Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present invention, the technical solution in the embodiment of the present invention is carried out clear, completeSite preparation describes, it is clear that described embodiment is part of the embodiment of the present invention, instead of all the embodiments.Based on this hairEmbodiment in bright, the every other implementation that those of ordinary skill in the art are obtained without making creative workExample, shall fall within the protection scope of the present invention.
It should be appreciated that ought use in this specification and in the appended claims, term " comprising " and "comprising" instructionDescribed feature, entirety, step, operation, the presence of element and/or component, but one or more of the other feature, whole is not precludedBody, step, operation, element, component and/or its presence or addition gathered.
It is also understood that the term used in this description of the invention is merely for the sake of the mesh for describing specific embodimentAnd be not intended to limit the present invention.As description of the invention and it is used in the attached claims, unless onOther situations are hereafter clearly indicated, otherwise " one " of singulative, "one" and "the" are intended to include plural form.
It will be further appreciated that the term "and/or" used in description of the invention and the appended claims isRefer to any combinations and all possible combinations of one or more of the associated item listed, and including these combinations.
As used in this specification and in the appended claims, term " if " can be according to context quiltBe construed to " when ... " or " once " or " in response to determining " or " in response to detecting ".Similarly, phrase " if it is determined that " or" if detecting [described condition or event] " can be interpreted to mean according to context " once it is determined that " or " in response to trueIt is fixed " or " once detecting [described condition or event] " or " in response to detecting [described condition or event] ".
At present, there are mainly two types of the implementations of speech recognition technology, one kind is to pass through application programming interface(Application Programming Interface, API) realizes speech recognition capabilities；Another kind is to pass through deep linkingDeep Linking technologies realize speech recognition capabilities.However, because API and Deep Linking technologies are required to voice knowledgeOther system and third company's both sides' combined debugging need the developer of both sides to unified api interface and deep linking sdkOr interface has very deep understanding.Meanwhile many details compiling problems are usually had in debugging process, such as the compiling ring of both sidesBorder difference can not compile after may result in API joint debuggings, very high so as to cause the cost of implementation of speech recognition, realize and know with voiceNot relevant ability also can be more difficult.
This programme proposes a kind of data processing method regarding to the issue above, wherein, which can be applied toVoice assistant equipment, the voice assistant equipment can be the intelligent terminals such as mobile phone, tablet computer, computer, smartwatch, also may be usedThink other equipment with speech identifying function.
In the embodiment of the present invention, the voice assistant equipment can be started by obtaining speech recognition triggering command first,The voice assistant equipment can obtain voice messaging input by user, by being analyzed and processed to the voice messaging, identifyEach symbol data in the voice messaging, and semantic dependency analysis is carried out to each symbol data, determine each symbol dataSemanteme, further identify the corresponding operational order of the semanteme, and perform the operational order.
It should be noted that, although the operator of voice assistant equipment can be understood that the meaning of symbol, but the voice helpsHand equipment in itself and does not understand the meaning of symbol, thus the voice assistant equipment need by the symbol data to getting intoRow semantic dependency analyzes the semanteme to determine the symbol data.Such as：People are appreciated and understood by " mobile phone " this signal symbolNumber, but in the true world, it is the strip object that people grasp on hand and people always hold it and against ear, simultaneouslyFace often says sound.Then people are unified refers to this kind of article using " mobile phone ".
In one embodiment, the symbol data includes the text in the voice messaging that the voice assistant equipment is gotWord, sentence and/or vocabulary.Symbol grounding treatment technology may be used in the semantic dependency analysis.Wherein, the symbol groundingRefer to symbol data (such as written word or spoken language) and object, thought or the events of symbol data meaning, instituteState symbol grounding processing refer at least one symbol data that will be got and at least one symbol data meaning object,Thought or events get up as semantic relation, the semantic relevancy of each symbol data are obtained, by each symbolSemantic relevancy between number is adjusted matching, determines the semanteme of each symbol data.Specifically can for example,Assuming that the voice messaging that voice assistant is got is " Zhang San says：Whether you will go or not tomorrow, and Li Si says：I goesTomorrow will have a meeting！",The voice assistant can get each symbol data according to the voice messaging, wherein, it is preferred that emphasis is analysis " I goes " this twoA word, that is, symbol data, there are two types of the semantemes of the two words expression, and one kind is to express certainly i.e. " I will go ", and another kind is expression" not going " according to the semantic relation of symbol data each in the voice messaging, can get the semanteme of expression " I will go "The degree of correlation is 90% and the semantic relevancy of expression " not going " is 95%, the two obtained semantic relevancies are comparedCompared with 95%>90%, the semanteme that may thereby determine that the symbol data " I goes " expression is " not going ".
In one embodiment, the unlatching of the voice assistant equipment can obtain triggering command to open by terminal,For example, terminal can receive wake-up word input by user, wherein, the terminal can be by setting the speech chip of low-power consumptionWake-up word input by user is received, as the terminal can start language by obtaining the wake-up word " your good friend " that user saysSound assistant's equipment.In another example the terminal can be by getting triggering of the user to Home key, earphone, touch screen or gyroscopeIt operates to open the voice assistant equipment.Wherein, the user got can be described in long-press to the operation of Home keyThe operation of Home key or the operation for pressing the Home key according to predetermined manner；The user got is to the behaviour of earphoneWork can be the operation of long-press earphone " play button or Pause key "；The user got can be double to the operation of touch screenHit the operation of screen；The user got can be the wake operation of " special track " to the operation of gyroscope.At oneIn embodiment, which can get the corresponding control signal of the user's operation according to user's operation, and the control is believedNumber low and high level signal is converted into, by, into row decoding, obtaining the corresponding triggering of the user's operation to the low and high level signalInstruction so as to trigger the speech recognition trigger for starting the voice assistant equipment, enables the identification function of the voice assistant equipment.
In one embodiment, which can obtain user after speech recognition triggering command is gotThe voice messaging of input, and being analyzed and processed to the voice messaging, identify word in the voice messaging, vocabulary and/Or sentence, and it is semantic dependency analysis that word, vocabulary and/or sentence to being identified, which carry out symbol grounding processing, is heldOperational order corresponding to the row voice messaging.Wherein, the explanation of the symbol grounding processing is as described above.Specific voiceIdentification process is as shown in Figure 1, Fig. 1 is a kind of flow diagram of voice recognition processing provided in an embodiment of the present invention.Using Fig. 1 asExample illustrates, and the process of the speech recognition includes the following steps：
S101：Voice messaging is obtained, specifically, voice assistant equipment can get voice messaging input by user, exampleSuch as, the voice assistant equipment can get voice messaging input by user (such as " I wants to listen the song of a schoolmate ").
S102：Coded treatment is carried out to the voice messaging and is converted to compressed file, specifically, the voice assistant is setIt is standby to carry out coded treatment to the voice messaging got and be converted into digital compression file, wherein, the number pressureInclude the relevant information of the voice messaging in contracting file.
S103：The compressed file is compared and analyzed, specifically, the voice assistant equipment can will be gotThe digital compression file is sent to local server or cloud server compares and analyzes.
S104：Identify the word and/or vocabulary in voice messaging, specifically, the voice assistant equipment can be according to upperThe comparative analysis to compressed file is stated, identifies word and/or vocabulary in the voice messaging.
S105：The word that identifies and/or vocabulary are transmitted to earthing module, specifically, the voice assistant equipment can be withThe earthing module that the word and/or vocabulary that will identify that are transferred to the voice assistant equipment is analyzed and processed.
S106：Analyze the semanteme of the word and/or vocabulary, the voice assistant equipment can to the word and/orVocabulary is analyzed, and obtains the semanteme of the voice messaging.
S107：The operational order corresponding to the semanteme is performed, the voice assistant equipment can be according to the language identifiedThe semanteme of message breath performs the operational order corresponding to the semanteme.
Below in conjunction with Fig. 2 to Fig. 8 to data processing method provided in an embodiment of the present invention, voice assistant equipment and calculatingMachine readable storage medium storing program for executing illustrates.
Fig. 2 specifically is referred to, Fig. 2 is a kind of network topological diagram of voice assistant system provided in an embodiment of the present invention, such asShown in Fig. 2, which may include：Client 21, server 22, communication network 23 and voice assistant management system 24.Wherein,The client 21 can be the intelligent terminals such as mobile phone, tablet computer, computer, the communication network 23 can be wireless network (such asWi-Fi), cable network (such as 4G networks), voice assistant management system 24 includes voice assistant management module 241, signal connectsGround module 242, personalized user data database 243, voice assistant content data base 244.Wherein, the personalized user numberAccording to database 243 for storing data related to user, the voice assistant content data base 244 helps for storing with voiceThe relevant data of hand content.
In the embodiment of the present invention, client 21 can get the triggering command for opening voice assistant management system 24, lead toIt crosses communication network 23 and the triggering command got is transferred to voice assistant management system 24, so as to open the voice assistant managementSystem 24.Client 21 can obtain voice messaging input by user, and client 21 can will be got by communication network 23User voice messaging be sent to voice assistant management system 24.Voice assistant pipe in the voice assistant management system 24Reason module 241 is used to the voice messaging got is carried out coded treatment, and be converted into the compressed file of the voice messaging.It is describedWhether voice assistant management module 241 can be prejudged can handle the voice messaging, by institute's predicate if it can handleSound assistant management module 241 is handled, and being otherwise sent to server 22 by communication network 23 is handled, so that the server 22 is to thisThe attribute of the voice messaging of user is analyzed and is handled, and identifies content, intonation etc. corresponding to the voice messaging.
In one embodiment, the voice assistant management module 241 can identify includes in the voice messaging of userSymbol data, wherein, the symbol data include word, sentence and/or vocabulary.The voice assistant management module 241 willEach symbol data got is transferred to signal ground module 242 so that the signal ground module 242 to each symbol data intoThe processing of row symbol grounding is semantic dependency analysis, and the semanteme of the determining voice messaging is handled according to the symbol grounding,Identify the corresponding operational order of the semanteme, perform the operational order, and pass through communication network 23 by implementing result export toClient 21.
Fig. 3 is referred to, Fig. 3 is a kind of flow diagram of data processing method provided in an embodiment of the present invention, such as Fig. 3 institutesShow, described method includes following steps for the embodiment of the present invention：
S301：Obtain voice messaging input by user.
In the embodiment of the present invention, voice assistant equipment can obtain voice messaging input by user, specifically, voice assistantEquipment can obtain voice messaging input by user by terminal.For example, voice assistant equipment can obtain user by terminalThe voice messaging " I wants to listen the song of a schoolmate " of input.
In one embodiment, the voice assistant equipment can pass through end before voice messaging input by user is obtainedEnd obtains the triggering command for opening the voice assistant equipment.The triggering command can be that voice input by user wakes up instruction(such as " your good friend ")；The triggering command can also be the operational order input by user to physical button in terminal (such as lengthPress or pressed according to predetermined manner the operational order of Home key)；The triggering command can also be input by user and outside setStandby operational order (such as the operational order of the broadcasting Pause key of pressing earphone)；The triggering command can also be other triggering sidesFormula (such as double-clicks touch screen, gyroscope " special track " wakes up mode).
S302：The voice messaging is analyzed, identifies at least one symbol data.
In the embodiment of the present invention, voice assistant equipment can analyze and process the voice messaging of the user got,Identify at least one of voice messaging symbol data, wherein, the symbol data includes word, sentence and/or wordIt converges.For example, it is assumed that the voice assistant equipment gets voice messaging input by user as " I wants to listen the song of a schoolmate ", it is describedVoice assistant equipment can analyze, and identify described the voice messaging " I wants to listen the song of a schoolmate " gotThe symbol data of voice messaging for " I, think, listen, schoolmate, song ".
S303：Semantic dependency analysis is carried out at least one symbol data, determines the language of at least one symbol dataJustice.
In the embodiment of the present invention, voice assistant equipment can carry out semantic dependency point at least one symbol dataAnalysis determines the semanteme of at least one symbol data.Specifically, the voice assistant equipment can be at least one symbol numberAccording to symbol grounding processing is carried out, the semanteme of at least one symbol data is determined.Wherein, the solution of the symbol grounding processingIt releases as described above, details are not described herein again.
In one embodiment, the voice assistant equipment can divide at least one symbol data identifiedClass calls preset neural network algorithm, and deep learning is carried out to sorted each symbol data, obtains between each symbol dataSemantic relevancy (semantic relevancy between i.e. each word and/or vocabulary).The voice assistant equipment is described to gettingEach symbol data of semantic relevancy carries out semantic dependency analysis, identifies the semanteme of at least one symbol data.ExampleSuch as, it is assumed that the symbol data that the voice assistant equipment identifies for " I, think, listen, schoolmate, song ".By to each symbolNumber is classified, and neural network algorithm is called to carry out deep learning to each symbol data, so as to obtain each symbol dataSemantic relevancy, then symbol grounding processing is carried out to each symbol data, identifies the semanteme of each symbol data, wherein, it is describedThe explanation of symbol grounding processing is as described above.As it can be seen that the embodiment of the present invention is called preset by classifying to symbol dataNeural network algorithm carries out deep learning to symbol data, can improve the ability of deep learning, and pass through deep learning and identifySemantic relevancy between each symbol data identifies the semanteme of each symbol data.
In one embodiment, the voice assistant equipment is calling neural network algorithm to carry out depth to each symbol dataDuring study, the voice assistant equipment can be related by the semanteme of the symbol data got during deep learningDegree is added in the voice assistant equipment, and the semanteme for obtaining the voice messaging is analyzed by semantic dependency, so as to the languageSound assistant equipment can identify when getting same voice information again and perform the behaviour corresponding to the semanteme of the voice messagingIt instructs.For example, it is assumed that the voice messaging that the voice assistant equipment is got is " I wants to listen the song of a schoolmate ", getThe voice messaging symbol data be " I, think, listen, schoolmate, song ", then the voice assistant equipment is being calledIt, can will be in deep learning during neural network algorithm carries out deep learning to each symbol data of the voice messagingThe semantic relevancy of the symbol got in the process is added in the voice assistant equipment, and is handled and obtained by symbol groundingThe semanteme of the voice messaging, so that the voice assistant equipment is getting the voice messaging " I wants to listen the song of a schoolmate " againWhen, it can identify and perform the operational order corresponding to the semanteme of the voice messaging.As it can be seen that implementation of the embodiment of the present inventionMode, by adding the semantic relevancy of symbol data, so that the voice assistant equipment is getting same voice information againWhen, it can rapidly identify and perform the operational order corresponding to the semanteme of the voice messaging, so as to improve the energy of deep learningPower, and then the attraction of user can be improved.
S304：Go out the corresponding operational order of semanteme of each symbol data according to the semantics recognition, and perform identify shouldOperational order.
In the embodiment of the present invention, voice assistant equipment can go out each symbol data according to the semantics recognition gotSemantic corresponding operational order, and perform the operational order identified.For example, it is assumed that the voice assistant equipment is gotSymbol data for " I, think, listen, schoolmate, song ", then the voice assistant equipment can be described each according to what is gotThe semanteme of symbol data, it is " song for playing a schoolmate " to identify the corresponding operational order of the semanteme, so as to perform the behaviourIt instructs.
In the embodiment of the present invention, voice assistant equipment by obtaining voice messaging input by user, to the voice messaging intoRow analyzing and processing, can identify at least one symbol data, and semantic dependency point is carried out at least one symbol dataAnalysis, determines the semanteme of at least one symbol data, and identify the operational order corresponding to the semanteme of each symbol data, withAnd the operational order identified is performed, realize the identification to the voice messaging, and perform the operation corresponding to the voice messaging,So as to improve the practicability of speech recognition and interest, people are met to the intelligence of speech recognition and the need of interestIt asks.
Fig. 4 is referred to, Fig. 4 is the flow diagram of another data processing method provided in an embodiment of the present invention.Such as Fig. 4Shown, described method includes following steps for the embodiment of the present invention：
S401：Obtain voice messaging input by user.
In the embodiment of the present invention, voice assistant equipment can obtain voice messaging input by user, specifically, the voiceAssistant's equipment can obtain voice messaging input by user by terminal.For example, voice assistant equipment can be obtained by terminalVoice messaging " I wants to listen the song of a schoolmate " input by user.
S402：The voice messaging is analyzed, identifies at least one symbol data.
In the embodiment of the present invention, voice assistant equipment can analyze and process the voice messaging got, knowDo not go out at least one of voice messaging symbol data, wherein, the symbol data includes word sentence, and/or vocabulary.ExampleSuch as, it is assumed that the voice assistant equipment gets voice messaging input by user as " I wants to listen the song of a schoolmate ", and the voice helpsHand equipment can analyze the voice messaging " I wants to listen the song of a schoolmate " got, identify the voice messagingSymbol data be " I, think, listen, schoolmate, song ".
S403：Classify at least one symbol data.
In the embodiment of the present invention, voice assistant equipment can at least one symbol data of the voice messaging identified intoRow classification.
In one embodiment, which can close domain of at least one symbol data in voice messagingSystem is analyzed and processed, wherein, the domain relationship includes any one or more in hierarchical relationship, spatial relationship, time relationship.The voice assistant equipment can divide each symbol data after the progress domain Automated generalization according to default ruleClass.
In one embodiment, the voice assistant equipment is after according to default rule to carrying out the domain Automated generalizationIt, can be according to preset static statistics structure, after carrying out the domain Automated generalization during each symbol data is classifiedSymbol data carry out semantic analysis with statistics, according to semantic analysis with count as a result, classifying to the symbol data.
In one embodiment, the voice assistant equipment is after according to default rule to carrying out the domain Automated generalizationDuring each symbol data is classified, if detecting multiple symbol datas with identical semantic relation, the voiceAssistant's equipment can obtain the object information of each symbol data, and according to the object information, to the progress domain Automated generalizationEach symbol data afterwards is classified.
In one embodiment, the voice assistant equipment is after according to default rule to carrying out the domain Automated generalizationIt, can be according to the environmental characteristic of each symbol data application, to the carry out domain during each symbol data is classifiedEach symbol data after Automated generalization is classified；Alternatively, according to the technical characteristic that each symbol data uses, to it is described intoEach symbol data after the Automated generalization of row domain is classified.
S404：Mobilism processing is carried out to sorted each symbol data, obtains the semantic phase between each symbol dataGuan Du.
In the embodiment of the present invention, which can carry out at mobilism sorted each symbol dataReason, obtains the semantic relevancy between each symbol data.
In one embodiment, which is carrying out mobilism processing to sorted each symbol dataDuring, preset neural network algorithm can be called, dynamically adjusts the parameter weights in the neural network algorithm, rootAccording to the adjustment of the parameter weights, deep learning is carried out to sorted each symbol data, is obtained between each symbol dataSemantic relevancy.
In one embodiment, which can call symbolic network deep neural network as shown in Figure 5Algorithm, wherein, Fig. 5 is a kind of structure diagram of neural network provided in an embodiment of the present invention.The embodiment of the present invention is using such asDeep learning neural network shown in fig. 5 dynamically adjusts the parameter weights in the neural network algorithm.Wherein, the godCloud server can be deployed in through network algorithm.For example, it is assumed that the voice assistant equipment gets two symbol datas, if shouldBetween two symbol datas under some scene, weights 0.6 are 0.4 under some scene, then can be weighed by the two parametersValue carrys out the semantic relevancy between dynamic regulation the two symbol datas.
In one embodiment, which is calling neural network algorithm to carry out depth to each symbol dataDuring habit, which can add the semantic relevancy of the symbol data got during deep learningIt adds in the voice assistant equipment.As it can be seen that the embodiment of the present invention, the semanteme of the voice messaging can be obtained in this way,And it can be identified when the voice assistant equipment gets same voice information again and correctly perform the voice messagingSemanteme corresponding to operational order.
S405：According to the semantic relevancy between each symbol data, the semanteme of at least one symbol data is determined.
In the embodiment of the present invention, voice assistant equipment can according to the semantic relevancy between each symbol data, determine toThe semanteme of a few symbol data.
In one embodiment, which is calling neural network algorithm to carry out depth to each symbol dataDuring habit, which can add the semantic relevancy of the symbol data got during deep learningIt adds in the voice assistant equipment, and the semanteme for obtaining the voice messaging is handled by symbol grounding, so that the voice assistant is setIt is standby to be identified when getting same voice information again and correctly perform the operation corresponding to the semanteme of the voice messaging.It canSee, the embodiment of the embodiment of the present invention, by adding the semantic relevancy of symbol data, which obtains againWhen getting same voice information, it can identify and correctly perform the operation corresponding to the semanteme of the voice messaging, so as to improve depthThe ability of study is spent, and then the attraction of user can be improved.
S406：Go out the corresponding operational order of semanteme of each symbol data according to the semantics recognition, and perform identify shouldOperational order.
In the embodiment of the present invention, the semanteme that voice assistant equipment can go out each symbol data according to the semantics recognition is correspondingOperational order, and perform the operational order identified.For example, it is assumed that each symbol data that the voice assistant equipment is gotFor " I, think, listen, schoolmate, song ", then the voice assistant equipment can be according to each symbol data gotSemantics recognition goes out the corresponding operational order of the semanteme for " song for playing a schoolmate ", and performs the operational order.In a realityIt applies in example, if voice assistant equipment detects that the semanteme of the symbol data identified is not deposited in the voice assistant equipmentIt is added in the voice assistant equipment in the semanteme of, the then symbol data that can be will identify that.
In one embodiment, the voice assistant equipment gone out by semantic dependency analysis and identification get voice letterAfter the semanteme of breath, the central processing unit (Central Processing Unit, CPU) of the voice assistant equipment can be passed throughOr microcontroller (Micro Controller Unit, MCU) calls miscellaneous function, performs the corresponding behaviour of semanteme of the voice messagingMake, and export to terminal.The specific implementation process of the miscellaneous function is as shown in fig. 6, Fig. 6 is one kind provided in an embodiment of the present inventionThe interface schematic diagram of accessibility aid operation process.As shown in fig. 6, the miscellaneous function of the terminal can be Accessibility, it shouldAccessibility miscellaneous functions can monitor the focus of the terminal, window variation, button click etc..Assuming that the voice assistantAfter equipment identifies voice messaging " I wants to listen the song of a schoolmate ", it can be called by the Accessibility miscellaneous functionsMusic is clicked at interface as shown in Figure 6, searches for and play the song of a schoolmate.
In the embodiment of the present invention, voice assistant equipment can believe the voice by obtaining voice messaging input by userBreath is analyzed and processed, and identifies at least one symbol data, and semantic dependency analysis is carried out at least one symbol data,The semanteme of at least one symbol data, and the corresponding operational order of semanteme by identifying each symbol data are obtained, is performedThe operational order identified to realize the identification to the voice messaging, and performs the operation corresponding to the voice messaging, so as toThe practicability and interest of speech recognition are improved, meets people to the intelligence of speech recognition and the demand of interest.
The embodiment of the present invention additionally provides a kind of voice assistant equipment, and the voice assistant equipment is any one of aforementioned for performingThe unit of the method.Specifically, referring to Fig. 7, Fig. 7 is a kind of signal of voice assistant equipment provided in an embodiment of the present inventionBlock diagram.The voice assistant equipment of the present embodiment includes：Acquiring unit 701, analytic unit 702, processing unit 703 and identification are singleMember 704.
Acquiring unit 701, for obtaining voice messaging input by user；
Analytic unit 702 for analyzing the voice messaging, identifies at least one symbol data, the symbolNumber includes word and/or vocabulary；
Processing unit 703, for carrying out semantic dependency analysis at least one symbol data, determine it is described at leastThe semanteme of one symbol data；
Recognition unit 704, for going out the corresponding behaviour of semanteme of at least one symbol data according to the semantics recognitionIt instructs, and performs the operational order identified.
Further, the processing unit 703, for classifying at least one symbol data；To described pointEach symbol data after class carries out mobilism processing, obtains the semantic relevancy between each symbol data；According to each symbolSemantic relevancy between data determines the semanteme of at least one symbol data.
Further, the processing unit 703, for being closed to domain of at least one symbol data in voice messagingSystem is analyzed and processed, and the domain relationship includes any one or more in hierarchical relationship, spatial relationship, time relationship；RootAccording to default rule, classify to each symbol data after the progress domain Automated generalization.
Further, the processing unit 703, for according to preset static statistics structure, to carrying out the domain relationshipTreated, and symbol data carries out semantic analysis and statistics；According to semantic analysis and statistical result, the symbol data is carried outClassification.
Further, the processing unit 703, if being additionally operable to detect multiple symbolic numbers with identical semantic relationAccording to obtaining the object information of each symbol data；According to the object information, to each symbolic number after the progress domain Automated generalizationAccording to classifying.
Further, the processing unit 703 is additionally operable to the environmental characteristic applied according to each symbol data, to instituteEach symbol data after carrying out domain Automated generalization is stated to classify；Alternatively, according to the technical characteristic that each symbol data uses,Classify to each symbol data after the progress domain Automated generalization.
Further, the processing unit 703 for calling preset neural network algorithm, dynamically adjusts the godThrough the parameter weights in network algorithm；According to the adjustment of the parameter weights, sorted each symbol data is carried out deepDegree study, obtains the semantic relevancy between each symbol data.
In the embodiment of the present invention, voice assistant equipment obtains voice messaging input by user by acquiring unit 701, passes throughAnalytic unit 702 analyzes the voice messaging, identifies at least one symbol data, by processing unit 703 to this extremelyA few symbol data carries out semantic dependency analysis, determines the semanteme of at least one symbol data, and pass through recognition unit704 identify the corresponding operational order of semanteme of each symbol data and perform the operational order identified, realize to the languageThe identification of message breath, and the operation corresponding to the voice messaging is performed, so as to improve the practicability of speech recognition and entertainingProperty, people are met to the intelligence of speech recognition and the demand of interest.
Fig. 8 is referred to, Fig. 8 is the schematic block diagram of another voice assistant equipment provided in an embodiment of the present invention.Such as Fig. 8 institutesVoice assistant equipment in the embodiment of the present invention shown can include：One or more processors 801；One or more input is setStandby 802, one or more output equipments 803 and memory 804.Above-mentioned processor 801, input equipment 802, output equipment 803It is connected with memory 804 by bus 805.For memory 804 for storing computer program, the computer program includes programInstruction, processor 801 are used to perform the program instruction of the storage of memory 804.Wherein, processor 801 is configured for calling instituteState program instruction execution：
Obtain voice messaging input by user；
The voice messaging is analyzed, identifies at least one symbol data, the symbol data include word and/Or vocabulary；
Semantic dependency analysis is carried out at least one symbol data, determines the language of at least one symbol dataJustice；
Go out the corresponding operational order of semanteme of at least one symbol data according to the semantics recognition, and perform identificationThe operational order gone out.
Further, processor 801 is configured for that described program instruction is called to perform following steps：
Classify at least one symbol data；
Mobilism processing is carried out to sorted each symbol data, the semanteme obtained between each symbol data is relatedDegree；
According to the semantic relevancy between each symbol data, the semanteme of at least one symbol data is determined.
Further, processor 801 is configured for that described program instruction is called to perform following steps：
Domain relationship of at least one symbol data in voice messaging is analyzed and processed, the domain relationship includesAny one or more in hierarchical relationship, spatial relationship, time relationship；
According to default rule, classify to each symbol data after the progress domain Automated generalization.
Further, processor 801 is configured for that described program instruction is called to perform following steps：
According to preset static statistics structure, to carry out the symbol data after the domain Automated generalization carry out semantic analysis withStatistics；
According to semantic analysis and statistical result, classify to the symbol data.
Further, processor 801 is configured for that described program instruction is called to perform following steps：
If detecting multiple symbol datas with identical semantic relation, the object information of each symbol data is obtained；
According to the object information, classify to each symbol data after the progress domain Automated generalization.
Further, processor 801 is configured for that described program instruction is called to perform following steps：
According to the environmental characteristic that each symbol data is applied, to each symbol data after the progress domain Automated generalization intoRow classification；
Alternatively, according to the technical characteristic that each symbol data uses, to each symbol after the progress domain Automated generalizationData are classified.
Further, processor 801 is configured for that described program instruction is called to perform following steps：
Preset neural network algorithm is called, dynamically adjusts the parameter weights in the neural network algorithm；
According to the adjustment of the parameter weights, deep learning is carried out to sorted each symbol data, obtains each symbolSemantic relevancy between number.
In the embodiment of the present invention, voice assistant equipment by obtaining voice messaging input by user, to the voice messaging intoRow analyzing and processing, identifies at least one symbol data, which is handled, it is at least one to obtain thisThe semanteme of symbol data, and the corresponding operational order of semanteme by identifying each symbol data, perform the operation identifiedInstruction to realize the identification to the voice messaging, and performs the operation corresponding to the voice messaging, so as to improve speech recognitionPracticability and interest, meet people to the intelligence of speech recognition and the demand of interest.
It should be appreciated that in embodiments of the present invention, alleged processor 801 can be central processing unit (CentralProcessing Unit, CPU), which can also be other general processors, digital signal processor (DigitalSignal Processor, DSP), application-specific integrated circuit (Application Specific Integrated Circuit,ASIC), ready-made programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logicDevice, discrete gate or transistor logic, discrete hardware components etc..General processor can be microprocessor or this atIt can also be any conventional processor etc. to manage device.
Input equipment 802 can include Trackpad, microphone etc., output equipment 803 can include display (LCD etc.),Loud speaker etc..
The memory 804 can include read-only memory and random access memory, and to processor 801 provide instruction andData.The a part of of memory 804 can also include nonvolatile RAM.For example, memory 804 can also be depositedStore up the information of device type.
In the specific implementation, processor 801, input equipment 802, the output equipment 803 described in the embodiment of the present invention canPerform embodiment described in Fig. 3 of data processing method provided in an embodiment of the present invention or the realization side described in Fig. 4 embodimentsFormula also can perform the realization method of the described voice assistant equipment of the embodiment of the present invention, and details are not described herein.
A kind of computer readable storage medium is additionally provided in the embodiment of the present invention, the computer readable storage medium is depositedComputer program is contained, is realized as described in Fig. 3 or embodiment as described in Figure 4 when the computer program is executed by processorRealization method, also can perform the realization method of described voice assistant equipment of the embodiment of the present invention, details are not described herein.
The computer readable storage medium can be that the inside of the voice assistant equipment described in aforementioned any embodiment is depositedStorage unit, such as the hard disk or memory of voice assistant equipment.The computer readable storage medium can also be that the voice helpsThe plug-in type hard disk being equipped on the External memory equipment of hand equipment, such as the voice assistant equipment, intelligent memory card (SmartMedia Card, SMC), secure digital (Secure Digital, SD) card, flash card (Flash Card) etc..Further,The computer readable storage medium can also both include the internal storage unit of the voice assistant equipment or be deposited including outsideStore up equipment.The computer readable storage medium is used to store needed for the computer program and the voice assistant equipmentOther programs and data.The computer readable storage medium, which can be also used for temporarily storing, have been exported or will exportData.
Those of ordinary skill in the art may realize that each exemplary lists described with reference to the embodiments described hereinMember and algorithm steps can be realized with the combination of electronic hardware, computer software or the two, in order to clearly demonstrate hardwareWith the interchangeability of software, each exemplary composition and step are generally described according to function in the above description.ThisA little functions are performed actually with hardware or software mode, specific application and design constraint depending on technical solution.SpeciallyIndustry technical staff can realize described function to each specific application using distinct methods, but this realization is notIt is considered as beyond the scope of this invention.
It is apparent to those skilled in the art that for convenience of description and succinctly, the language of foregoing descriptionThe specific work process of sound assistant equipment and unit can refer to the corresponding process in preceding method embodiment, no longer superfluous hereinIt states.
In several embodiments provided herein, it should be understood that disclosed voice assistant device and method, it canTo realize by another way.For example, the apparatus embodiments described above are merely exemplary, for example, the unitDivision, only a kind of division of logic function can have other dividing mode, such as multiple units or group in actual implementationPart may be combined or can be integrated into another system or some features can be ignored or does not perform.In addition, it is shown orThe mutual coupling, direct-coupling or communication connection discussed can be the indirect coupling by some interfaces, device or unitIt closes or communication connection or electricity, the connection of mechanical or other forms.
The unit illustrated as separating component may or may not be physically separate, be shown as unitThe component shown may or may not be physical unit, you can be located at a place or can also be distributed to multipleIn network element.Some or all of unit therein can be selected according to the actual needs to realize the embodiment of the present inventionPurpose.
In addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit, it can alsoIt is that each unit is individually physically present or two or more units integrate in a unit.It is above-mentioned integratedThe form that hardware had both may be used in unit is realized, can also be realized in the form of SFU software functional unit.
If the integrated unit is realized in the form of SFU software functional unit and is independent product sale or usesWhen, it can be stored in a computer read/write memory medium.Based on such understanding, technical scheme of the present invention is substantiallyThe part to contribute in other words to the prior art or all or part of the technical solution can be in the form of software productsIt embodies, which is stored in a storage medium, is used including some instructions so that a computerEquipment (can be personal computer, server or the network equipment etc.) performs the complete of each embodiment the method for the present inventionPortion or part steps.And aforementioned storage medium includes：USB flash disk, mobile hard disk, read-only memory (ROM, Read-OnlyMemory), random access memory (RAM, Random Access Memory), magnetic disc or CD etc. are various can store journeyThe medium of sequence code.
The above description is merely a specific embodiment, but protection scope of the present invention is not limited thereto, anyThose familiar with the art in the technical scope disclosed by the present invention, can readily occur in various equivalent modifications or replaceIt changes, these modifications or substitutions should be covered by the protection scope of the present invention.Therefore, protection scope of the present invention should be with rightIt is required that protection domain subject to.