A kind of robot service awakening method and deviceTechnical field
The present invention relates to robotics, specifically, are related to a kind of robot service awakening method and device.
Background technology
With the popularization of smart machine, people are more and more frequent to the usage frequency of smart machine, the power consumption of these equipmentAlso increasing, the standby time that this also results in smart machine is shorter and shorter.
All newborn smart machines, and past smart mobile phone and apparatus such as computer, are all using battery as energyAmount source.The power consumption of these or equipment is less, and its stand-by time is also longer.But run counter to therewith, the performance of equipmentOperating is stronger, and its power consumption is bigger, while the usage frequency of equipment can also cause the power consumption of equipment increasing.Therefore in order toThe use duration of extension device, when user does not use smart machine, the related service in equipment can stop/disabling/suspend, thisSample also just can effectively preserve electricity.
However, when the service in smart machine is in stopping/deactivation/halted state, if user reactivates (called outWake up) service, the method adopted by prior art generally require user be manually entered again enabled instruction come realize service weightOpen, this does not obviously utilize the effective use of the service.
Content of the invention
For solving the above problems, the invention provides a kind of awakening method of robot service, which includes：
Interactive voice information acquiring step, obtains the interactive voice information of user input；
Interactive voice information analysis step, parses to the interactive voice information, obtains servicing wake-up information harmonyLine characteristic information；
Robot services wake-up step, judges whether the service wake-up information indicates that robot service wakes up, if it is,Then determine whether whether the vocal print feature information is mated with default vocal print feature set, if it does, then waking up robotService.
According to one embodiment of present invention, in the interactive voice information analysis step, using local automatic speechIdentification engine and/or high in the clouds automatic speech recognition engine, and local Application on Voiceprint Recognition engine and/or high in the clouds Application on Voiceprint Recognition engine,The interactive voice information is parsed, the service wake-up information and vocal print characteristic information is obtained.
According to one embodiment of present invention, in the interactive voice information analysis step, by handing over to the voiceMutual information is parsed, and obtains the wake-up word of user input；
In robot service wake-up step, judge whether the wake-up word of user input is included in default wake-up word setIn conjunction, if it is, judging that the wake-up word indicates that robot service wakes up.
According to one embodiment of present invention, in the interactive voice information analysis step, by handing over to the voiceMutual information is parsed, and obtains user intent information；
In robot service wake-up step, the user intent information and default wake-up intent data are carried outMatch somebody with somebody, if the match is successful, judge that the intent information indicates that robot service wakes up.
According to one embodiment of present invention,
In the interactive voice information analysis step, by parsing to the interactive voice information, user is obtainedThe wake-up word of input；
In robot service wake-up step, judge whether the wake-up word of user input is included in default wake-up word setIn conjunction, if it is not, then re-executing the Semantic interaction information analysis step to parse the interactive voice information, obtainArrive user intent information；
In robot service wake-up step, the user intent information and default wake-up intent data are carried outMatch somebody with somebody, if the match is successful, judge that the user intent information indicates that robot service wakes up.
According to one embodiment of present invention, the robot service includes any one of item set forth below or or several?：
The maincenter service of robot operating system, robot application, robot realize function and robot interactive interface.
Present invention also offers a kind of Rouser of robot service, which includes：
Interactive voice data obtaining module, its are used for the interactive voice information for obtaining user input；
Interactive voice information analysis module, its are used for parsing the interactive voice information, obtain service and wake up letterBreath and vocal print characteristic information；
Robot services wake module, and which is used for judging whether the service wake-up information indicates that robot service wakes up,If it is, determine whether whether the vocal print feature information is mated with default vocal print feature set, if it does, then wake upRobot is serviced.
According to one embodiment of present invention, the interactive voice information analysis module is configured to using local automatic speechIdentification engine and/or high in the clouds automatic speech recognition engine, and local Application on Voiceprint Recognition engine and/or high in the clouds Application on Voiceprint Recognition engine,The interactive voice information is parsed, the service wake-up information and vocal print characteristic information is obtained.
According to one embodiment of present invention, the interactive voice information analysis module is configured to pass and the voice is handed overMutual information is parsed, and obtains the wake-up word of user input；
The robot service wake module is configured to judge whether the wake-up word of user input is included in default wake-up wordIn set, if it is, judging that the wake-up word indicates that robot service wakes up.
According to one embodiment of present invention, the interactive voice information analysis module is configured to pass and the voice is handed overMutual information is parsed, and obtains user intent information；
The robot service wake module is configured to carry out the user intent information with default wake-up intent dataCoupling, if the match is successful, judges that the intent information indicates that robot service wakes up.
According to one embodiment of present invention,
The interactive voice information analysis module is configured to pass and the interactive voice information is parsed, and obtains userThe wake-up word of input, and the wake-up word is transmitted to robot service wake module；
The robot service wake module is configured to judge whether the wake-up word of user input is included in default wake-up wordIn set, if not, the interactive voice information analysis module is then parsed to the interactive voice information again, usedFamily intent information, and the user intent information is transmitted to robot service wake module；
The robot service wake module is configured to carry out the user intent information with default wake-up intent dataCoupling, if the match is successful, judges that the user intent information indicates that robot service wakes up.
According to one embodiment of present invention, the robot service includes any one of item set forth below or or several?：
The maincenter service of robot operating system, robot application, robot realize function and robot interactive interface.
Robot service awakening method provided by the present invention wakes up the correlation machine in dead status by voicePeople services, and it is double that this robot service wake-up mode can liberate user in the case that aspect is not operated manually in userHand, meanwhile, the method can also realize the Remote Wake Up of robot service.Awakening method is serviced compared to existing robot,The wakeup process that this method enables to robot service is more efficiently and convenient.
Meanwhile, with people for itself privacy attention and inherently safe realize raising, increasing peopleStart to wish that the smart machine of oneself is not carried out some operations by incoherent people.For the problem, side provided by the present inventionMethod determines the identity information of user using vocal print feature information, so as to ensure that the security of wakeup process.
Other features and advantages of the present invention will be illustrated in the following description, also, partly be become from specificationObtain it is clear that or being understood by implementing the present invention.The purpose of the present invention and other advantages can pass through in specification, rightIn claim and accompanying drawing, specifically noted structure is realizing and obtain.
Description of the drawings
In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existingThere is the accompanying drawing that wants needed for technology description to do simple introduction：
Fig. 1 is the flowchart that robot according to an embodiment of the invention services awakening method；
Fig. 2 is the flowchart that robot in accordance with another embodiment of the present invention services awakening method；
Fig. 3 is the structural representation that robot according to an embodiment of the invention services Rouser.
Specific embodiment
Describe embodiments of the present invention below with reference to drawings and Examples in detail, whereby how the present invention is appliedTechnological means solving technical problem, and reach technique effect realize that process can fully understand and implement according to this.Need to illustrateAs long as not constituting conflict, each embodiment and each feature in each embodiment in the present invention can be combined with each other,The technical scheme for being formed is within protection scope of the present invention.
Meanwhile, in the following description, elaborate many details for illustrative purposes, to provide to of the invention realApply the thorough understanding of example.It will be apparent, however, to one skilled in the art, that the present invention can be without tool hereBody details or described ad hoc fashion are implementing.
In addition, can be in the department of computer science of such as one group of computer executable instructions the step of the flow process of accompanying drawing is illustratedExecute in system, and, although show logical order in flow charts, but in some cases, can be being different from hereinOrder execute shown or described step.
Embodiment one：
Fig. 1 shows the flowchart of the awakening method of the robot service provided by the present embodiment.
As shown in figure 1, the awakening method that the robot provided by the present embodiment is serviced obtains use first in step S101The interactive voice information of family input.Specifically, in the present embodiment, the method is preferably filled by voice collecting in step S101Put (such as microphone) to obtain the interactive voice information of user input.
After obtaining interactive voice information, the method is by step s 102 to accessed interactive voice in step S101Information is parsed, so as to obtain servicing wake-up information.Specifically, in the present embodiment, the method is in step s 102 using thisGround automatic speech recognition engine comes and local Application on Voiceprint Recognition engine, to enter to accessed interactive voice information in step S101Row parsing, so that respectively obtain service wake-up information and vocal print characteristic information.
Application on Voiceprint Recognition belongs to the one kind in biological identification technology, be one according in speech waveform reflection speak human physiology andThe speech parameter of behavioural characteristic, the technology of automatic identification speaker's identity.From unlike speech recognition, Application on Voiceprint Recognition is utilizedIt is the speaker information in voice signal, and does not consider that the words in voice looks like, it emphasizes the individual character of speaker；And voice is knownOther purpose is to identify the speech content in voice signal, it is not intended that whom speaker is, it emphasizes general character.
In the present embodiment, the accessed interactive voice information in step S101 of local Application on Voiceprint Recognition engine is solvedDuring analysis, pitch contour, formant frequency bandwidth and its track in preferably extracting included in voice messaging are used as vocal printCharacteristic information.This kind of being characterized in that based on the physiological structure that organ (such as glottis, sound channel and nasal cavity) occurs and the parameter extracted.
It is pointed out that in other embodiments of the invention, the local Application on Voiceprint Recognition engine used by the method is alsoOther rational methods can be adopted come corresponding vocal print feature in extracting from interactive voice information, the invention is not restricted to this.
For example, in one embodiment of the invention, the local Application on Voiceprint Recognition engine used by the method can be with from languageSpectrum envelop parameter is extracted in sound interactive information is used as vocal print feature information；In one embodiment of the invention, the method instituteUsing local Application on Voiceprint Recognition engine can also extract corresponding vocal print from interactive voice information using linear predictor coefficientCharacteristic information, specifically, as linear prediction is consistent with channel parameters model, therefore the Application on Voiceprint Recognition engine can be withAll kinds of parameters (such as linear predictor coefficient, auto-correlation coefficient, emission ratio, log-area, linear prediction that linear prediction is derivedThe parameters such as residual error and combinations thereof) it is used as Application on Voiceprint Recognition feature；In one embodiment of the invention, the sheet used by the methodGround Application on Voiceprint Recognition engine can be extracting the parameter that can reflect auditory properties (for example from accessed interactive voice informationU.S. cepstrum coefficient perceives linear forecasting parameter etc.) it is used as vocal print feature information.
It is also desirable to, it is noted that in other embodiments of the invention, the method can both be adopted in step s 102With high in the clouds automatic speech recognition engine and high in the clouds Application on Voiceprint Recognition engine come to accessed interactive voice information in step S101Parsed and correspondingly obtained servicing wake-up information and vocal print characteristic information, it would however also be possible to employ automatic speech recognition listed above is drawnHold up a certain or two kinds are obtained servicing wake-up information to carry out parsing to above-mentioned interactive voice information, and using listed aboveA certain or two kinds the combination of Application on Voiceprint Recognition engine obtains vocal print feature information to carry out parsing to above-mentioned voice messaging, thisInvention is similarly not so limited to.
As shown in figure 1, in the present embodiment, after obtaining servicing wake-up information and vocal print characteristic information, the method can be in stepService obtained by judging in step S102 in rapid S103 wakes up whether information indicates that robot service wakes up.
Interactive voice information analysis are obtained by automatic speech recognition engine due to servicing wake-up information, therefore the clothesContain, in business wake-up information, the relevant information for characterizing whether user expects to carry out robot service wake-up, the method is according to the letterWhether also it may determine that servicing wake-up information, whether Mailbot service wakes up breath, i.e., currently need to carry out robot serviceWake up.
If the method in step s 103 in determination step S102 obtained by service wake up information and indicate robot clothesBusiness wakes up, then party's rule can determine whether in step S104 in step S102 obtained by vocal print feature information with pre-If whether vocal print feature set mates.If the method in step s 103 in determination step S102 obtained by service wake up letterBreath does not indicate that robot service wakes up, then then represent that active user is not intended to wake up robot service, therefore the methodThe current state of robot service will be kept in step s 106, that is, continue to disable/suspend/stop the robot service.
In step S104, the method can be by obtained vocal print feature information in step S102 and default vocal print feature collectionConjunction is mated.If the two can mate, (the vocal print feature information obtained by such as step S102 is included in factor harmonyIn line characteristic set), then then represent that active user is the validated user of robot service, therefore the method can be in step S105The service of middle wake-up robot；And if the vocal print feature information obtained by step S102 is mismatched with default vocal print set, thatThen represent that active user is not the validated user of robot service, the user do not possess the authority serviced using robot, becauseThis method also will keep the current state of robot service in step s 106, that is, continue to disable/suspend/stop the machineDevice people services.
It is pointed out that the robot service referred to by the present invention both may refer to the maincenter clothes of robot operating systemBusiness, it is also possible to refer to robot application, can also refer to robot and realizes function and/or robot interactive interface, and the present invention is notIt is limited to this.
It is also desirable to, it is noted that in other embodiments of the invention, the method can also be by interactive voiceInformation is parsed, and determines robot service to be waken up.For example, when being obtained by parsing to interactive voice informationDuring the word for such as " shooting ", the method will judge that robot service to be waken up now is taken for greatly possibly imagingBusiness.
Embodiment two：
Fig. 2 shows the flowchart of the awakening method of this robot service for being provided.
As shown in Fig. 2 the method provided by the present embodiment obtains the interactive voice of user input first in step s 201Information.Wherein, in the present embodiment, step S101 institute in the realizing principle and realize process and above-described embodiment one of step S201The content being related to is identical, therefore here is no longer repeated to step S201.
After interactive voice information is obtained, the method can in step S202 to step S201 in accessed voice hand overMutual information is parsed, so as to obtain the wake-up word and vocal print characteristic information of user input.Specifically, in the present embodiment, the partyMethod preferably in step S202 using local automatic speech recognition engine come to step S201 in collect interactive voice letterBreath is parsed, so as to obtain the wake-up word included in above-mentioned interactive voice information.It is pointed out that the present embodiment is carriedAnd wake-up word refer not only to generation and prestore wake-up word, can also refer to contain the phrase/short sentence for waking up word, being capable of order robotWaken up instructs word, characterizes intention sentence of wake-up etc..
Specifically, in the present embodiment, the method is in step S202 first with local automatic speech recognition engine to languageSound interactive information is identified, and obtains the text message corresponding to interactive voice information, is subsequently carried according to text informationTake out corresponding wake-up word.
In step S202, the method is preferably by local Application on Voiceprint Recognition engine to carry out above-mentioned interactive voice informationParsing, so that obtain the vocal print feature information of user.It is pointed out that the method utilizes local Application on Voiceprint Recognition engine to voiceInteractive information carries out parsing in the implementing principle and realize process and embodiment one of the vocal print feature information for obtaining userContent involved by step S102 is similar to, therefore will not be described here.
It is also desirable to, it is noted that in other embodiments of the invention, the method can be both adopted in step S202With high in the clouds automatic speech recognition engine and high in the clouds Application on Voiceprint Recognition engine come to accessed interactive voice information in step S201Parsed and correspondingly obtained servicing wake-up information and vocal print characteristic information, it would however also be possible to employ automatic speech recognition listed above is drawnHold up a certain or two kinds are obtained servicing wake-up information to carry out parsing to above-mentioned interactive voice information, and using listed aboveA certain or two kinds the combination of Application on Voiceprint Recognition engine obtains vocal print feature information to carry out parsing to above-mentioned voice messaging, thisInvention is similarly not so limited to.
After the wake-up word for obtaining user input, the method will judge obtained in step S202 calling out in step S203Whether awake word is included in default wake-up in set of words.Specifically, preset to wake up to contain in set of words and multiple can characterize userExpect to carry out word or sentence (for example, " open ", " startup " or " restarting ") of robot service wake-up etc..It may be noted thatBe, in different embodiments of the invention, above-mentioned default wake-up set of words both can Shi You associated productions manufacturer in advance of dispatching from the factoryRow sets, it is also possible to is set during daily interaction by user, the invention is not restricted to this.
In the present embodiment, the method judges obtained wake-up word in step S202 and default wake-up word in step S203Whether set when mating, wake up word and default wake-up set of words the match is successful can only not be default wake up include in set of words withThe identical word of obtained wake-up word in step S202, can also be that default wake-up in set of words includes and step S202In obtained by wake up word in semantically same or like word.This mode can effectively improve the relevance grade of this method,So that the wakeup process more hommization serviced by robot and personalization.
Parsing is carried out by the interactive voice information to user input for example and has obtained wake-up word " waking up ", and if defaultWake up, but contain the word that such as " does not sleep ", then due to the semantic phase of the twoWith therefore with this, the method also just can judge that the wake-up word being capable of successful match with default wake-up set of words.
If the wake-up word obtained by step S202 is included in default wake-up in set of words, then then represent active userWish to wake up correlation machine people service, therefore now the method by execution step S206 determining whether gained in step S202To vocal print feature information whether mate with default vocal print feature set；And if the wake-up word obtained by step S202 is not wrappedIt is contained in default wake-up in set of words, then party's rule can re-use local automatic speech recognition engine in step S204Interactive voice information obtained by step S201 is parsed, and obtains user intent information.
During daily interaction, it is likely in interactive voice information and not comprising related wake-up word, but the interaction languageMessage breath still represents that user expects to carry out robot service wake-up.For example, user is input into interactive voice information " my godBright, this is got up ", the method can be obtained by parsing to the interactive voice information, still in step S204 currentlyUser intent information be " wake up robot service ".
After user intent information is obtained, as shown in Fig. 2 the method provided by the present embodiment will will be walked in step S205User intent information obtained by rapid S204 is mated with default wake-up intent data, and judges whether the two can mateSuccess.
If the user intent information obtained by step S204 is with default wake-up, and intent data can the match is successful, thenParty's rule is special with the vocal print feature information obtained by determining whether in step S202 and default vocal print by execution step S206Whether collection conjunction mates；And if the user intent information obtained by step S204 cannot be mated with default wake-up intent dataSuccess, then then represent user be not intended to carry out robot service wake-up, therefore now the method by execution step S208 withKeep the current state of robot service, that is, continue to disable/suspend/stop the robot service.
In step S206, whether the vocal print feature information obtained by the method will determine that is mated with default vocal print set.If if the two can mate, (the vocal print feature information obtained by such as step S202 is included in factor and vocal print feature setIn conjunction), then then represent that active user is the validated user of robot service, therefore the method can wake up machine in step S207Device people services；And if the vocal print feature information obtained by step S202 is mismatched with default vocal print set, then then representActive user is not the validated user of robot service, and the user do not possess the authority serviced using robot, therefore the methodAlso the current state of robot service will be kept in step S208, that is, continue to disable/suspend/stop robot clothesBusiness.
It is pointed out that in other embodiments of the invention, the method can with according only to wake up word or according only toUser intent information judges whether user is desired with robot judging currently whether need to carry out robot service wake-upService wakes up, and the invention is not restricted to this.
It is also desirable to, it is noted that the robot service referred to by the present invention both may refer to robot operating systemMaincenter service, it is also possible to refer to robot application, can also refer to robot and realize function and/or robot interactive interface,The present invention is similarly not so limited to.
With popularization of the smart machine in people's life, occur in that a lot of users are inconvenient to be manually entered instruction to wake upRobot service and need Remote Wake Up robot service scene.For example, when user needs to operate vapour in startup procedureDuring car navigation feature, it is extremely dangerous that in vehicle traveling process, both hands departure direction disk removes manual Electrode Operation screen again.AndAnd, in some cases, smart machine is arranged and is not equipped with the input hardwares such as screen, keyboard or mouse, in the case, is passedThe mode of operation of system obviously cannot be implemented.
Robot service awakening method provided by the present invention wakes up the correlation machine in dead status by voicePeople services, and it is double that this robot service wake-up mode can liberate user in the case that aspect is not operated manually in userHand, meanwhile, the method can also realize the Remote Wake Up of robot service.Awakening method is serviced compared to existing robot,The wakeup process that this method enables to robot service is more efficiently and convenient.
Meanwhile, with people for itself privacy attention and inherently safe realize raising, increasing peopleStart to wish that the smart machine of oneself is not carried out some operations by incoherent people.Example is directed to the problem, provided by the present inventionMethod determines the identity information of user using vocal print feature information, so as to ensure that the security of wakeup process.
The present embodiment additionally provides a kind of robot service Rouser, and Fig. 3 outputs the structural representation of the device.
As shown in figure 3, the robot service Rouser provided by the present embodiment is preferably included：Interactive voice information is obtainedDelivery block 301, interactive voice information analysis module 302 and robot service wake module 303.Wherein, interactive voice information301 user of acquisition module obtains the interactive voice information of user input.Specifically, in the present embodiment, interactive voice acquisition of informationModule 301 preferably can be realized using voice acquisition device (such as microphone).
After the interactive voice information for obtaining user input, interactive voice data obtaining module 301 can be by the interactive voiceInformation transfer to interactive voice information analysis module 302, to be entered to the interactive information by interactive voice information analysis module 302Row parsing is so as to obtaining servicing wake-up information and vocal print characteristic information.
Interactive voice information analysis module 302 includes automatic speech recognition engine and Application on Voiceprint Recognition engine.Wherein, automatic languageSound identification engine can be parsed to above-mentioned interactive voice information and be obtained servicing wake-up information, and Application on Voiceprint Recognition engine then canReach the vocal print feature information that user is parsed and is obtained to above-mentioned interactive voice information.
It is pointed out that in different embodiments of the invention, above-mentioned automatic speech recognition engine can both adopt thisGround automatic speech recognition engine is realizing, it would however also be possible to employ automatic speech recognition engine in high in the clouds can also be adopted local realizingAutomatic speech recognition engine and high in the clouds automatic speech recognition engine realized jointly, the invention is not restricted to this.Meanwhile, the present invention'sIn different embodiments, above-mentioned Application on Voiceprint Recognition engine both can be realized using local Application on Voiceprint Recognition engine, it would however also be possible to employ high in the cloudsApplication on Voiceprint Recognition engine can also be realized using local Application on Voiceprint Recognition engine and high in the clouds Application on Voiceprint Recognition engine jointly realizing, thisBright it is similarly not so limited to.
In the present embodiment, interactive voice information analysis module 302 after obtaining servicing wake-up information and vocal print characteristic information,The service can be waken up information and vocal print characteristic information is transmitted to robot and services wake module 303.Robot service wakes up mouldBlock 303 can judge whether above-mentioned service wakes up information after above-mentioned service wake-up information and vocal print characteristic information is received firstIndicate that robot service wakes up, if it is, determining whether whether are above-mentioned vocal print feature information and default vocal print feature setCoupling.If above-mentioned vocal print feature information and default vocal print feature sets match, then robot service wake module 303 is then called outAwake correlation machine people service, otherwise then keeps the current state of robot service.
Wherein, in the present embodiment, the tool of interactive voice information analysis module 302 and robot service wake module 303Body realizes the content class that realizes in process and above-described embodiment one involved by step S102 to step S106 of principle and its functionSeemingly, thus here no longer to interactive voice information analysis module 302 and robot service wake module 303 repeat.
It is pointed out that in other embodiments of the invention, interactive voice information analysis module 302 can also be adoptedContent as involved by step S202 to step S208 in Fig. 2 by being taken up in order of priority judgement wake-up word with set of words is waken up it isWhether no coupling and user intent information and default wake-up intent data mate to determine service wake-up information whether knowledge machineDevice people service wakes up, or is only to pass through to judge to wake up word and wake up whether set of words is mated or only judged and wake-up set of wordsCoupling is determining whether Mailbot service wakes up service wake-up information, and the present invention is similarly not so limited to.
It should be understood that disclosed embodiment of this invention is not limited to ad hoc structure disclosed herein or processes stepSuddenly, the equivalent substitute of these features that those of ordinary skill in the related art are understood should be extended to.It should also be understood thatIt is that term as used herein is only used for the purpose for describing specific embodiment, and is not intended to limit.
" one embodiment " or " embodiment " that mentions in specification means special characteristic, the structure for describing in conjunction with the embodimentsOr characteristic is included at least one embodiment of the present invention.Therefore, the phrase " reality that specification various places throughout occursApply example " or " embodiment " same embodiment might not be referred both to.
Although above-mentioned example is used for principle of the present invention in one or more applications to be described, for the technology of this areaFor personnel, in the case of the principle and thought without departing substantially from the present invention, hence it is evident that can in form, the details of usage and enforcementUpper various modifications may be made and without paying creative work.Therefore, the present invention is defined by the appended claims.