user by way of voice content.In one embodiment, user is got
The corresponding scene of demand information " air ticket " feature vector, it is A that its inner product, which is calculated,1、A2、A3、A4, inner product is obtained after sequence
In maximum value be A2, then by A2Corresponding voice content (for example, its voice content be " it is good, to the plane ticket in Qingdao, may I ask
When you leave ") it exports to user.
More scene management methods of voice dialogue provided in an embodiment of the present invention, it is defeated by obtaining user from Textual information
The demand information entered obtains at least one scene in scene according to the demand information of user's input, thus for for providing desire
Scene switching that is executing and being suitable for user demand acts, and shows voice content corresponding with the scene after switching, energy
The problem of enough well solving more scene switchings of voice dialogue in conversational system.In addition, indicating that scene makes by feature vector
Conversational system has good generalization ability, can quickly increase new scene into system, and then be effectively performed more
Scape handover management, moreover it is possible to fully understand the true service condition of user, provide most reasonable movement decision for user, enhance use
Family experience.
Embodiment two:
Fig. 3 is the flow diagram of more scene management methods of voice dialogue provided by Embodiment 2 of the present invention, and Fig. 4 is this
The schematic diagram of scene switching action in inventive embodiments two；As shown in figure 3, the embodiment of the present invention includes following steps:
Step 201, the demand information of user's input is obtained from text information, wherein voice of the text information from user
Text identification is carried out in information to obtain.
Step 202, scene classification is carried out to voice dialogue according to the demand information identified in step 201, is needed
At least one scene in the scene for asking information to be applicable in.
Step 203, at least one scene according to obtained in step 202 carries out scene characteristic extraction to demand information, obtains
To at least one feature vector corresponding at least one scene.
Step 204, the inner product for obtaining at least one feature vector Yu corresponding weight vectors, obtains at least one
Product.
Step 205, at least one inner product is ranked up, obtains the maximum value in all inner products.
Step 206, scene switching movement is carried out to demand information according to the corresponding scene characteristic of maximum value, and shows and cuts
The corresponding speech response of scene after changing.
In step 201, it can be no longer described in further detail herein with the description of the step 101 in reference implementation example one.
In step 202, scene classification is carried out to voice dialogue according to the demand information obtained in step 201, obtained
Suitable at least one scene in scene, for example, the demand information of user's input is " Qingdao ", " air ticket ", it can be by the voice
Dialogue is categorized into the sub-scene air ticket scene of travelling scene.After classification obtains multiple scenes, in step 203, according to this
Scape carries out scene characteristic extraction to demand information, gets corresponding feature vector.
In step 203 and step 204, can with the step 102 and step 103 in reference implementation example one, herein no longer into
One step is described in detail.
In step 205, at least one inner product obtained in step 204 is ranked up, obtains the maximum in inner product
Value, for example, the feature vector of the corresponding scene of the demand information " air ticket " for getting user, it is A that its inner product, which is calculated,1, A2, A3,
It is A that the maximum value in inner product is obtained after sequence2。
In step 206, Fig. 4 is that the schematic diagram of scene switching action in the embodiment of the present invention two is corresponding according to maximum value
Scene characteristic, the voice messaging that response is adapted with demand information, and voice content is fed back into user, for example, in step
Maximum value A in the inner product referred in 2052Corresponding voice content be " it is good, to the plane ticket in Qingdao, may I ask your what when
Time is left ", during voice dialogue, then this section of voice content is fed back into user.
It will be appreciated by persons skilled in the art that setting and study for scene are not in actual application process
It may be exhaustive, it is also possible to the scene characteristic (the outer feature of scene) outside default scene occur, one implements according to the present invention
Example generates the feature vector of scene confirmation movement according to feature outside scene and at least one scene characteristic, scene confirmation movement
Feature vector is one at least one feature vector, further, if maximum value obtained in step 205 corresponds to scene
An interior scene characteristic, responds demand information according to the scene characteristic；If maximum value corresponds to two in scene
Above feature vector clarifies demand information according to more than two feature vectors；If maximum value corresponds to outside scene
Scene characteristic in feature and scene, confirms feature outside scene and the scene characteristic in scene.
In scene clarifying process, the difference of at least two corresponding scene vector of scene characteristic of acquisition can be passed through
Value, obtains the exponent arithmetic of the difference, is determined according to exponent arithmetic result and clarifies feature vector to two scenes；For example, having two
The feature vector f_1 and feature vector f_2 of a scene, calculate the difference f_1-f_2 of two scene characteristics, further calculate the difference
It is worth corresponding exponent e ^ (f_1-f_2), wherein e indicates natural constant, it is, of course, also possible to transport using other numerical value as index
The truth of a matter of calculation.It is determined according to the operation result of the index and clear feature vector is carried out to two scenes, it is specifically, scene is clear
The clear weight vector computation inner product of clear feature vector and scene, obtains the clear score of the two scenes, when the score value
When maximum, two scenes are clarified.
For example, during above-mentioned more wheel voice dialogues, when " Qingdao has the voice messaging that mobile terminal inputs user
When the Sichuan cuisine eaten " identifies " Sichuan cuisine for having nice acid ", mobile terminal passes through this according to text information at this time
Inventive embodiments two, mobile terminal can be used in conjunction with contextual information and parsing information when executing scene switching movement
Scene confirmation, and illustrate scene and confirm corresponding speech response " you will be after renewal air tickets ", so that user be made to carry out field
Scape confirmation.
Further, after user confirms "no", mobile terminal combination contextual information and parsing information are executing field
When scape switching action, use scene and clarify, and illustrate scene clarification confirm corresponding speech response " hear have it is nice
The Sichuan cuisine of acid ", so that user be made to clarify scene.
As shown in figure 5, being the flow diagram of the more scene management methods for the voice dialogue that the embodiment of the present invention three provides；
In embodiments of the present invention, exemplary theory is carried out so that mobile terminal specifically executes more scene management methods of voice dialogue as an example
It is bright, as shown in figure 5, the embodiment of the present invention includes following steps:
During off-line learning in step 501, during crowd surveys, multiple scene objects can be set, allow user
More wheel interactive voices are carried out with mobile terminal, so that mobile terminal has certain sticgastuc deicision；Wherein, many measured data
It is one of foundation of the mobile terminal training data in the embodiment of the present invention, the embodiment of the present invention can be made to be based on training
Data can be realized on-line prediction.
During on-line study in step 502, if voice dialogue is related to take turns (that is, user and mobile terminal more
Carry out multiple voice dialogue), the contextual information and parsing information of user and mobile terminal can be collected, to get spy
Vector is levied to indicate the significant condition of scene, enhances learning model to feature vector and weight vector computation inner product；Pass through the mistake
Journey enables to the embodiment of the present invention to reach global gain maximum, and by multiple groups comparative experiments, experiment effect is more than existing skill
Rule-based more scene managements in art.In addition, the feature vector that the embodiment of the present invention is unrelated with scene field by selection,
Scene characteristic is indicated using feature vector, to maximumlly cover factor relevant to scene switching, is improved extensive
Ability.The signal of feature vector may refer to Fig. 4.
In scene switching movement in step 503, the embodiment of the present invention is acted using 4 classes shown in table 1 as example
Property explanation, including but not limited to: it is outer (present (NULL)) to show scene, shows scene (present (d)), scene confirmation
(clarify (d1, d2)) is clarified between (confirm (d)) and scene.By scene confirmation with scene clarify enhance it is man-machine whole
Interaction capabilities in a more wheel dialog procedures.
Table 1
It, can using the enhancing learning model after the optimization of training in step 502 in the movement selection course of step 503
With according to the demand information of active user, prediction executes which class movement in table 1.
By the above process, the feedback information of user can be made full use of, the long-term gain that can predict user is maximum
Movement；Further, since feature vector chooses the feature unrelated with concrete scene, it is special so as to quickly introduce new scene
Sign, so that scheme has good scalability.
Example IV:
Fig. 6 is the structural schematic diagram of the more scene management devices for the voice dialogue that the embodiment of the present invention four provides；Such as Fig. 6 institute
Show, the embodiment of the present invention includes
First obtains module 41, for obtaining the demand information of user's input from text information, wherein the text envelope
Breath carries out text identification from the voice messaging of the user and obtains；
Second obtains module 42, corresponding for obtaining at least one scene in scene according to the demand information
At least one score value；
Switching module 43 for determining the scene switching to be executed movement according at least one described score value, and is shown
Voice content corresponding with the scene after switching.
Wherein, the second acquisition module 42 includes:
First acquisition unit 421, for according to the demand information obtain scene at least one scene, and obtain with
At least one corresponding feature vector of described at least one scene；
Second acquisition unit 422, for obtaining commenting at least one described feature vector and corresponding weight vectors
Score value obtains at least one score value.
Further, the first acquisition unit includes:
Scene classification subelement (not shown), for carrying out scene point to the voice dialogue according to the demand information
Class obtains at least one scene in the scene that the demand information is applicable in；
Feature extraction subelement (not shown), for carrying out field to the demand information according at least one described scene
Scape feature extraction obtains at least one feature vector corresponding at least one described scene.
The detailed description and advantageous effects of the embodiment of the present invention can be with reference to the associated descriptions in above-described embodiment one
And advantageous effects, details are not described herein.
Embodiment five:
Fig. 7 is the structural schematic diagram of the more scene management devices for the voice dialogue that the embodiment of the present invention five provides；Such as Fig. 7 institute
Show, if also getting feature outside scene from the demand information, the embodiment of the invention also includes:
Third obtains module 44, for according to feature outside the scene and at least one described scene characteristic from it is described at least
The feature vector of scene confirmation movement is obtained in one scene characteristic.
Switching module 43 includes:
Sequencing unit 431 obtains the maximum in all score values for being ranked up at least one described score value
Value；
Determination unit 432, for determining that the scene switching to be executed acts according to the corresponding scene characteristic of the maximum value,
And show the voice content of scene characteristic corresponding with the maximum value.
Further, the determination unit includes:
First responds subelement (not shown), if corresponding to a scene spy in the scene for the maximum value
Sign, responds the demand information according to the scene characteristic；
Second responds subelement (not shown), if corresponded to for the maximum value more than two in the scene
Feature vector clarifies the demand information according to described two above feature vectors；
Third responds subelement (not shown), if corresponding to the outer feature of the scene and the field for the maximum value
Scene characteristic in scape confirms feature outside the scene with the scene characteristic in the scene.
Further, third response subelement (not shown) includes:
Difference obtains subelement, for obtaining the difference of the corresponding scene vector of at least two scene characteristic
Value；
Subelement is clarified, for obtaining the exponent arithmetic of the difference, is determined according to exponent arithmetic result to described two
Above scene characteristic is clarified.
Further, the device further include:
4th obtains module 45, for obtaining the target signature of at least one scene during crowd surveys, passes through system
It counts model and more wheel voice trainings is carried out to the target signature；
5th obtains module 46, for obtaining the weight vectors when the statistical model has sticgastuc deicision
Initial value.
The detailed description and advantageous effects of the embodiment of the present invention can be with reference to the associated descriptions in above-described embodiment two
And advantageous effects, details are not described herein.
To sum up, the embodiment of the present invention can make full use of the feedback information of user, can predict the long-term gain of user
Maximum movement；Further, since feature vector chooses the feature unrelated with concrete scene, so as to quickly introduce new field
Scape feature, so that scheme has good scalability.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can easily think of the change or the replacement, and should all contain
Lid is within protection scope of the present invention.Therefore, protection scope of the present invention should be based on the protection scope of the described claims.
Claims (10)
1. a kind of more scene management methods of voice dialogue, which is characterized in that the described method includes:
The demand information of user's input is obtained from text information, wherein voice messaging of the text information from the user
Middle progress text identification obtains；
At least one corresponding score value of at least one scene in scene is obtained according to the demand information；
At least one described score value is ranked up, the maximum value in all score values is obtained；
The scene switching to be executed movement is determined according to the corresponding scene characteristic of the maximum value, and is shown and the maximum value pair
The voice content for the scene characteristic answered；
The step that at least one corresponding score value of at least one scene in scene is obtained according to the demand information
Suddenly include:
Scene classification is carried out to the voice dialogue according to the demand information, is obtained in the scene that the demand information is applicable in
At least one scene；
Scene characteristic extraction is carried out to the demand information according at least one described scene, is obtained and at least one described scene
At least one corresponding feature vector；
Obtain the corresponding weight vectors of at least one described feature vector；
The inner product for obtaining described at least one feature vector and corresponding weight vectors, obtains at least one inner product, the inner product
As score value；
The step of voice content for showing corresponding with maximum value scene characteristic includes:
If the maximum value corresponds to more than two feature vectors in the scene, according to described two above features to
Amount clarifies the demand information；
If the maximum value corresponds to the scene characteristic in the outer feature of the scene and the scene, to feature outside the scene with
Scene characteristic in the scene is confirmed.
2. the method according to claim 1, wherein if also being got from the demand information special outside scene
Sign, the method also includes:
Scene is obtained from least one described scene characteristic according to feature outside the scene and at least one described scene characteristic
The feature vector of confirmation movement.
3. the method according to claim 1, wherein the method also includes:
The target signature that at least one scene is obtained during crowd surveys carries out the target signature by statistical model
More wheel voice trainings；
When the statistical model has sticgastuc deicision, the initial value of the weight vectors is obtained.
4. the method according to claim 1, wherein described show corresponding with maximum value scene characteristic
The step of voice content further include:
If the maximum value corresponds to a scene characteristic in the scene, according to the scene characteristic to the demand information into
Row is responded.
5. the method according to claim 1, wherein it is described according to described two above scene characteristics to described
The step of demand information is clarified include:
Obtain the difference of the corresponding scene vector of at least two scene characteristic；
The exponent arithmetic for obtaining the difference determines clear to described two above scene characteristics progress according to exponent arithmetic result
Clearly.
6. a kind of more scene management devices of voice dialogue, which is characterized in that described device includes:
First obtains module, for obtaining the demand information of user's input from text information, wherein the text information is from institute
Progress text identification in the voice messaging of user is stated to obtain；
Second obtains module, for obtaining at least one scene corresponding at least one in scene according to the demand information
A score value；
Switching module, for determining the scene switching to be executed movement, and displaying and switching according at least one described score value
The corresponding voice content of scene afterwards；
Described second, which obtains module, includes:
First acquisition unit, for according to the demand information obtain scene at least one scene, and obtain with it is described extremely
Few at least one corresponding feature vector of a scene；
Second acquisition unit, for obtaining the corresponding weight vectors of at least one described feature vector；
Third acquiring unit obtains at least for obtaining the inner product of described at least one feature vector and corresponding weight vectors
One score value；
The first acquisition unit includes:
Scene classification subelement obtains the need for carrying out scene classification to the voice dialogue according to the demand information
At least one scene in the scene for asking information to be applicable in；
Feature extraction subelement is obtained for carrying out scene characteristic extraction to the demand information according at least one described scene
To at least one feature vector corresponding at least one described scene；
The switching module includes:
Sequencing unit obtains the maximum value in all score values for being ranked up at least one described score value；
Determination unit for determining that the scene switching to be executed acts according to the corresponding scene characteristic of the maximum value, and is shown
The voice content of scene characteristic corresponding with the maximum value.
The determination unit includes:
Second responds subelement, if corresponding to more than two feature vectors in the scene for the maximum value, according to
Described two above feature vectors clarify the demand information；
Third responds subelement, if it is special to correspond to the scene outside the scene in feature and the scene for the maximum value
Sign, feature outside the scene is confirmed with the scene characteristic in the scene.
7. device according to claim 6, which is characterized in that if also got from the demand information special outside scene
Sign, described device further include:
Third obtains module, for according to feature outside the scene and at least one described scene characteristic from least one described field
The feature vector of scene confirmation movement is obtained in scape feature.
8. device according to claim 6, which is characterized in that described device further include:
4th obtains module, for obtaining the target signature of at least one scene during crowd surveys, passes through statistical model
More wheel voice trainings are carried out to the target signature；
5th obtains module, for obtaining the initial value of the weight vectors when the statistical model has sticgastuc deicision.
9. device according to claim 6, which is characterized in that the determination unit further include:
First responds subelement, if a scene characteristic in the scene is corresponded to for the maximum value, according to the scene
Feature responds the demand information.
10. device according to claim 6, which is characterized in that the third responds subelement and includes:
Difference obtains subelement, for obtaining the difference of the corresponding scene vector of at least two scene characteristic；
Subelement is clarified, for obtaining the exponent arithmetic of the difference, is determined according to exponent arithmetic result to more than described two
Scene characteristic clarified.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410590076.4A CN104464733B (en) | 2014-10-28 | 2014-10-28 | A kind of more scene management method and devices of voice dialogue |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410590076.4A CN104464733B (en) | 2014-10-28 | 2014-10-28 | A kind of more scene management method and devices of voice dialogue |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104464733A CN104464733A (en) | 2015-03-25 |
| CN104464733B true CN104464733B (en) | 2019-09-20 |
Family
ID=52910681
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410590076.4A Active CN104464733B (en) | 2014-10-28 | 2014-10-28 | A kind of more scene management method and devices of voice dialogue |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104464733B (en) |
Families Citing this family (61)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8977255B2 (en) | 2007-04-03 | 2015-03-10 | Apple Inc. | Method and system for operating a multi-function portable electronic device using voice-activation |
| US8676904B2 (en) | 2008-10-02 | 2014-03-18 | Apple Inc. | Electronic devices with voice command and contextual data processing capabilities |
| US10417037B2 (en) | 2012-05-15 | 2019-09-17 | Apple Inc. | Systems and methods for integrating third party services with a digital assistant |
| DE112014000709B4 (en) | 2013-02-07 | 2021-12-30 | Apple Inc. | METHOD AND DEVICE FOR OPERATING A VOICE TRIGGER FOR A DIGITAL ASSISTANT |
| EP3937002A1 (en) | 2013-06-09 | 2022-01-12 | Apple Inc. | Device, method, and graphical user interface for enabling conversation persistence across two or more instances of a digital assistant |
| US9715875B2 (en) | 2014-05-30 | 2017-07-25 | Apple Inc. | Reducing the need for manual start/end-pointing and trigger phrases |
| US10170123B2 (en) | 2014-05-30 | 2019-01-01 | Apple Inc. | Intelligent assistant for home automation |
| US9338493B2 (en) | 2014-06-30 | 2016-05-10 | Apple Inc. | Intelligent automated assistant for TV user interactions |
| US9886953B2 (en) | 2015-03-08 | 2018-02-06 | Apple Inc. | Virtual assistant activation |
| US10460227B2 (en) | 2015-05-15 | 2019-10-29 | Apple Inc. | Virtual assistant in a communication session |
| US20160378747A1 (en) | 2015-06-29 | 2016-12-29 | Apple Inc. | Virtual assistant for media playback |
| US10331312B2 (en) | 2015-09-08 | 2019-06-25 | Apple Inc. | Intelligent automated assistant in a media environment |
| US10740384B2 (en) | 2015-09-08 | 2020-08-11 | Apple Inc. | Intelligent automated assistant for media search and playback |
| US10747498B2 (en) | 2015-09-08 | 2020-08-18 | Apple Inc. | Zero latency digital assistant |
| US11587559B2 (en) | 2015-09-30 | 2023-02-21 | Apple Inc. | Intelligent device identification |
| US10691473B2 (en) | 2015-11-06 | 2020-06-23 | Apple Inc. | Intelligent automated assistant in a messaging environment |
| US11295736B2 (en) | 2016-01-25 | 2022-04-05 | Sony Corporation | Communication system and communication control method |
| CN105913039B (en) * | 2016-04-26 | 2020-08-18 | 北京光年无限科技有限公司 | Interactive processing method and device for dialogue data based on vision and voice |
| CN106023991B (en) * | 2016-05-23 | 2019-12-03 | 丽水学院 | A kind of hand-held voice interaction device and exchange method towards multitask interaction |
| US10586535B2 (en) | 2016-06-10 | 2020-03-10 | Apple Inc. | Intelligent digital assistant in a multi-tasking environment |
| DK201670540A1 (en) | 2016-06-11 | 2018-01-08 | Apple Inc | Application integration with a digital assistant |
| DK179415B1 (en) | 2016-06-11 | 2018-06-14 | Apple Inc | Intelligent device arbitration and control |
| CN106528522A (en) * | 2016-08-26 | 2017-03-22 | 南京威卡尔软件有限公司 | Scenarized semantic comprehension and dialogue generation method and system |
| CN108628908B (en) * | 2017-03-24 | 2021-02-26 | 北京京东尚科信息技术有限公司 | Method, device and electronic equipment for classifying user question-answer boundaries |
| CN107169034B (en) * | 2017-04-19 | 2020-08-04 | 畅捷通信息技术股份有限公司 | Multi-round human-computer interaction method and system |
| CN107292696B (en) * | 2017-04-27 | 2020-12-22 | 深圳车包包智能科技有限公司 | Intelligent automobile shopping guide system and implementation method |
| DK180048B1 (en) | 2017-05-11 | 2020-02-04 | Apple Inc. | MAINTAINING THE DATA PROTECTION OF PERSONAL INFORMATION |
| DK179496B1 (en) | 2017-05-12 | 2019-01-15 | Apple Inc. | USER-SPECIFIC Acoustic Models |
| DK201770428A1 (en) | 2017-05-12 | 2019-02-18 | Apple Inc. | Low-latency intelligent automated assistant |
| DK201770411A1 (en) | 2017-05-15 | 2018-12-20 | Apple Inc. | Multi-modal interfaces |
| US20180336275A1 (en) | 2017-05-16 | 2018-11-22 | Apple Inc. | Intelligent automated assistant for media exploration |
| CN107026940B (en) * | 2017-05-18 | 2018-09-11 | 北京神州泰岳软件股份有限公司 | A kind of method and apparatus of determining session feedback information |
| CN107133349B (en) * | 2017-05-24 | 2018-02-23 | 北京无忧创新科技有限公司 | One kind dialogue robot system |
| CN109086282A (en) * | 2017-06-14 | 2018-12-25 | 杭州方得智能科技有限公司 | A kind of method and system for the more wheels dialogue having multitask driving capability |
| CN107357855B (en) * | 2017-06-29 | 2018-06-08 | 北京神州泰岳软件股份有限公司 | Support the intelligent answer method and device of scene relating |
| CN107437143A (en) * | 2017-07-26 | 2017-12-05 | 携程计算机技术（上海）有限公司 | Tourism ordering method, system, equipment and storage medium based on virtual customer service |
| CN107644641B (en) * | 2017-07-28 | 2021-04-13 | 深圳前海微众银行股份有限公司 | Dialog scene recognition method, terminal and computer-readable storage medium |
| US10928918B2 (en) | 2018-05-07 | 2021-02-23 | Apple Inc. | Raise to speak |
| US11145294B2 (en) | 2018-05-07 | 2021-10-12 | Apple Inc. | Intelligent automated assistant for delivering content from user experiences |
| DK201870355A1 (en) | 2018-06-01 | 2019-12-16 | Apple Inc. | Virtual assistant operation in multi-device environments |
| DK180639B1 (en) | 2018-06-01 | 2021-11-04 | Apple Inc | DISABILITY OF ATTENTION-ATTENTIVE VIRTUAL ASSISTANT |
| DK179822B1 (en) | 2018-06-01 | 2019-07-12 | Apple Inc. | Voice interaction at a primary device to access call functionality of a companion device |
| CN108874967B (en) * | 2018-06-07 | 2023-06-23 | 腾讯科技（深圳）有限公司 | Dialogue state determining method and device, dialogue system, terminal and storage medium |
| CN109086419B (en) * | 2018-08-07 | 2020-11-13 | 广州小鹏汽车科技有限公司 | Social communication method and system based on scene and voice distribution |
| US11462215B2 (en) | 2018-09-28 | 2022-10-04 | Apple Inc. | Multi-modal inputs for voice commands |
| US11348573B2 (en) | 2019-03-18 | 2022-05-31 | Apple Inc. | Multimodality in digital assistant systems |
| CN109754806A (en) * | 2019-03-21 | 2019-05-14 | 问众智能信息科技（北京）有限公司 | A kind of processing method, device and the terminal of more wheel dialogues |
| DK201970509A1 (en) | 2019-05-06 | 2021-01-15 | Apple Inc | Spoken notifications |
| US11307752B2 (en) | 2019-05-06 | 2022-04-19 | Apple Inc. | User configurable task triggers |
| US11140099B2 (en) | 2019-05-21 | 2021-10-05 | Apple Inc. | Providing message response suggestions |
| US11468890B2 (en) | 2019-06-01 | 2022-10-11 | Apple Inc. | Methods and user interfaces for voice-based control of electronic devices |
| CN110837543A (en) * | 2019-10-14 | 2020-02-25 | 深圳和而泰家居在线网络科技有限公司 | Conversation interaction method, device and equipment |
| CN110880324A (en) * | 2019-10-31 | 2020-03-13 | 北京大米科技有限公司 | Voice data processing method and device, storage medium and electronic equipment |
| CN111161739B (en) * | 2019-12-28 | 2023-01-17 | 科大讯飞股份有限公司 | Speech recognition method and related product |
| CN113450786B (en) * | 2020-03-25 | 2024-07-26 | 阿里巴巴集团控股有限公司 | Network model obtaining method, information processing method, device and electronic equipment |
| US11038934B1 (en) | 2020-05-11 | 2021-06-15 | Apple Inc. | Digital assistant hardware abstraction |
| US11061543B1 (en) | 2020-05-11 | 2021-07-13 | Apple Inc. | Providing relevant data items based on context |
| CN113488036A (en) * | 2020-06-10 | 2021-10-08 | 海信集团有限公司 | Multi-round voice interaction method, terminal and server |
| US11490204B2 (en) | 2020-07-20 | 2022-11-01 | Apple Inc. | Multi-device audio adjustment coordination |
| US11438683B2 (en) | 2020-07-21 | 2022-09-06 | Apple Inc. | User identification using headphones |
| CN115083412B (en) * | 2022-08-11 | 2023-01-17 | 科大讯飞股份有限公司 | Voice interaction method and related device, electronic equipment and storage medium |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101067930A (en) * | 2007-06-07 | 2007-11-07 | 深圳先进技术研究院 | Intelligent audio frequency identifying system and identifying method |
| CN102831446A (en) * | 2012-08-20 | 2012-12-19 | 南京邮电大学 | Image appearance based loop closure detecting method in monocular vision SLAM (simultaneous localization and mapping) |
| CN103187058A (en) * | 2011-12-28 | 2013-07-03 | 上海博泰悦臻电子设备制造有限公司 | Speech conversational system in vehicle |
| CN103413549A (en) * | 2013-07-31 | 2013-11-27 | 深圳创维－Rgb电子有限公司 | Voice interaction method and system and interaction terminal |
| CN103456301A (en) * | 2012-05-28 | 2013-12-18 | 中兴通讯股份有限公司 | Ambient sound based scene recognition method and device and mobile terminal |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8041570B2 (en) * | 2005-05-31 | 2011-10-18 | Robert Bosch Corporation | Dialogue management using scripts |
-
2014
- 2014-10-28 CN CN201410590076.4A patent/CN104464733B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101067930A (en) * | 2007-06-07 | 2007-11-07 | 深圳先进技术研究院 | Intelligent audio frequency identifying system and identifying method |
| CN103187058A (en) * | 2011-12-28 | 2013-07-03 | 上海博泰悦臻电子设备制造有限公司 | Speech conversational system in vehicle |
| CN103456301A (en) * | 2012-05-28 | 2013-12-18 | 中兴通讯股份有限公司 | Ambient sound based scene recognition method and device and mobile terminal |
| CN102831446A (en) * | 2012-08-20 | 2012-12-19 | 南京邮电大学 | Image appearance based loop closure detecting method in monocular vision SLAM (simultaneous localization and mapping) |
| CN103413549A (en) * | 2013-07-31 | 2013-11-27 | 深圳创维－Rgb电子有限公司 | Voice interaction method and system and interaction terminal |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104464733A (en) | 2015-03-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104464733B (en) | A kind of more scene management method and devices of voice dialogue | |
| Hancock et al. | On the future of transportation in an era of automated and autonomous vehicles | |
| CN109684478B (en) | Classification model training method, classification device, classification equipment and medium | |
| CN109964271B (en) | Apparatus and method for providing response message to voice input of user | |
| CN111914551B (en) | Natural language processing method, device, electronic equipment and storage medium | |
| CN109918673A (en) | Semantic referee method, device, electronic equipment and computer readable storage medium | |
| CN111145721B (en) | Personalized prompt generation method, device and equipment | |
| CN108415932A (en) | Interactive method and electronic equipment | |
| EP2112621A2 (en) | Apparatus for forming good feeling of robot and method therefor | |
| CN107665706A (en) | Rapid Speech exchange method and system | |
| CN107831903A (en) | The man-machine interaction method and device that more people participate in | |
| CN106020488A (en) | Man-machine interaction method and device for conversation system | |
| CN110083110A (en) | End to end control method and control system based on natural intelligence | |
| CN107944911A (en) | A kind of recommendation method of the commending system based on text analyzing | |
| CN109408710A (en) | Search result optimization method, device, system and storage medium | |
| CN104731874B (en) | A kind of evaluation information generation method and device | |
| US11468247B2 (en) | Artificial intelligence apparatus for learning natural language understanding models | |
| CN112884235B (en) | Travel recommendation method, and training method and device of travel recommendation model | |
| JP2023036574A (en) | Conversational recommendation method, method and device of training model, electronic apparatus, storage medium, and computer program | |
| DE112020003033T5 (en) | Method and apparatus for improving a geolocation database | |
| Shibata et al. | Chat-like conversational system based on selection of reply generating module with reinforcement learning | |
| US20180018403A1 (en) | Method and system for analyzing emotion on basis of position related document, and non-transitory computer-readable recording medium | |
| US20210334461A1 (en) | Artificial intelligence apparatus and method for generating named entity table | |
| KR102039293B1 (en) | Method, apparatus and program for obtaining answer based on artificial intelligence by using clustering | |
| Doğan et al. | Artificial Intelligence (AI) in Tourism |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |