CN106023995A - Voice recognition method and wearable voice control device using the method - Google Patents
Voice recognition method and wearable voice control device using the method Download PDFInfo
- Publication number
- CN106023995A CN106023995A CN201610535986.1A CN201610535986A CN106023995A CN 106023995 A CN106023995 A CN 106023995A CN 201610535986 A CN201610535986 A CN 201610535986A CN 106023995 A CN106023995 A CN 106023995A
- Authority
- CN
- China
- Prior art keywords
- model
- voice
- infrared
- language
- order
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 35
- 230000006870 function Effects 0.000 claims abstract description 7
- 238000004891 communication Methods 0.000 claims description 13
- 238000012952 Resampling Methods 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000003909 pattern recognition Methods 0.000 claims description 6
- 238000004458 analytical method Methods 0.000 claims description 4
- 230000033228 biological regulation Effects 0.000 claims description 3
- 238000003066 decision tree Methods 0.000 claims description 3
- 230000007935 neutral effect Effects 0.000 claims description 3
- 239000002245 particle Substances 0.000 claims description 3
- 230000007613 environmental effect Effects 0.000 abstract description 4
- 238000004378 air conditioning Methods 0.000 description 14
- 238000005516 engineering process Methods 0.000 description 13
- 230000005540 biological transmission Effects 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000015572 biosynthetic process Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 230000005611 electricity Effects 0.000 description 2
- 239000000047 product Substances 0.000 description 2
- 230000004044 response Effects 0.000 description 2
- 238000003786 synthesis reaction Methods 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 238000013477 bayesian statistics method Methods 0.000 description 1
- 239000007795 chemical reaction product Substances 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 239000011521 glass Substances 0.000 description 1
- 230000008676 import Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000003825 pressing Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/26—Speech to text systems
-
- G—PHYSICS
- G08—SIGNALLING
- G08C—TRANSMISSION SYSTEMS FOR MEASURED VALUES, CONTROL OR SIMILAR SIGNALS
- G08C23/00—Non-electrical signal transmission systems, e.g. optical systems
- G08C23/04—Non-electrical signal transmission systems, e.g. optical systems using light waves, e.g. infrared
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L15/14—Speech classification or search using statistical models, e.g. Hidden Markov Models [HMMs]
- G10L15/142—Hidden Markov Models [HMMs]
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L15/16—Speech classification or search using artificial neural networks
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Multimedia (AREA)
- Probability & Statistics with Applications (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Selective Calling Equipment (AREA)
Abstract
The present invention provides a voice recognition method and a wearable voice control device using the method wherein the wearable voice control device comprises a household appliance control device, a user operation device and a storage device; the user operation device is provided with a voice recognition system, a voice learning system and an infrared learning system. The voice recognition system is provided with only two basic commands. With the system, it is possible to reduce the erroneous recognition rate, to greatly reduce the interference of environmental noise, and to improve the voice recognition rate by enabling a user to control the home appliances according to the self-defined voice commands conveniently, thus bringing good user experience. Further, the present invention provides a storage device that stores relevant commands and infrared coding commands; compared with the prior art, without network support, control over various functions of home appliances is made possible.
Description
Technical field
The present invention relates to voice field, particularly relate to a kind of Intelligent housing Wearable sound control method and dress
Put.
Background technology
Speech recognition technology, also referred to as automatic speech recognition (Automatic Speech Recognition, ASR),
Its target is to be computer-readable input by the vocabulary Content Transformation in the voice of the mankind, such as button, binary coding or
Person's character string.
The ultimate principle of speech recognition is theoretical frame based on Bayesian statistic modeling, accurate with maximum a posteriori probability decision-making
Then judge acoustic features.Firstly the need of setting up language database and speech database, wherein deposit character library and the word of certain language
Storehouse also associates the acoustic features of standard pronunciation of correspondence.By based on genetic algorithm, it is aided with voice, linguistry, signal
Data base is trained by treatment technology, data mining technology and statistical modeling method.Make data base from general storage number
Acoustic model repository and the language model storehouse possessing certain screening function is become according to storehouse.The acoustics of voice is extracted by equipment such as microphones
After feature, comparing with the model in data base, compared result carries out statistical modeling analysis, according to the spy being previously set
Value indicative obtains optimal result.
Wearable is equipped, and refers to apply wearable technology, daily wearing equipment is carried out intelligentized design, develops permissible
The general name of the equipment dressed.Both can be the equipment independently realizing complete function, it is also possible to be with other equipment with the use of with
Realize the equipment of complete function.
Smart Home, is with house as platform, utilizes comprehensive wiring technology, the network communications technology, Smart Home-system
Design security precautions technology, automatic control technology, audio frequency and video technology, by integrated for facility relevant for life staying idle at home, build efficiently
The management system of housing facilities and family's schedule affairs, promote house security, convenience, comfortableness, artistry, and realize
The living environment of environmental protection and energy saving.
In Smart Home field, Voice command man's power technology the most gradually occupies one seat.Such as 122 Guangzhou Export Commodities Fairs
On (China import and export commodities fair i.e. Guangzhou Trade Fair), Haier's air-conditioning is proposed first " voice remote controller ", and user needs
Voice remote controller facing to hand size is spoken, and can control air-conditioning and carry out the operation such as switching on and shutting down, homoiothermic.Company of Konka is proposed
Voice electric cooker, voice microwave oven, the most all achieve the Voice command of electrical equipment.
But, Wearable field product in terms of voice is less, and price is up to the high-end product paddy of 1500 dollars
Song glasses, can be taken pictures by sound control, video calling and distinguish direction, and surf the web, process Word message and electricity
Sub-mail etc..And the Wearable product in terms of intelligent appliance control, it is still blank out the most so far.
Summary of the invention
In order to overcome the problems referred to above, the invention provides one can imperfect at phonetic order or sequentially make a mistake
Shi Jinhang effectively corrects the audio recognition method with fuzzy matching, additionally provide a kind of can processed offline nonspecific voice
Phonetic order, and then realize the Wearable phonetic controller of the coupling to different household electrical appliances and remote control.
For achieving the above object, present invention provide the technical scheme that
A kind of sound control method, it is characterised in that it comprises the following steps:
1) hidden Markov model and neutral net is used to carry out probability calculation during acoustic features is extracted;
2) resampling is carried out by the particle filter signal to extracting;
3) probability Distribution Model of voice and certain language regulation is set up;
4) to step 2) acoustic features after resampling gives weight, and is weighted;
5) with regard to the probability Distribution Model of comparison of computational results step 3) of step 4), when the result of calculation of step 4) does not meets step
3) probability Distribution Model, shields worn-out step 2) resampling correspondence voice.
Multiplex mode recognition methods is used to identify further for meeting the acoustic features of probability Distribution Model, described multiple
Mode identification method includes including following steps:
1) target sent according to phonetic order is different, devises independent speech model and language model group;
2) when receiving phonetic order, its acoustic features and different types of Model Group are compared, first phonetic order is divided to
In a certain big class model, then carry out pattern recognition with the Model Group within this big class model, until last and least model
Concrete model in Qun compares, and completes the identification of phonetic order.
Use pca method partiting step 2) Model Group size and the foundation of level, described principle component analysis include with
Lower step:
1) several acoustic features is extracted;
2) in multiple features, an identification is found the highest, main as pattern recognition of the feature that model can be distinguished
Foundation, and as pivot;
3) all models are classified, obtain ground floor Model Group;
4) from remaining feature, a topmost feature is found as pivot;
5) all models are reclassified, obtain the Model Group of next level;
6) step 4) and 5 is repeated) until all features are all used complete, and complete the division of multilayered model group.
Speech model described in step 1) includes N-Gram model, decision-tree model, CFG language model
On polyglot model, and perform according to following steps:
1) voice signal obtained is mated with the principal character in all of language model;
2) it is ranked up with the matching degree of all language models, selects five kinds of models that wherein matching degree is the highest by ballot
Mode determine.
The Wearable voice control device of a kind of audio recognition method, it includes domestic electric appliances controller, user operation dress
Put and store device;Wherein, described domestic electric appliances controller is provided with signal emission module and signal receiving module, home wiring control
Device utilizes signal emission module and signal receiving module to be remotely connected with user operation device, storage device;Described storage dress
Being provided with two or more in putting for storing the memory block of different home wiring control order, described signal emission module is provided with red
External transmitter, is provided with speech recognition system, for learning the phonetic study body of voice command in described user operation device
Unite and for learning the infrared learning system of infrared coding;Described language learning system coordinates with infrared learning system, infrared
Learning system is connected by the signal receiving module of domestic electric appliances controller and is connected with storage device;Described speech recognition system is arranged
Having identification engine, this identification engine includes starting order, study order and phonetic storage district, and language learning system is deposited with voice
Storage area and storage device connect.
Described memory block is both provided with some storage communities.
It is additionally provided with self stick circuit, in this self stick circuit, is provided with timing means.
Described control equipment also includes bluetooth communication, main control chip, function key and display module, phonetic synthesis mould
Block, speech recognition circuit, single-chip microcomputer and wireless communication module;Described bluetooth communication includes Bluetooth audio frequency emitter and bluetooth
Audio receiver;Described Bluetooth audio frequency emitter and Bluetooth audio receiver wireless connections;Described Bluetooth audio receiver connects
Speech recognition circuit audio input end;Described speech recognition circuit data-interface is connected with single-chip microcomputer input/output interface.
Having the beneficial effect that of technique scheme
1, the audio recognition method that the present invention provides can or order imperfect at phonetic order be carried out when making a mistake effectively
Correction and fuzzy matching.
2, the invention provides the Wearable voice control device of a kind of audio recognition method, it includes home wiring control dress
Put, user operation device and storage device, this user operation device is provided with speech recognition system, the phonetic study decorum and infrared
Learning system;Speech recognition system only arranges two basic commands, can reduce the error rate that identifies of speech recognition system, permissible
Being greatly reduced the interference of environmental noise, improve phonetic recognization rate, user can also control household electrical appliances according to the voice command made by oneself,
Easy to use, there is more preferable Consumer's Experience.And the present invention is provided with storage device, utilize the relevant life of this storage device storage
Order and infrared coding order, compared to existing technology, it may not be necessary to the support of network can realize various household electrical appliances difference in functionalitys
Manipulation.
3, the present invention controls device and accepts the phonetic order of operator, matches according to instruction, and passes through infrared coding
Signal exports.Use bluetooth wireless voice frequency transmission device or realized controlling coded modulation to radio frequency by WIFI communication module
On radio carrier, operator carries Bluetooth audio frequency emitter, is realized home wiring control by voice.
Below in conjunction with specific embodiment, the present invention is further illustrated.
Detailed description of the invention
Embodiment 1
A kind of sound control method, it comprises the following steps:
1) hidden Markov model and neutral net is used to carry out probability calculation during acoustic features is extracted;
2) resampling is carried out by the particle filter signal to extracting;
3) probability Distribution Model of voice and certain language regulation is set up;
4) to step 2) acoustic features after resampling gives weight, and is weighted;
5) with regard to the probability Distribution Model of comparison of computational results step 3) of step 4), when the result of calculation of step 4) does not meets step
3) probability Distribution Model, shields worn-out step 2) resampling correspondence voice;
6) for meet probability Distribution Model acoustic features use multiplex mode recognition methods identify further, described many molalities
Formula recognition methods includes including following steps:
The 6.1 target differences sent according to phonetic order, devise independent speech model and language model group;
Its acoustic features and different types of Model Group are compared when receiving phonetic order by 6.2, are first divided by phonetic order
To a certain big class model, then carry out pattern recognition with the Model Group within this big class model, until last and minimum modulus
Concrete model in type group compares, and completes the identification of phonetic order;
7) use pca method partiting step 6.2 Model Group size and the foundation of level, described principle component analysis include with
Lower step:
7.1 extract several acoustic featuress;
7.2 to find an identification in multiple features the highest, and the feature that can be distinguished by model is as the master of pattern recognition
Will foundation, and as pivot;
7.3 pairs of all models are classified, and obtain ground floor Model Group;
7.4 find a topmost feature as pivot from remaining feature;
7.5 pairs of all models reclassify, and obtain the Model Group of next level;
7.6 repetition steps 7.4 and 7.5 are until all using complete by all features, and complete the division of multilayered model group.
In the present embodiment, the speech model described in step 6.1 includes N-Gram model, decision-tree model, context-free language
The polyglot models such as syntax language model, and perform according to following steps:
1) voice signal obtained is mated with the principal character in all of language model;
2) it is ranked up with the matching degree of all language models, selects five kinds of models that wherein matching degree is the highest by ballot
Mode determine.
Embodiment 2
A kind of Wearable voice control device of audio recognition method, its include domestic electric appliances controller, user operation device and
Storage device;Wherein, described domestic electric appliances controller is provided with signal emission module and signal receiving module, domestic electric appliances controller
Signal emission module and signal receiving module is utilized remotely to be connected with user operation device, storage device.
Described user operation is for identifying the instruction that user sends, and is transferred to domestic electric appliances controller, and storage device is used
In the infrared coding used of the different domestic electrical equipment remote controller of storage, domestic electric appliances controller is for received encoded signal and to send coding letter
Number, in i.e. receiving the sent order of user operation device, then transfer storage device, corresponding coding is sent to corresponding household electrical appliances, it is achieved right
The control of household electrical appliances.Can also be after reception user operation device " learns " order, then receive the infrared coding of domestic electrical equipment remote controller
Waveform, and it is stored in storage device.
When being controlled, user operation device receives the instruction that user sends, and is identified judging, is then emitted to
Domestic electric appliances controller, in domestic electric appliances controller transfers storage device according to the instruction received, corresponding infrared coding utilizes signal
Transmitter module sends.
Concrete, it is provided with memory block, the infrared volume of the different domestic electrical equipment remote controller of memory block storage in described storage device
Code, is provided with some storage communities in memory block, the quantity of storage community be 5-100 or more than, some storage communities are i.e. used for
Storage controls the different infrared codings of household electrical appliances work;By storage device being divided into memory block and the setting of storage community, household electrical appliances
After controlling the order that device receives user operation device, can first determine to should the memory block of household electrical appliances infrared coding, then from this
Memory block is found out the storage community of corresponding received command, the most then the infrared coding in this storage community is passed through emitter
Launching, this mode can make domestic electric appliances controller precisely and quickly determine the infrared coding of corresponding order and launch
Going, accuracy rate is high, and response speed is fast;Such as, memory block includes refrigerator memory block, and it is little that this air-conditioning memory block includes startup storage
District, stop storage community, rise storage community, decline storage community, pattern switching storage community, timing storage community and temperature
Degree 10-50 totally 44 storage communities, wherein start storage community and i.e. store the startup infrared coding of air-conditioning, stop storage community i.e.
The stopping infrared coding of storage air-conditioning, raising and lowering storage community i.e. stores and increases temperature and reduce the infrared coding of temperature,
The pattern switching i.e. infrared coding of storage switching air conditioning mode, timing storage community i.e. stores the infrared volume of timing of air-conditioning remote control
Code, temperature 10-50 stores the infrared coding of 10-50 ° respectively;User, when operation, sends order such as " starting of air conditioner 24 ° ", family
Controller for electric consumption then can first send the startup infrared coding of air-conditioning and start air-conditioning, sends the infrared coding of 24 ° the most again by air-conditioning
It is adjusted to 24 °, if order is " air-conditioner temperature decline ", then sends the infrared coding declined in storage community, make under air-conditioner temperature
Drop 1 °.
Concrete, described signal emission module is provided with an infrared transmitter, the signal launch party of infrared transmitter
To being positioned at the front of Wearable phonetic controller, when i.e. user normally wears Wearable phonetic controller, the front of user
It is the transmitting direction of infrared transmitter;By the setting of infrared transmitter, domestic electric appliances controller directly initiates after receiving order
Corresponding infrared transmitter, determine in storage device the most again to should infrared transmitter memory block, find out in this memory block
The infrared coding of corresponding order is sent by infrared transmitter, and response speed is fast, and accuracy is high;When user uses, user is only
Need just to the household electrical appliances that need to manipulate, then send order, corresponding household electrical appliances can be manipulated accurately, outside this mode is relative to Different Red
The omnibearing transmitting signal of emitter 360 ° is the most accurate, i.e. can improve accurate for home wiring control of the present invention further
Rate.
Concrete, it is provided with speech recognition system, language learning system and infrared learning in described user operation device
System;Described infrared learning system is the infrared coding for learning other domestic electrical equipment remote controllers, and this infrared learning system includes:
Step 1, user sends " study " order, and is recognized by;
Step 2, user operation device starts infrared learning system, starts the signal receiving module of domestic electric appliances controller simultaneously;
Step 3, user operation need to learn the remote controller of household electrical appliances, and domestic electric appliances controller receives the infrared coding that this remote controller is sent
Wave data, and this infrared coding Wave data is stored in storage device.
Further, in step 3, after domestic electric appliances controller receives infrared learning order, vision can be passed through or listen
Feeling and feed back, such as arranging display lamp and continue bright a period of time (such as 6s), user need to (6s waits altogether before display lamp extinguishes
Time) will need to learn the button corresponding to remote controller alignment domestic electric appliances controller press.
Described language learning system is for learning voice command, coordinates with infrared learning system, it is achieved to not consanguinity
The various orders of electrical remote control device learn, concrete:
Step 1, step 1, user sends " study " order, and is recognized by;
Step 2, user operation device starts infrared learning system and language learning system, starts domestic electric appliances controller simultaneously
Signal receiving module;
Step 3, user operation need to learn the remote controller of household electrical appliances, and domestic electric appliances controller receives the infrared coding that this remote controller is sent
Wave data, simultaneously user send to should the order of Wave data of infrared coding, this order is remembered by language learning system
Record, the order of record stores in community be stored into the infrared coding Wave data ordered in storage device.
Such as: air-conditioning remote control need to be learnt, user says " study " order, and infrared learning system and language learning system open
Dynamic, then to press any button of air-conditioning remote control, such as, " open ", " unlatching " infrared coding that air-conditioning remote control sends is by family
Controller for electric consumption receives, and meanwhile, user sends voice and " opens " by language learning system record, and domestic electric appliances controller is by phonology
The voice of learning system record " is opened " order and is stored in voice command memory block, simultaneously also by this voice command and infrared learning
Systematic learning to should voice " open " infrared coding ordered simultaneously be stored in storage device in one storage community in,
Another button that then can press remote controller learns, and carries out learning the various remote controls life of air-conditioning remote control successively
Order.
Wherein, described speech recognition system is provided with identification engine, and this identification engine includes: start order and study life
Order, and voice command memory block, start order and be used for starting Wearable voice control device of the present invention, and study order is used for making
Wearable voice control device of the present invention enters learning state;By only arranging two basic commands, speech recognition can be reduced
System identify error rate, can be greatly reduced the interference of environmental noise, improve phonetic recognization rate, user can also basis make by oneself
Voice command control household electrical appliances, easy to use, there is more preferable Consumer's Experience.And the present invention is provided with storage device, profit
Related command and infrared coding order is stored, compared to existing technology, it may not be necessary to the support of network with this storage device
Realize the manipulation to various household electrical appliances difference in functionalitys.
More specifically, in the present invention, it is additionally provided with self stick circuit, in this self stick circuit, is provided with timing means, self-closing electricity
Road is connected with the electric power system of the present invention, and user is not when operating, and timing means enters countdown state or time status, falls
After timing or timing, self stick circuit cuts off electric power system, i.e. closes the present invention, reduces the consumption of the energy.
Described control equipment also includes bluetooth communication, main control chip, function key and display module, phonetic synthesis mould
Block, speech recognition circuit, single-chip microcomputer and wireless communication module;Described bluetooth communication includes Bluetooth audio frequency emitter and bluetooth
Audio receiver;Described Bluetooth audio frequency emitter and Bluetooth audio receiver wireless connections;Described Bluetooth audio receiver connects
Speech recognition circuit audio input end;Described speech recognition circuit data-interface is connected with single-chip microcomputer input/output interface.
During work, microphone receives the sound of user and is converted into the signal of telecommunication, enters sound identification module, by amplifying,
After anti-sidetone process, digitized processing, compare according to specific algorithm with the data of pre-stored criteria sound bank, identify phase
The statement answered or control instruction are come, and send main control chip to, and the design of multiple microphones is beneficial to eliminate background noise to carry
High s/n ratio;Main control chip goes to perform corresponding program according to the instruction received, and shows relevant information by display module, simultaneously
Again the directive statement received is converted into default voice by voice synthetic module, says corresponding language by earphone or loudspeaker
Sound instruction comes, and user is after hearing feedback voice messaging, it can be determined that the phonetic order above sent is the most correct.This mode
Just can improve speech discrimination score and reliability by man-machine interaction.
It addition, use voice remote controller first, need all electrical equipment to be carried out adaptability pairing, so passing through infrared receiver
Module carries out intellectual learning and pairing, and concrete mode is exactly main control chip when performing learning model, is used by voice reminder
Family, the remote controller of controlled electric appliance facing to infrared receiving module, the keyboard of pressing remote controller, sends corresponding infrared information and compiles
Code, after infrared receiving module receives infrared signal and filters carrier wave, sends main control chip to the keyboard coding information obtained, main
Control chip analyzes this remote controller codes type by the standard code storehouse that memory element prestores, then by man-machine communication, this is distant
Control-register code type is saved in the middle of memory element with associating of controlled electric appliance, for controlling to call later.If compiled at pre-stored criteria
Corresponding type of coding is can not find it is necessary to use coding copy mode, coding letter corresponding for each button of remote controller in code storehouse
Breath is saved in memory element, same for controlling to call later.
In the present embodiment, after main control chip receives and analyzes the phonetic order that user sends, this instruction from
Taking out corresponding infrared remote coding in memory element, pass to infrared transmission module, infrared transmission module realizes infrared coding
Modulate on infrared carrier wave, then send infrared remote-controlled signal by infrared transmitting tube to controlled electric appliance.Due to memory element capacity
Restriction, and numerous and diverse degree of standard language, when the process by Wearable phonetic controller is the most not ideal enough, so language
Identification cannot depend merely on sound identification module and realize, and be also to the high in the clouds language identification server by the Internet of plugging into
Realize.This mode will be docked by WIFI communication module with router and be connected to the Internet and be realized.
Claims (7)
1. a sound control method, it is characterised in that it comprises the following steps:
1) hidden Markov model and neutral net is used to carry out probability calculation during acoustic features is extracted;
2) resampling is carried out by the particle filter signal to extracting;
3) probability Distribution Model of voice and certain language regulation is set up;
4) to step 2) acoustic features after resampling gives weight, and is weighted;
5) with regard to the probability Distribution Model of comparison of computational results step 3) of step 4), when the result of calculation of step 4) does not meets step
3) probability Distribution Model, shields worn-out step 2) resampling correspondence voice.
2. a kind of sound control method as claimed in claim 1, it is characterised in that: for meeting the acoustics of probability Distribution Model
Feature uses multiplex mode recognition methods to identify further, and described multiplex mode recognition methods comprises the following steps:
1) target sent according to phonetic order is different, devises independent speech model and language model group;
2) when receiving phonetic order, its acoustic features and different types of Model Group are compared, first phonetic order is divided to
In a certain big class model, then carry out pattern recognition with the Model Group within this big class model, until last and least model
Concrete model in Qun compares, and completes the identification of phonetic order.
3. a kind of audio recognition method as claimed in claim 2, it is characterised in that use pca method partiting step 2)
Model Group size and the foundation of level, described principle component analysis comprises the following steps:
1) several acoustic features is extracted;
2) in multiple features, an identification is found the highest, main as pattern recognition of the feature that model can be distinguished
Foundation, and as pivot;
3) all models are classified, obtain ground floor Model Group;
4) from remaining feature, a topmost feature is found as pivot;
5) all models are reclassified, obtain the Model Group of next level;
6) step 4) and 5 is repeated) until all features are all used complete, and complete the division of multilayered model group.
4. a kind of audio recognition method as claimed in claim 2, it is characterised in that: the speech model described in step 1) includes N-
The polyglot models such as Gram model, decision-tree model, CFG language model, and perform according to following steps:
1) voice signal obtained is mated with the principal character in all of language model;
2) it is ranked up with the matching degree of all language models, selects five kinds of models that wherein matching degree is the highest by ballot
Mode determine.
5. the Wearable voice control device using audio recognition method described in claim 1, it is characterised in that: it includes
There are domestic electric appliances controller, user operation device and storage device;Wherein, described domestic electric appliances controller is provided with signal and launches mould
Block and signal receiving module, domestic electric appliances controller utilizes signal emission module and signal receiving module and user operation device and deposits
Storage device remotely connects;The memory block for storing different home wiring control order, described letter it is provided with in described storage device
Number receiver module is provided with infrared transmitter;Speech recognition system it is provided with, for learning language in described user operation device
The phonetic study decorum of sound order and for learning the infrared learning system of infrared coding;Described language learning system is with infrared
Learning system coordinates, and infrared learning system is connected by the signal receiving module of domestic electric appliances controller and is connected with storage device;Institute
Stating speech recognition system and be provided with identification engine, this identification engine includes starting order, study order and phonetic storage district, language
Sound learning system is connected with phonetic storage district and storage device.
The Wearable voice control device of a kind of audio recognition method the most as claimed in claim 5, it is characterised in that deposit described in:
Storage area is both provided with some storage communities;Described control equipment is additionally provided with self stick circuit, is provided with timing dress in this self stick circuit
Put.
The Wearable voice control device of a kind of audio recognition method the most as claimed in claim 5, it is characterised in that: described control
Control equipment also include bluetooth communication, main control chip, function key and display module, voice synthetic module, speech recognition circuit,
Single-chip microcomputer and wireless communication module;Described bluetooth communication includes Bluetooth audio frequency emitter and Bluetooth audio receiver;Described
Bluetooth audio frequency emitter and Bluetooth audio receiver wireless connections;Described Bluetooth audio receiver connects speech recognition circuit audio frequency
Input;Described speech recognition circuit data-interface is connected with single-chip microcomputer input/output interface.
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2015206307314 | 2015-08-20 | ||
CN201520630731 | 2015-08-20 |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106023995A true CN106023995A (en) | 2016-10-12 |
Family
ID=57108612
Family Applications (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610535986.1A Pending CN106023995A (en) | 2015-08-20 | 2016-07-10 | Voice recognition method and wearable voice control device using the method |
CN201620718248.6U Active CN206097455U (en) | 2015-08-20 | 2016-07-10 | Speech recognition controlgear |
Family Applications After (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201620718248.6U Active CN206097455U (en) | 2015-08-20 | 2016-07-10 | Speech recognition controlgear |
Country Status (1)
Country | Link |
---|---|
CN (2) | CN106023995A (en) |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107347111A (en) * | 2017-05-16 | 2017-11-14 | 上海与德科技有限公司 | The control method and terminal of terminal |
CN108011789A (en) * | 2017-12-14 | 2018-05-08 | 四川声达创新科技有限公司 | A kind of artificial intelligence deep learning neutral net embedded control system |
CN108022592A (en) * | 2018-01-17 | 2018-05-11 | 深圳市沃特沃德股份有限公司 | Voice remote control method and device |
CN108492545A (en) * | 2018-03-15 | 2018-09-04 | 北京金山安全软件有限公司 | Infrared remote control method, device, remote control equipment and medium |
CN108648747A (en) * | 2018-03-21 | 2018-10-12 | 清华大学 | Language recognition system |
CN109903769A (en) * | 2017-12-08 | 2019-06-18 | Tcl集团股份有限公司 | A kind of method, apparatus and terminal device of terminal device interaction |
WO2019161686A1 (en) * | 2018-02-22 | 2019-08-29 | Midea Group Co., Ltd. | Machine generation of context-free grammar for intent deduction |
CN110428815A (en) * | 2019-08-01 | 2019-11-08 | 格音智能科技(上海)有限公司 | A kind of intelligent home voice interactive system and method based on speech recognition technology |
CN111092798A (en) * | 2019-12-24 | 2020-05-01 | 东华大学 | Wearable system based on spoken language understanding |
CN111240540A (en) * | 2019-12-27 | 2020-06-05 | 咪咕视讯科技有限公司 | Video adjusting method, terminal with flexible screen and storage medium |
CN114495920A (en) * | 2022-02-12 | 2022-05-13 | 深圳市宏芯达科技有限公司 | AI all-in-one chip |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108199938A (en) * | 2018-03-07 | 2018-06-22 | 佛山市云米电器科技有限公司 | A kind of voice control formula intelligent sound box |
CN108346207A (en) * | 2018-04-12 | 2018-07-31 | 伟创亚洲有限公司 | The speech control system and method for intelligent gradevin |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
GB2240203A (en) * | 1990-01-18 | 1991-07-24 | Apple Computer | Automated speech recognition system |
CN1349213A (en) * | 2001-12-18 | 2002-05-15 | 蔡磊 | Speech sound control system of household electric appliance and lighting device |
CN101794126A (en) * | 2009-12-15 | 2010-08-04 | 广东工业大学 | Wireless intelligent home appliance voice control system |
CN102411931A (en) * | 2010-09-15 | 2012-04-11 | 微软公司 | Deep belief network for large vocabulary continuous speech recognition |
CN103117060A (en) * | 2013-01-18 | 2013-05-22 | 中国科学院声学研究所 | Modeling approach and modeling system of acoustic model used in speech recognition |
-
2016
- 2016-07-10 CN CN201610535986.1A patent/CN106023995A/en active Pending
- 2016-07-10 CN CN201620718248.6U patent/CN206097455U/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
GB2240203A (en) * | 1990-01-18 | 1991-07-24 | Apple Computer | Automated speech recognition system |
CN1349213A (en) * | 2001-12-18 | 2002-05-15 | 蔡磊 | Speech sound control system of household electric appliance and lighting device |
CN101794126A (en) * | 2009-12-15 | 2010-08-04 | 广东工业大学 | Wireless intelligent home appliance voice control system |
CN102411931A (en) * | 2010-09-15 | 2012-04-11 | 微软公司 | Deep belief network for large vocabulary continuous speech recognition |
CN103117060A (en) * | 2013-01-18 | 2013-05-22 | 中国科学院声学研究所 | Modeling approach and modeling system of acoustic model used in speech recognition |
Non-Patent Citations (3)
Title |
---|
俞利强等: "基于PCA技术的神经网络说话人识别研究", 《计算机工程与应用》 * |
杨海峰等: "模式识别理论和技术在语音识别研究中的应用 ", 《合肥学院学报(自然科学版)》 * |
林坤辉等: "基于HMM与神经网络的声学模型研究", 《厦门大学学报(自然科学版)》 * |
Cited By (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107347111A (en) * | 2017-05-16 | 2017-11-14 | 上海与德科技有限公司 | The control method and terminal of terminal |
CN109903769A (en) * | 2017-12-08 | 2019-06-18 | Tcl集团股份有限公司 | A kind of method, apparatus and terminal device of terminal device interaction |
CN108011789A (en) * | 2017-12-14 | 2018-05-08 | 四川声达创新科技有限公司 | A kind of artificial intelligence deep learning neutral net embedded control system |
CN108022592A (en) * | 2018-01-17 | 2018-05-11 | 深圳市沃特沃德股份有限公司 | Voice remote control method and device |
CN112074898A (en) * | 2018-02-22 | 2020-12-11 | 美的集团股份有限公司 | Machine generation of context-free grammars for intent inference |
WO2019161686A1 (en) * | 2018-02-22 | 2019-08-29 | Midea Group Co., Ltd. | Machine generation of context-free grammar for intent deduction |
US10496382B2 (en) | 2018-02-22 | 2019-12-03 | Midea Group Co., Ltd. | Machine generation of context-free grammar for intent deduction |
CN112074898B (en) * | 2018-02-22 | 2023-12-01 | 美的集团股份有限公司 | Machine generation of context-free grammar for intent reasoning |
CN108492545A (en) * | 2018-03-15 | 2018-09-04 | 北京金山安全软件有限公司 | Infrared remote control method, device, remote control equipment and medium |
CN108648747A (en) * | 2018-03-21 | 2018-10-12 | 清华大学 | Language recognition system |
CN108648747B (en) * | 2018-03-21 | 2020-06-02 | 清华大学 | Language identification system |
CN110428815A (en) * | 2019-08-01 | 2019-11-08 | 格音智能科技(上海)有限公司 | A kind of intelligent home voice interactive system and method based on speech recognition technology |
CN111092798A (en) * | 2019-12-24 | 2020-05-01 | 东华大学 | Wearable system based on spoken language understanding |
CN111240540A (en) * | 2019-12-27 | 2020-06-05 | 咪咕视讯科技有限公司 | Video adjusting method, terminal with flexible screen and storage medium |
CN111240540B (en) * | 2019-12-27 | 2023-11-10 | 咪咕视讯科技有限公司 | Video adjustment method, terminal with flexible screen and storage medium |
CN114495920A (en) * | 2022-02-12 | 2022-05-13 | 深圳市宏芯达科技有限公司 | AI all-in-one chip |
Also Published As
Publication number | Publication date |
---|---|
CN206097455U (en) | 2017-04-12 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106023995A (en) | Voice recognition method and wearable voice control device using the method | |
CN108091331B (en) | Intelligent household system based on voice control and control method thereof | |
CN201129826Y (en) | Air conditioner control device | |
CN102902253B (en) | Intelligent switch with voice control function and intelligent control system | |
CN105068460B (en) | A kind of intelligence control system | |
US20020123896A1 (en) | Control module for HVAC systems | |
CN109974235A (en) | Control the method, apparatus and household appliance of household appliance | |
CN101794126A (en) | Wireless intelligent home appliance voice control system | |
CN107818782B (en) | Method and system for realizing intelligent control of household appliances | |
CN105045122A (en) | Intelligent household natural interaction system based on audios and videos | |
US20200082827A1 (en) | Artificial intelligence-based appliance control apparatus and appliance controlling system including the same | |
CN106647311A (en) | Intelligent central control system and equipment, server and intelligent equipment control method | |
CN110942773A (en) | Method and device for controlling intelligent household equipment through voice | |
CN104133387A (en) | Intelligent kitchen host, and work method and control method of intelligent kitchen host | |
CN109474658A (en) | Electronic equipment, server and the recording medium of task run are supported with external equipment | |
CN109617772A (en) | A kind of smart home system based on speech recognition | |
CN108534297A (en) | A kind of intelligent air-conditioning system and control method based on speech recognition | |
KR20150053447A (en) | Wireless extender for interworking smart device and employment method thereof | |
CN109949803B (en) | Building service facility control method and system based on semantic instruction intelligent identification | |
CN106764965A (en) | Lighting apparatus control method and lighting apparatus | |
CN107230479A (en) | A kind of smart home multi-kind control device with the artificial maincenter of machine | |
CN111580406A (en) | Intelligent home control system based on Internet of things | |
CN113516975A (en) | Intelligent household voice-operated switch system and control method | |
WO2020175293A1 (en) | Apparatus control system, apparatus control method, and program | |
CN109947067A (en) | Control method applied to intelligent home control system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
CB02 | Change of applicant information | ||
CB02 | Change of applicant information |
Address after: 363000 Six Blocks, 28 Longxiang North Road, Lantian Development Zone, Longwen District, Zhangzhou City, Fujian Province Applicant after: Fujian Kaibang Electronics Co., Ltd. Address before: 363000 No. 16 Longtengbei Road, Longwen District, Zhangzhou City, Fujian Province Applicant before: Zhangzhou Kaibang Electronics Co.,Ltd. |
|
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20161012 |