A kind of control system and method can passing through Voice command household electrical appliancesTechnical field
The present invention relates to a kind of appliance control system and method, be specifically related to a kind of can by the System and method for of Voice command household electrical appliances.
Background technology
Current household electrical appliances are towards more and more intelligent direction development.In home wiring control, speech control is one of most important intelligentized means.For speech control household electrical appliances, Current patents has related example, but all there is certain problem:
Be in the utility model patent " a kind of voice interaction device for household electrical appliances " of ZL 2,011 2 0534127.3 in the patent No., relate to a kind of voice interaction device, can control household electrical appliances.It is mutual that this novel speech control for household electrical appliances needs to do two steps, and the first step needs to say one and wakes the voice activated identifying of word up.Second step, after speech recognition starts, user besides go out the fill order of household electrical appliances.Such as, user says " intelligent air condition ", and air-conditioning can be said " owner please tell ", and then user says " opening " again, and air-conditioning is said " good, air-conditioning is opened ", then opens air-conditioning.The shortcoming of this method needs two steps mutual exactly, and reciprocal process is more loaded down with trivial details.
Be in the patent of invention " air-conditioning and control method " of CN 103712311 A at application publication number, relate to a kind of method of Voice command air-conditioning, Voice command can be carried out to air-conditioning.This invention only needs a step mutual to the speech control of air-conditioning, but need to say the voice command containing three keywords: the first keyword can activate speech recognition simultaneously, second keyword is airconditioning control order, and the 3rd keyword can stop speech recognition.Such as, user says " prepare air-conditioning open air-conditioner standby air-conditioning ", and voice activated identification afterwards that air-conditioning recognizes " preparation air-conditioning ", recognizes the opening action that " opening air-conditioning " performs air-conditioning afterwards, recognizes " until air conditioner " and afterwards stops current identification.This System and method for some extent solves two step interaction problems in utility model patent " a kind of voice interaction device for household electrical appliances ", but still there is the oversize problem of order word.Start a speech recognition, need in an order word, comprise three keywords.
Summary of the invention
The present invention is directed to the deficiencies in the prior art, provide a kind of control system by Voice command household electrical appliances and method, the present invention can help that user realizes carrying out fast household electrical appliances by voice, the control of hommization.The present invention by comprising a voice command words of both keyword, can complete the control to home appliance.Wherein the first half of voice command words is the first key word, for home appliance wakes word up; The latter half of voice command words is the second key word, is home appliance operational order word.As " intelligent air condition start " order word, conditioning control unit can wake up by the first key word " intelligent air condition "; Second key word " start " instruction conditioning control unit, completes boot action by conditioning control unit.The present invention can help that user realizes carrying out fast household electrical appliances by voice, the control of hommization.
The technical solution used in the present invention is: a kind of control system that can pass through Voice command household electrical appliances, and this control system comprises main control module, voice input device, audio output device and electronic appliance control unit; The output terminal of described voice input device is connected with the input end of main control module; The input end of described audio output device is connected with the output terminal of main control module; The communication interface of described electronic appliance control unit and main control module is bi-directionally connected.
Further, described voice input device comprises microphone, note amplifier and AD converter, the sound that microphone collects, and after note amplifier amplifies, then through AD converter, is converted into digital signal, and outputs to main control module.
Further, described audio output device comprises D/A converter, note amplifier and loudspeaker; The audio frequency of the digital format that main control module exports, through D/A converter, is converted into analog format, and then after note amplifier amplifies, is broadcasted by loudspeaker.
Further, described electronic appliance control unit is common household electrical appliances main control unit, can control household electrical appliances, can carry out communication with main control module by communication interface.
Further, described main control module can adopt general microprocessor chip, needs to comprise voice data buffer cell, voice wake up procedure unit, speech recognition program unit, audio playing unit and communication interface.
Further, the voice data buffer cell described in described main control module can audio reception input media DAB input, the voice data of digital format is kept in, and quick position can be carried out to voice data.
Further, if voice wake up procedure unit and speech recognition program unit run noise-filtering algorithm, by the filtering of non-human voice, thus non-human spoken sounds is got rid of to the interference of whole system.
Further, the voice wake up procedure unit described in described main control module runs the voice wake up procedure had based on unspecified person, detects the particular keywords of unspecified person from the digital audio-frequency data of input continuously.
Further, speech recognition program unit described in described main control module runs the speech recognition program had based on unspecified person, the target recognition command word of preset some, can process a piece of digital voice data, and judge whether this piece of digital voice data meets one in target recognition command word according to result.
Further, the audio frequency prestored can be outputted to audio playing apparatus and play by the audio playing unit described in described main control module, to point out the state of current system and various operation information etc.
Further, communication interface described in described main control module is general communication interface, various communication interface such as such as I2C, UART etc., communication interface is connected with household electrical appliances main control unit two-phase, namely can send various control command to household electrical appliances main control unit, also can obtain the various information of household electrical appliances main control unit from household electrical appliances main control unit.
Further, during normal use in the most of the time, only have speech input device and voice data buffer cell, voice wake up procedure unit to keep duty, other parts are in dormancy, low-power consumption or off-mode, thus reduce the power consumption of whole system.
Further, the microphone in voice input device can be positioned over outside the housing of household electrical appliances, easily receives the place of user speech.
The present invention provide in addition a kind of can by the control method of Voice command household electrical appliances, utilize above-mentioned a kind of can by the control system of Voice command household electrical appliances, user can complete control to home appliance by a voice command, voice command comprises the first keyword and the second keyword, first keyword for waking home appliance word up, as " xx fan ", " xx air-conditioning ", " xx TV " etc.Second keyword is home appliance operational order, as " opening ", " closedown ", " timing xx minute " etc.First keyword and the second keyword combine, and are a voice command, as " beautiful air-conditioning is opened ", " form air cooler is shaken the head ", " Haier's TV set timing 30 minutes " etc.
Further, if user wants by speech control air-conditioning, only need to say to control voice command XXXX-YY, wherein XXXX is the first keyword, and YY is the second keyword, such as " intelligent air condition start ".Voice wake up procedure unit is being intercepted and the voice data detected in voice data buffer cell always, detect the first keyword, once there is the first keyword XXXX in voice data, then voice wake up procedure unit wakes speech recognition program unit up, and the end position of XXXX order word in data buffer unit is informed speech recognition program unit, after speech recognition program unit starting, the second keyword voice data is read in voice data buffer cell, start the identification of the second keyword, voice recognition processing is carried out to these data, obtain the recognition result meeting predetermined order YY, speech recognition program unit starting communication interface, and trigger household electrical appliances main control unit by communication interface, order YY is sent to household electrical appliances main control unit, household electrical appliances main control unit carries out various household electrical appliances according to YY order and manipulates, if said process is various, modules and unit are in running order, then do not need the starting operation performing modules and unit.
Further, mute time after speech recognition program unit is terminated by the second keyword audio frequency in detection voice data buffer cell, can the voice data of the second keyword in automatic decision Audio Buffer unit at the end position of voice data, and voice wake up procedure unit is informed in this position, thus inform that voice wake up procedure unit restarts the detection of the first keyword and intercepts process in time, voice wake up procedure unit and speech recognition program unit can not work simultaneously, but speech recognition program unit is once after obtaining recognition result, voice wake up procedure unit just can work, restart the detection of the first keyword and intercept process.
Further, if speech recognition program unit detects that in voice data buffer cell, the first keyword audio frequency is quiet after terminating always, then do not start identifying, after certain hour (6 second), inform that voice wake up procedure unit restarts the detection of the first keyword and intercepts process.
Further, if speech recognition program unit to detect in voice data buffer cell the first keyword audio frequency after terminating always for there being voice pronunciation state, or identify less than the second preset keyword, then after a certain time (6 second) force afterwards terminate identifying, start audio playing unit, playing alert tones, points out the second current keyword of user to detect, and informs that voice wake up procedure unit restarts the detection of the first keyword and intercepts process.
Further, speech recognition program unit can terminate current speech recognition process automatically, does not need the 3rd keyword to go to terminate current identifying.
Principle of the present invention is:
If wish by Voice command household electrical appliances, then must comprise two keywords, first keyword is the name information of home appliance, and the second keyword is the manipulation order of home appliance.These two order words are necessary:
1, for first keyword, the title needing to manipulate home appliance is explicitly pointed out.Speech recognition must stress environment for use, and the grammer environment of household electrical appliances is mainly home environment.In home environment, multiple household electrical appliances can be there are, if so do not use first keyword, only say the second keyword, ambiguity can be brought.As, user only says " start ", and possible fan, air-conditioning, washing machine all can receive this order.
2, for second keyword, the manipulation order of household electrical appliances is explicitly pointed out.This is also necessary, could control to household electrical appliances with clear and definite order to household electrical appliances.
The present invention by comprising a voice command words of both keyword, can complete the control to home appliance.Wherein the first half of voice command words is the first key word, for home appliance wakes word up; The latter half of voice command words is the second key word, is home appliance operational order word.As " intelligent air condition start " order word, conditioning control unit can wake up by the first key word " intelligent air condition "; Second key word " start " instruction conditioning control unit, completes boot action by conditioning control unit.The present invention can help that user realizes carrying out fast household electrical appliances by voice, the control of hommization.
The present invention's advantage is compared with prior art:
(1) by Voice command household electrical appliances, for home wiring control provides one method of operating more easily.
(2) in the patent No. be ZL 2,011 2 0534127.3 utility model patent " a kind of voice interaction device for household electrical appliances " in, relate to a kind of voice interaction device, can control household electrical appliances.It is mutual that this novel speech control for household electrical appliances needs to do two steps, and the first step needs to say one and wakes the voice activated identifying of word up.Second step, after speech recognition starts, user besides go out the fill order of household electrical appliances.Such as, user says " intelligent air condition ", and air-conditioning can be said " owner please tell ", and then user says " opening " again, and air-conditioning is said " good, air-conditioning is opened ", then opens air-conditioning.The shortcoming of this method needs two steps mutual exactly, and reciprocal process is more loaded down with trivial details.The present invention only needs a step alternately, and only need an order word, reciprocal process becomes simple.
(3) at application publication number be CN 103712311 A patent of invention " air-conditioning and control method " in, relate to a kind of method of Voice command air-conditioning, Voice command can be carried out to air-conditioning.This invention only needs a step mutual to the speech control of air-conditioning, but need to say the voice command containing three keywords: the first keyword can activate speech recognition simultaneously, second keyword is airconditioning control order, and the 3rd keyword can stop speech recognition.Such as, user says " prepare air-conditioning open air-conditioner standby air-conditioning ", and voice activated identification afterwards that air-conditioning recognizes " preparation air-conditioning ", recognizes the opening action that " opening air-conditioning " performs air-conditioning afterwards, recognizes " until air conditioner " and afterwards stops current identification.This System and method for some extent solves two step interaction problems in utility model patent " a kind of voice interaction device for household electrical appliances ", but still there is the oversize problem of order word.Start a speech recognition, need in an order word, comprise three keywords.The present invention only needs in an order word, comprise two keywords.
Accompanying drawing explanation
Fig. 1 is that the present invention is a kind of can by the control system block diagram of Voice command household electrical appliances.
Fig. 2 is that the present invention is a kind of can by the control system main control module system chart of Voice command household electrical appliances.
To be that the present invention is a kind of can describe block diagram by the control system of Voice command household electrical appliances and method to Fig. 3.
Embodiment
The present invention is further illustrated below in conjunction with accompanying drawing and specific embodiment.
The present invention is a kind of by the control system of Voice command household electrical appliances, can comprise main control module 1, voice input device 2, audio output device 3, electronic appliance control unit 4; The output terminal of described voice input device 2 is connected with the input end of main control module 1; The input end of described audio output device 3 is connected with the output terminal of main control module 1; Described electronic appliance control unit 4 is bi-directionally connected with the communication interface of main control module 1.
Voice input device 2 comprises microphone, note amplifier and AD converter.The sound that microphone collects, after note amplifier amplifies, then through AD converter, is converted into digital signal, and outputs to main control module 1.Microphone can be positioned over outside the housing of household electrical appliances, easily receives the place of user speech.
Audio output device 3 comprises D/A converter, note amplifier and loudspeaker; The audio frequency of the digital format that main control module 1 exports, through D/A converter, is converted into analog format, and then after note amplifier amplifies, is broadcasted by loudspeaker.
Household electrical appliances main control unit 4 is common household electrical appliances main control unit, can control household electrical appliances, can carry out communication with main control module 1 by communication interface 15.
Main control module 1 can adopt general microprocessor chip.Need to comprise voice data buffer cell 11, voice wake up procedure unit 12, speech recognition program unit 13, audio playing unit 14 and communication interface 15.
As Fig. 2, voice data buffer cell 11 can audio reception input media 2 DAB input, the voice data of digital format is kept in, and quick position can be carried out to voice data.Digital audio-frequency data is ceaselessly input to voice data buffer cell 11 by voice input device 2.Voice data can be outputted to voice wake up procedure unit 12 by voice data buffer cell 11, also voice data can be outputted to speech recognition program unit 13.Voice wake up procedure unit 12 and speech recognition program unit 13 can carry out quick position and reading to the voice data in voice data buffer cell 11, and the load position that real-time update is current.Voice wake up procedure unit 12 and speech recognition program unit 13 can not read the voice data in voice data buffer cell 11 simultaneously, after each program element reads, the position of reading and terminating must be marked, read follow-up voice data to facilitate another program element.
As Fig. 2, voice wake up procedure unit 12 and speech recognition program unit 13 run noise-filtering algorithm, by the filtering of non-human voice, thus gets rid of non-human spoken sounds to the interference of whole system.Voice wake up procedure unit 12 runs the voice wake up procedure had based on unspecified person, detects the particular keywords of unspecified person from the digital audio-frequency data of input continuously.
As Fig. 2, voice wake up procedure unit 12 ceaselessly reads the voice data of voice data buffer cell 11, and detects the keyword of wherein unspecified person, as " intelligent air condition ".From the digital audio-frequency data of input continuously, detect the particular keywords of unspecified person, be current proven technique and algorithm.
As Fig. 2, speech recognition program unit 13 runs the speech recognition program had based on unspecified person, the target recognition command word of preset some.Can process a piece of digital voice data, and judge whether this piece of digital voice data meets one in target recognition command word according to result.Speech recognition program based on unspecified person is current proven technique and algorithm.Suppose current preset 20 order words, this recognizer can process section audio data, the starting position of automatic detection voice, if these section audio data are quiet always, then real speech recognition can not start, once the sound of speech of user be detected, recognizer just can real startup optimization.In identifying, the order that user does not need speech recognition to terminate, as long as user pipes down exceed certain hour (0.5 second), then recognizer then thinks that order word is finished by user, automatically terminates to fetch data from voice data buffer cell 11.According to result, recognizer can judge whether this section of current speech data meets 20 predetermined order words.If met, then export recognition result.If do not met, following three kinds of reasons may be there are:
(1) user does not speak always, is quiet always.Then after certain hour (6 seconds), speech recognition program unit 13 no longer detects the starting position of voice, forces the identifying terminating this, does not export recognition result.
(2) user is speaking always.General predetermined order word all can not be oversize, and need user to say in 6 seconds at certain hour, if do not said within this time, then the order that user says is wrong certainly.If there is the situation that user is speaking always, then force the identifying terminating this, do not export recognition result.
(3) the order word that user says is wrong.Recognizer, according to result, can detect that order word that user says is not one in preset 20 order words, then terminate this identifying, do not export recognition result.
As Fig. 2, the audio frequency prestored can be outputted to audio playing apparatus 3 and play by audio playing unit 14, to point out the state of current system and various operation information etc.Communication interface 15 is general communication interface, communication interface as various in I2C, UART etc.Communication interface 15 is connected with household electrical appliances main control unit 4 two-phase, namely can send various control command to household electrical appliances main control unit 4, also can obtain the various information of household electrical appliances main control unit 4 from household electrical appliances main control unit 4.
During normal use in the most of the time, only have speech input device 1 and voice data buffer cell 11, voice wake up procedure unit 12 to keep duty, other parts are in dormancy, low-power consumption or off-mode.Thus reduce the power consumption of whole system.
As Fig. 3, describe the method applied in the present invention in detail:
User can complete control to home appliance by a voice command, and voice command comprises the first keyword and the second keyword.First keyword for waking home appliance word up, as " xx fan ", " xx air-conditioning ", " xx TV " etc.Second keyword is home appliance operational order, as " opening ", " closedown ", " timing xx minute " etc.First keyword and the second keyword combine, and are a voice command, as " beautiful air-conditioning is opened ", " form air cooler is shaken the head ", " Haier's TV set timing 30 minutes " etc.
Such as user says " intelligent air condition startup " this order word, the digital audio-frequency data that voice input device 2 inputs, is kept in by voice data buffer cell 11.Voice data in the continual detection voice data buffer cell 11 of voice wake up procedure unit 12.If keyword " intelligent air condition " never detected, then voice wake up procedure unit 12 can cycle detection.Once the first keyword " intelligent air condition " be detected, then voice wake up procedure unit 12 needs to arrange the second keyword starting position in voice data buffer cell 13, then starts speech recognition program unit 13.After speech recognition program unit 13 starts, from voice data buffer cell 13, voice data is read in the second keyword starting position, and starts identifying.In identifying, the order that user does not need speech recognition to terminate, as long as user pipes down exceed certain hour (0.5 second), then recognizer then thinks that order word is finished by user, automatically terminates to fetch data from voice data buffer cell 11.In identifying, if recognition result meets an order word " startup " in predetermined order word, then think and obtain recognition result, then speech recognition program unit 13 starts communication interface 15, predetermined order " startup " is sent to communication interface 15, activate and start household electrical appliances main control unit 4, household electrical appliances main control unit 4 is according to predetermined order " startup ", start the operation of whole air-conditioning, speech recognition program unit 13 also needs the starting position arranging the first keyword simultaneously, restarts the listening detection of the first keyword.Be finished after above-mentioned action, speech recognition program unit 13, communication interface 15 and household electrical appliances main control unit 4 reenter dormancy, shutdown or low power consumpting state.
At above-mentioned speech recognition program unit 13 in the process of identification second keyword, if do not obtain predetermined order word, then need to take concrete measure according to concrete reason.
(1) if the voice data after the first keyword is quiet always, then after certain hour (6 seconds), then think that user does not say the second keyword, then current voice command manipulation is false.After speech recognition program unit 13 arranges the starting position of the first keyword, restart the listening detection of the first keyword.Be finished after above-mentioned action, speech recognition program unit 13 reenters dormancy, shutdown or low power consumpting state.
(2) if the audio frequency input after user's first keyword of saying is the state being in voice always.General predetermined order word all can not be oversize, needs user to say in certain hour (6 seconds), if do not said within this time, then the order that user says is wrong certainly.Then force to terminate current identifying, start audio playing unit 14, play be prompted to user accordingly by audio playing apparatus 3, prompting user says the voice command words striven for.Be finished after above-mentioned action, the starting position of first keyword is set, restart the listening detection of first keyword, speech recognition program unit 13, audio playing unit 14, audio playing apparatus 3 reenter dormancy, shutdown or low power consumpting state.
By each above-mentioned process, speech recognition program unit 13 can terminate current speech recognition process automatically, does not need the 3rd keyword to go to terminate current identifying:
(1) in identifying, the order that user does not need speech recognition to terminate, as long as user pipes down exceed certain hour (0.5 second), then recognizer then thinks that order word is finished by user, automatically terminates to fetch data from voice data buffer cell 11.
(2) if user does not speak always, then after certain hour (6 seconds), force to terminate current identification after the first keyword terminates.Because user does not speak, also can not impact Consumer's Experience.
(3) if the first keyword terminates ceaselessly to speak in (6 seconds) in rear certain hour always, then the order word of saying of user is certainly not right, because predetermined order can be designed as in design be no more than this Preset Time (6 seconds).There is this situation, force to terminate current identification.Because user speaks always, also can not impact Consumer's Experience.
Non-elaborated part of the present invention belongs to techniques well known.