CN108647346A - A kind of the elderly's voice interactive method and system for wearable electronic - Google Patents
A kind of the elderly's voice interactive method and system for wearable electronic Download PDFInfo
- Publication number
- CN108647346A CN108647346A CN201810460399.XA CN201810460399A CN108647346A CN 108647346 A CN108647346 A CN 108647346A CN 201810460399 A CN201810460399 A CN 201810460399A CN 108647346 A CN108647346 A CN 108647346A
- Authority
- CN
- China
- Prior art keywords
- word
- elderly
- voice
- user
- wearable electronic
- 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.)
- Granted
Links
- 230000002452 interceptive effect Effects 0.000 title claims abstract description 73
- 238000000034 method Methods 0.000 title claims abstract description 16
- 238000012937 correction Methods 0.000 claims abstract description 24
- 238000011109 contamination Methods 0.000 claims description 30
- 238000012549 training Methods 0.000 claims description 25
- 230000036760 body temperature Effects 0.000 claims description 9
- 238000001514 detection method Methods 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 4
- 230000015572 biosynthetic process Effects 0.000 claims 1
- 230000003993 interaction Effects 0.000 claims 1
- 238000003786 synthesis reaction Methods 0.000 claims 1
- 230000000875 corresponding effect Effects 0.000 description 15
- 230000006870 function Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000036387 respiratory rate Effects 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 230000002618 waking effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/284—Lexical analysis, e.g. tokenisation or collocates
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/12—Fingerprints or palmprints
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/06—Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice
- G10L15/063—Training
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; 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]
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/06—Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice
- G10L15/063—Training
- G10L2015/0631—Creating reference templates; Clustering
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L2015/086—Recognition of spelled words
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L2015/088—Word spotting
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/22—Procedures used during a speech recognition process, e.g. man-machine dialogue
- G10L2015/223—Execution procedure of a spoken command
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Probability & Statistics with Applications (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- User Interface Of Digital Computer (AREA)
Abstract
The present invention relates to a kind of the elderly's voice interactive methods and system for wearable electronic, after getting the fingerprint of user, until Cloud Server inquiry user identity and age of user;If the age of user is in default the elderly's the range of age, the wearable electronic enters the elderly's interactive mode;Under the elderly's interactive mode, the speed for playing voice is slowed down preset multiple by the wearable electronic;The volume for playing voice is improved preset multiple by the wearable electronic;And the wearable electronic carries out error correction when getting the control voice of user to the control voice of user.The present invention fully takes into account the demand that interactive voice is carried out with the elderly, is adjusted to playing voice, and carry out error correction to control voice, ensures the control instruction for correctly getting old user, to execute corresponding task;Dictionary and language model are created simultaneously, is more effectively carried out error correction, improves the accuracy of voice error correction.
Description
Technical field
The present invention relates to technical field of data processing, in particular to a kind of the elderly's voices for wearable electronic
Exchange method and system.
Background technology
With the development of society, the type and function of wearable electronic are more and more, it can not only detect user's
Physiological parameter, such as detection user body temperature, detection user heart rate, detection user's respiratory rate etc. are used as user's body health assessment
Reference, can also further receive the instruction of user to execute corresponding action, such as receive the phonetic order of user, play
Song receives the phonetic order of user, actively measures body temperature etc., so as to the use of more convenient user.
However, in practical applications, the wearable electronic of the prior art does not advantageously take into account different user but
Demand.For example, for old user, for wearable electronic when playing voice prompt, possible volume is too small, word speed
It is too fast and cause old user that can not catch.In addition, old user is using voice mode input control when being instructed, it can also
Because it is ambiguous to speak, and make wearable electronic that can not correctly obtain control instruction.
Invention content
The present invention provides a kind of the elderly's voice interactive method and system for wearable electronic, purpose exists
In overcoming defect in the prior art, the demand for carrying out interactive voice with the elderly is fully taken into account, is adjusted to playing voice
It is whole, and error correction is carried out to control voice, ensure the control instruction for correctly getting old user, to execute corresponding task.
To achieve the goals above, the present invention has following constitute:
This is used for the elderly's voice interactive method of wearable electronic, includes the following steps:
Training dictionary is generated, the training dictionary includes the preset association relationship between multiple words and each word;
It includes arbitrary three word combinations to obtain training dictionary and sentence library, train language model, the language model
The probability occurred at a sentence;
When temperature sensor in the wearable electronic detects the temperature within the scope of default body temperature,
Judgement has user to wear the wearable electronic;
The wearable electronic prompts user to carry out fingerprint authentication by display screen and loud speaker;
After the wearable electronic gets the fingerprint of user, until Cloud Server inquiry user identity and user year
Age;
If the age of user is in default the elderly's the range of age, the wearable electronic enters the elderly
Interactive mode;
Under the elderly's interactive mode, the speed for playing voice is slowed down preset multiple by the wearable electronic;
Under the elderly's interactive mode, the volume for playing voice is improved preset multiple by the wearable electronic;
Under the elderly's interactive mode, when the wearable electronic gets the control voice of user, to user
Control voice carry out error correction;
The control voice to user carries out error correction, includes the following steps:
The control voice of user is resolved to multiple words, and selection and each word from dictionary by the wearable electronic
Other associated close words;
Each word and its associated word are combined by the wearable electronic with other words, and combination is put into institute
It states in language model and is trained, obtain the highest contamination of probability of occurrence, and the sequence of the control voice according to user, it will
Each word is ranked up in the highest contamination of probability of occurrence, obtains correct control statement;
The wearable electronic executes corresponding task according to the correct control statement.
Optionally, multiple words in the training dictionary and the preset association relationship between each word, including each word
And the word respectively with the similarity value of other words;
Described selection other close words associated with each word from dictionary, including:
The degree of approximation of selection and each word is more than the word of default degree of approximation threshold value, the close word as the word from dictionary.
Optionally, train language model, including arbitrary three morphologies are combined at one, and calculate the general of combination appearance
Rate;
The probability for calculating the combination and occurring, includes the following steps:
It calculates and occurs the sentence number of the combination in the sentence library;
Total sentence number in the sentence number of the combination divided by the sentence library to occur, obtains the probability of combination appearance.
Optionally, described to be combined each word and its associated word with other words, include the following steps:
Determine that the number of total word included by the control voice is n;
Multiple combinations are obtained from extraction in the associated word of n word and each word, each combination includes each in n word
Word of word itself or an associated word.
Optionally, described combination is put into the language model is trained, and obtains the group of the highest word of probability of occurrence
It closes, includes the following steps:
N word in each combination is divided into multiple three lexons combinations, each subgroup is searched from the language model
The probability occurred is closed, and the probability that each sub-portfolio is found is multiplied, obtains the probability of combination appearance;
The probability that relatively more each combination occurs, selects the highest contamination of probability of occurrence.
Optionally, the sequence of the control voice according to user, by each word in the highest contamination of probability of occurrence
It is ranked up, obtains correct control statement, include the following steps:
N positions of composing a poem to a given tune of ci are generated in correct control statement, select the highest contamination of probability of occurrence, it successively will combination
In each word be filled into corresponding n and compose a poem to a given tune of ci in positions;Institute's predicate is corresponding with position of composing a poem to a given tune of ci to refer to that the word is the word of the position of composing a poem to a given tune of ci
Itself or associated word.
The embodiment of the present invention also provides a kind of the elderly's voice interactive system of wearable electronic, including:
Language model training module, for generating training dictionary, the training dictionary includes multiple words and each
Preset association relationship between word;And training dictionary and sentence library are obtained, train language model wraps in the language model
It includes arbitrary three phrases and synthesizes the probability that a sentence occurs;
Human detection module, when for detecting the temperature being within the scope of default body temperature, judgement has user's wearing
The wearable electronic;
User authentication module, for prompting user to carry out fingerprint authentication by display screen and loud speaker;And it obtains
To after the fingerprint of user, until Cloud Server inquiry user identity and age of user;
The elderly's interactive mode control module is preset if the age for user is in the elderly's the range of age,
The wearable electronic enters the elderly's interactive mode;The speed for playing voice is slowed down preset multiple by control;Control will
The volume for playing voice improves preset multiple;And when getting the control voice of user, the control voice of user is entangled
It is wrong;
It is described that error correction is carried out to control voice, it is carried out according to following steps:
The control voice of user is resolved to multiple words by the elderly's interactive mode control module, and is selected from dictionary
Other close words associated with each word;
Each word and its associated word are combined by the elderly's interactive mode control module with other words, and by group
Conjunction is put into the language model and is trained, and obtains the highest contamination of probability of occurrence, and according to the control voice of user
Sequence, each word in the highest contamination of probability of occurrence is ranked up, correct control statement is obtained;
Control instruction execution module, the correct control language for being exported according to the elderly's interactive mode control module
Sentence executes corresponding task.
Optionally, each word and its associated word are carried out group by the elderly's interactive mode control module with other words
It closes, is executed according to following steps:
The elderly's interactive mode control module determines that the number of total word included by the control voice is n;
The elderly's interactive mode control module is extracted from the associated word of n word and each word and obtains multiple groups
It closes, each combination includes the word itself or an associated word of each word in n word.
Optionally, combination is put into the language model by the elderly's interactive mode control module is trained, and obtains
To the highest contamination of probability of occurrence, include the following steps:
N word in each combination is divided into multiple three lexons and combined by the elderly's interactive mode control module, from
The probability that each sub-portfolio occurs is searched in the language model, and the probability that each sub-portfolio is found is multiplied, and is obtained
The probability occurred to the combination;
The elderly's interactive mode control module more each combines the probability of appearance, selects the highest word of probability of occurrence
Combination.
Optionally, the sequence of the control voice according to user, by each word in the highest contamination of probability of occurrence
It is ranked up, obtains correct control statement, include the following steps:
The elderly's interactive mode control module generates n positions of composing a poem to a given tune of ci in correct control statement, and selection occurs general
Each word in combination is filled into corresponding n positions of composing a poem to a given tune of ci by the highest contamination of rate successively;Institute's predicate and position of composing a poem to a given tune of ci are right
It should refer to the word itself or associated word that the word is the position of composing a poem to a given tune of ci.
The elderly's voice interactive method and system for wearable electronic in the invention are used, is had as follows
Advantageous effect:
(1) demand for carrying out interactive voice with the elderly is fully taken into account, is adjusted to playing voice, and to control
Voice carries out error correction, ensures the control instruction for correctly getting old user, to execute corresponding task;
(2) it since wearable electronic is typically all to be close to the users what skin used, is examined using temperature sensor
When measuring user temperature, you can to show to have user's use, user fingerprints can be obtained at this time, further carried out to user identity
Verification, and use pattern control is carried out according to the age of different user or use habit;
(3) dictionary and language model are created, error correction more effectively is carried out to the control voice of user, improves voice error correction
Accuracy is more applicable for large-scale promotion application.
Description of the drawings
Fig. 1 is the flow chart of the elderly's voice interactive method for wearable electronic of the present invention;
Fig. 2 is that the control voice to user of the present invention carries out the flow chart of error correction;
Fig. 3 is the structural schematic diagram of the elderly's voice interactive system for wearable electronic of the present invention.
Specific implementation mode
In order to more clearly describe the technology contents of the present invention, carried out with reference to specific embodiment further
Description.
The elderly's voice interactive method and system for wearable electronic as shown in Figure 1, includes the following steps:
S100:Training dictionary is generated, the training dictionary includes the default association between multiple words and each word
Relationship;
S200:It includes arbitrary three words to obtain training dictionary and sentence library, train language model, the language model
It is combined into the probability of sentence appearance;Training dictionary and sentence library can be relatively high from accuracy such as novel, magazine, newspapers
Text in carry out plucking choosing and extraction;
S300:Temperature sensor in the wearable electronic detects the temperature within the scope of default body temperature
When spending, judgement has user to wear the wearable electronic;Since wearable electronic is typically all skin of being close to the users
When using, therefore user temperature detected using temperature sensor, you can to show to have user's use, therefore passed using temperature
Sensor carries out automatically waking up for wearable electronic, and application is very convenient, and is waken up manually again without user;
S400:The wearable electronic prompts user to carry out fingerprint authentication by display screen and loud speaker, and in obtaining
After getting the fingerprint of user, until Cloud Server inquiry user identity and age of user;
S500:If the age of user is in default the elderly's the range of age, the wearable electronic enters
The speed for playing voice is slowed down preset multiple, and the volume for playing voice is improved preset multiple by the elderly's interactive mode;
For example, it is contemplated that declining to the possible understandability of the elderly, Listening Ability of Ethnic also declines, and can will play the speed of voice
Slow down to 0.8 times of former speed, the volume for playing voice is improved to 1.2 times of former volume, more convenient user hears wearable
The voice that electronic equipment plays, the voice that wearable electronic plays is different and different according to function, such as can play day
Gas information, play cuing voice, broadcasting music, broadcasting physiological data measurement result etc. etc.;
And it can be adjusted with intelligent recognition broadcasting content, such as when playing music, it is undistorted in order to avoid playing, then
Only raising volume is without slow-down, then can not only slow-down but also raising volume etc. when playing Weather information.
S600:It is right when the wearable electronic gets the control voice of user under the elderly's interactive mode
The control voice of user carries out error correction;
The control voice to user carries out error correction, includes the following steps:
S610:The control voice of user is resolved to multiple words by the wearable electronic, and from dictionary selection with
Other associated close words of each word;
S620:Each word and its associated word are combined by the wearable electronic with other words, and will combination
It is put into the language model and is trained, obtain the highest contamination of probability of occurrence;
S630:According to the sequence of the control voice of user, each word in the highest contamination of probability of occurrence is arranged
Sequence obtains correct control statement;
S700:The wearable electronic executes corresponding task according to the correct control statement.
Further, the present invention can also count old user usually wrong some words, by it on the basis of error correction
Corresponding word is as substitute in correct control statement, when the word that detect a word that old user says be usually wrong
When, its substitute is searched automatically, you can with the demand realized self study with adapt to old user.
In this embodiment, in the step S100, the default pass between multiple words and each word in training dictionary
Connection relationship, including each word and the word respectively with the similarity value of other words;
Described selection other close words associated with each word from dictionary, including:
The degree of approximation of selection and each word is more than the word of default degree of approximation threshold value, the close word as the word from dictionary.
Optionally, train language model, including arbitrary three morphologies are combined at one, and calculate the general of combination appearance
Rate;
The probability for calculating the combination and occurring, includes the following steps:
It calculates and occurs the sentence number of the combination in the sentence library;
Total sentence number in the sentence number of the combination divided by the sentence library to occur, obtains the probability of combination appearance.
Optionally, described to be combined each word and its associated word with other words, include the following steps:
Determine that the number of total word included by the control voice is n;
Multiple combinations are obtained from extraction in the associated word of n word and each word, each combination includes each in n word
Word of word itself or an associated word.
For example, a user wants to say " me please be helped to play third song " originally, but because speak ambiguous
Or accent problem, say to may be " me is gently helped to play the 30th music ", then cutting is " light " " side " " I " " broadcasting " " the
30 " " music ".For wearable electronic, possibly correct sentence can not be correctly recognized.
At this point, searching light associated word, find two " clear " " asking ", searches the associated word of " side ", find two
" list ", " pound " etc., and so on, calculate combine when, 6 words, the quantity of the associated word of each word be respectively m1, m2,
M3, m4, m5 and m6, the then combination that can be formed have the combination of m1*m2*m3*m4*m5*m6 kinds.
Further, described combination is put into the language model is trained, and obtains the highest word of probability of occurrence
Combination, includes the following steps:
N word in each combination is divided into multiple three lexons combinations, each subgroup is searched from the language model
The probability occurred is closed, and the probability that each sub-portfolio is found is multiplied, obtains the probability of combination appearance;
Such as being divided into two sub-portfolio " asking " " side " " I " sub-portfolios in combination " me please be help to play the 30th music "
And " broadcasting " " the 30th " " music " sub-portfolio, the probability of two sub-portfolios are respectively 0.6 and 0.05, what which occurred
Probability is 0.6*0.05;And for be two sub-portfolio " asking " " side " " I " sub-portfolios in " me please be help to play third song " with
And " broadcasting " " third is first " " music " sub-portfolio, the probability of two sub-portfolios is respectively 0.6 and 0.5, which occurs general
Rate is 0.6*0.5, the probability occurred much larger than " me please be helped to play the 30th music ".
In addition, when the word split out in a sentence is not just 3 multiple, it can be in finally one common-use words of supplement
Gas word " " " OK " " can with " " " etc., to supply the last one sub-portfolio.Such as " me please be help to play the first English song of third
It is bent ", seven words are split as, then can be divided into three sub-portfolios:" asking " " side " " I " sub-portfolio, " broadcasting " " third is first " " English "
Sub-portfolio and " music " " can with " " " sub-portfolio.
The probability that relatively more each combination occurs, selects the highest contamination of probability of occurrence.
Further, the sequence of the control voice according to user, will be each in the highest contamination of probability of occurrence
Word is ranked up, and is obtained correct control statement, is included the following steps:
N positions of composing a poem to a given tune of ci are generated in correct control statement, select the highest contamination of probability of occurrence, it successively will combination
In each word be filled into corresponding n and compose a poem to a given tune of ci in positions;Institute's predicate is corresponding with position of composing a poem to a given tune of ci to refer to that the word is the word of the position of composing a poem to a given tune of ci
Itself or associated word.
For example, in " me is gently helped to play the 30th music ", there are six position of composing a poem to a given tune of ci, the finally obtained highest combination of probability is
" me please be helped to play third song " " will then ask " position for being put into former " light ", and " side " will be put into the position of former " side ", with such
It pushes away, " third is first " is put into the position of former " the 30th ".
As shown in figure 3, the embodiment of the present invention also provides a kind of the elderly's voice interactive system of wearable electronic, packet
It includes:
Language model training module 100, for generating training dictionary, the training dictionary includes multiple words and each
Preset association relationship between a word;And training dictionary and sentence library are obtained, train language model, in the language model
The probability that a sentence occurs is synthesized including arbitrary three phrases;
Human detection module 200, when for detecting the temperature being within the scope of default body temperature, judgement has user's pendant
Wear the wearable electronic;
User authentication module 300, for prompting user to carry out fingerprint authentication by display screen and loud speaker;And it obtains
After getting the fingerprint of user, until Cloud Server inquiry user identity and age of user;
The elderly's interactive mode control module 400 is preset if the age for user is in the elderly's the range of age,
Then the wearable electronic enters the elderly's interactive mode;The speed for playing voice is slowed down preset multiple by control;Control
The volume for playing voice is improved into preset multiple;And when getting the control voice of user, the control voice of user is carried out
Error correction;
It is described that error correction is carried out to control voice, it is carried out according to following steps:
The control voice of user is resolved to multiple words by the elderly's interactive mode control module, and is selected from dictionary
Other close words associated with each word;
Each word and its associated word are combined by the elderly's interactive mode control module with other words, and by group
Conjunction is put into the language model and is trained, and obtains the highest contamination of probability of occurrence, and according to the control voice of user
Sequence, each word in the highest contamination of probability of occurrence is ranked up, correct control statement is obtained;
Control instruction execution module 500, the correct control for being exported according to the elderly's interactive mode control module
Sentence processed executes corresponding task.Executing task herein can be the playing music of the task, the reporting weather of the task, report body temperature
The task of measurement data, the task dispatching etc. for reporting heart rate, when report according to after adjustment word speed and volume carry out.
As described above, each word and its associated word are carried out group by the elderly's interactive mode control module with other words
It closes, is executed according to following steps:
The elderly's interactive mode control module determines that the number of total word included by the control voice is n;
The elderly's interactive mode control module is extracted from the associated word of n word and each word and obtains multiple groups
It closes, each combination includes the word itself or an associated word of each word in n word.
Introduction has been carried out above for the mode that concrete application is implemented, and details are not described herein.
As described above, combination is put into the language model and is trained by the elderly's interactive mode control module,
The highest contamination of probability of occurrence is obtained, is included the following steps:
N word in each combination is divided into multiple three lexons and combined by the elderly's interactive mode control module, from
The probability that each sub-portfolio occurs is searched in the language model, and the probability that each sub-portfolio is found is multiplied, and is obtained
The probability occurred to the combination;
The elderly's interactive mode control module more each combines the probability of appearance, selects the highest word of probability of occurrence
Combination.
Introduction has been carried out above for the mode that concrete application is implemented, and details are not described herein.
As described above, the sequence of the control voice according to user, it will be each in the highest contamination of probability of occurrence
Word is ranked up, and is obtained correct control statement, is included the following steps:
The elderly's interactive mode control module generates n positions of composing a poem to a given tune of ci in correct control statement, and selection occurs general
Each word in combination is filled into corresponding n positions of composing a poem to a given tune of ci by the highest contamination of rate successively;Institute's predicate and position of composing a poem to a given tune of ci are right
It should refer to the word itself or associated word that the word is the position of composing a poem to a given tune of ci.
Introduction has been carried out above for the mode that concrete application is implemented, and details are not described herein.
Compared with prior art, the elderly's voice interactive method for wearable electronic in the invention is used
And system, it has the advantages that:
(1) demand for carrying out interactive voice with the elderly is fully taken into account, is adjusted to playing voice, and to control
Voice carries out error correction, ensures the control instruction for correctly getting old user, to execute corresponding task;
(2) it since wearable electronic is typically all to be close to the users what skin used, is examined using temperature sensor
When measuring user temperature, you can to show to have user's use, user fingerprints can be obtained at this time, further carried out to user identity
Verification, and use pattern control is carried out according to the age of different user or use habit;
(3) dictionary and language model are created, error correction more effectively is carried out to the control voice of user, improves voice error correction
Accuracy is more applicable for large-scale promotion application.
In this description, the present invention is described with reference to its specific embodiment.But it is clear that can still make
Various modifications and alterations are without departing from the spirit and scope of the invention.Therefore, the description and the appended drawings should be considered as illustrative
And not restrictive.
Claims (10)
1. a kind of the elderly's voice interactive method for wearable electronic, which is characterized in that include the following steps:
Training dictionary is generated, the training dictionary includes the preset association relationship between multiple words and each word;
Training dictionary and sentence library, train language model are obtained, the language model includes arbitrary three phrases synthesis one
The probability that a sentence occurs;
When temperature sensor in the wearable electronic detects the temperature within the scope of default body temperature, judge
There is user to wear the wearable electronic;
The wearable electronic prompts user to carry out fingerprint authentication by display screen and loud speaker;
After the wearable electronic gets the fingerprint of user, until Cloud Server inquiry user identity and age of user;
If the age of user is in default the elderly's the range of age, the wearable electronic enters the elderly's interaction
Pattern;
Under the elderly's interactive mode, the speed for playing voice is slowed down preset multiple by the wearable electronic;
Under the elderly's interactive mode, the volume for playing voice is improved preset multiple by the wearable electronic;
Under the elderly's interactive mode, when the wearable electronic gets the control voice of user, the control to user
Voice processed carries out error correction;
The control voice to user carries out error correction, includes the following steps:
The control voice of user is resolved to multiple words by the wearable electronic, and selection is related to each word from dictionary
Other close words of connection;
Each word and its associated word are combined by the wearable electronic with other words, and combination is put into institute's predicate
It is trained in speech model, obtains the highest contamination of probability of occurrence, and the sequence of the control voice according to user, will occur
Each word is ranked up in the highest contamination of probability, obtains correct control statement;
The wearable electronic executes corresponding task according to the correct control statement.
2. the elderly's voice interactive method according to claim 1 for wearable electronic, which is characterized in that institute
State multiple words in training dictionary and the preset association relationship between each word, including each word and the word respectively with other
The similarity value of word;
Described selection other close words associated with each word from dictionary, including:
The degree of approximation of selection and each word is more than the word of default degree of approximation threshold value, the close word as the word from dictionary.
3. the elderly's voice interactive method according to claim 1 for wearable electronic, which is characterized in that instruction
Practice language model, including arbitrary three morphologies are combined at one, and calculates the probability of combination appearance;
The probability for calculating the combination and occurring, includes the following steps:
It calculates and occurs the sentence number of the combination in the sentence library;
Total sentence number in the sentence number of the combination divided by the sentence library to occur, obtains the probability of combination appearance.
4. the elderly's voice interactive method according to claim 1 for wearable electronic, which is characterized in that institute
It states and is combined each word and its associated word with other words, include the following steps:
Determine that the number of total word included by the control voice is n;
Multiple combinations are obtained from extraction in the associated word of n word and each word, each combination includes each word in n word
Word itself or an associated word.
5. the elderly's voice interactive method according to claim 4 for wearable electronic, which is characterized in that institute
It states combination being put into the language model and be trained, obtain the highest contamination of probability of occurrence, include the following steps:
N word in each combination is divided into multiple three lexons combinations, each sub-portfolio is searched from the language model and is gone out
Existing probability, and the probability that each sub-portfolio is found is multiplied, obtain the probability of combination appearance;
The probability that relatively more each combination occurs, selects the highest contamination of probability of occurrence.
6. the elderly's voice interactive method according to claim 5 for wearable electronic, which is characterized in that institute
Each word in the highest contamination of probability of occurrence is ranked up by the sequence for stating the control voice according to user, is obtained correct
Control statement, include the following steps:
N positions of composing a poem to a given tune of ci are generated in correct control statement, select the highest contamination of probability of occurrence, it successively will be in combination
Each word is filled into corresponding n positions of composing a poem to a given tune of ci;Institute's predicate is corresponding with position of composing a poem to a given tune of ci to refer to that the word is the word of the position of composing a poem to a given tune of ci itself
Or associated word.
7. a kind of the elderly's voice interactive system of wearable electronic, which is characterized in that including:
Language model training module, for generating training dictionary, the training dictionary include multiple words and each word it
Between preset association relationship;And training dictionary and sentence library are obtained, train language model, the language model includes appointing
Three phrases of meaning synthesize the probability that a sentence occurs;
Human detection module, when for detecting the temperature being within the scope of default body temperature, judgement has described in user's wearing
Wearable electronic;
User authentication module, for prompting user to carry out fingerprint authentication by display screen and loud speaker;And get use
After the fingerprint at family, until Cloud Server inquiry user identity and age of user;
The elderly's interactive mode control module is preset if the age for user is in the elderly's the range of age, described
Wearable electronic enters the elderly's interactive mode;The speed for playing voice is slowed down preset multiple by control;Control will play
The volume of voice improves preset multiple;And when getting the control voice of user, error correction is carried out to the control voice of user;
It is described that error correction is carried out to control voice, it is carried out according to following steps:
The control voice of user is resolved to multiple words by the elderly's interactive mode control module, and from dictionary selection with it is each
Other associated close words of a word;
Each word and its associated word are combined by the elderly's interactive mode control module with other words, and combination is put
Enter and be trained in the language model, obtains the highest contamination of probability of occurrence, and according to the suitable of the control voice of user
Each word in the highest contamination of probability of occurrence is ranked up, obtains correct control statement by sequence;
Control instruction execution module, the correct control statement for being exported according to the elderly's interactive mode control module are held
The corresponding task of row.
8. the elderly's voice interactive system of wearable electronic according to claim 7, which is characterized in that described old
Year people's interactive mode control module by each word and its be associated word and be combined with other words, executed according to following steps:
The elderly's interactive mode control module determines that the number of total word included by the control voice is n;
The elderly's interactive mode control module is extracted from the associated word of n word and each word and obtains multiple combinations, often
A combination includes the word itself or an associated word of each word in n word.
9. the elderly's voice interactive system of wearable electronic according to claim 8, which is characterized in that described old
Combination is put into the language model by year people's interactive mode control module to be trained, and the group of the highest word of probability of occurrence is obtained
It closes, includes the following steps:
N word in each combination is divided into multiple three lexons and combined by the elderly's interactive mode control module, from described
The probability that each sub-portfolio occurs is searched in language model, and the probability that each sub-portfolio is found is multiplied, and is somebody's turn to do
Combine the probability occurred;
The elderly's interactive mode control module more each combines the probability of appearance, selects the group of the highest word of probability of occurrence
It closes.
10. the elderly's voice interactive system of wearable electronic according to claim 9, which is characterized in that described
According to the sequence of the control voice of user, each word in the highest contamination of probability of occurrence is ranked up, is obtained correctly
Control statement includes the following steps:
The elderly's interactive mode control module generates n positions of composing a poem to a given tune of ci in correct control statement, selects probability of occurrence most
Each word in combination is filled into corresponding n positions of composing a poem to a given tune of ci by high contamination successively;Institute's predicate is corresponding with position of composing a poem to a given tune of ci to be referred to
Be the word be the position of composing a poem to a given tune of ci word itself or associated word.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810460399.XA CN108647346B (en) | 2018-05-15 | 2018-05-15 | Old people voice interaction method and system for wearable electronic equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810460399.XA CN108647346B (en) | 2018-05-15 | 2018-05-15 | Old people voice interaction method and system for wearable electronic equipment |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108647346A true CN108647346A (en) | 2018-10-12 |
CN108647346B CN108647346B (en) | 2021-10-29 |
Family
ID=63755521
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810460399.XA Active CN108647346B (en) | 2018-05-15 | 2018-05-15 | Old people voice interaction method and system for wearable electronic equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108647346B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109360554A (en) * | 2018-12-10 | 2019-02-19 | 广东潮庭集团有限公司 | A kind of language identification method based on language deep neural network |
CN110880316A (en) * | 2019-10-16 | 2020-03-13 | 苏宁云计算有限公司 | Audio output method and system |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20080270133A1 (en) * | 2007-04-24 | 2008-10-30 | Microsoft Corporation | Speech model refinement with transcription error detection |
CN101655837A (en) * | 2009-09-08 | 2010-02-24 | 北京邮电大学 | Method for detecting and correcting error on text after voice recognition |
CN104681023A (en) * | 2015-02-15 | 2015-06-03 | 联想(北京)有限公司 | Information processing method and electronic equipment |
CN105957525A (en) * | 2016-04-26 | 2016-09-21 | 珠海市魅族科技有限公司 | Interactive method of a voice assistant and user equipment |
CN106598939A (en) * | 2016-10-21 | 2017-04-26 | 北京三快在线科技有限公司 | Method and device for text error correction, server and storage medium |
CN106710592A (en) * | 2016-12-29 | 2017-05-24 | 北京奇虎科技有限公司 | Speech recognition error correction method and speech recognition error correction device used for intelligent hardware equipment |
CN107122346A (en) * | 2016-12-28 | 2017-09-01 | 平安科技(深圳)有限公司 | The error correction method and device of a kind of read statement |
CN107977356A (en) * | 2017-11-21 | 2018-05-01 | 新疆科大讯飞信息科技有限责任公司 | Method and device for correcting recognized text |
-
2018
- 2018-05-15 CN CN201810460399.XA patent/CN108647346B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20080270133A1 (en) * | 2007-04-24 | 2008-10-30 | Microsoft Corporation | Speech model refinement with transcription error detection |
CN101655837A (en) * | 2009-09-08 | 2010-02-24 | 北京邮电大学 | Method for detecting and correcting error on text after voice recognition |
CN104681023A (en) * | 2015-02-15 | 2015-06-03 | 联想(北京)有限公司 | Information processing method and electronic equipment |
CN105957525A (en) * | 2016-04-26 | 2016-09-21 | 珠海市魅族科技有限公司 | Interactive method of a voice assistant and user equipment |
CN106598939A (en) * | 2016-10-21 | 2017-04-26 | 北京三快在线科技有限公司 | Method and device for text error correction, server and storage medium |
CN107122346A (en) * | 2016-12-28 | 2017-09-01 | 平安科技(深圳)有限公司 | The error correction method and device of a kind of read statement |
CN106710592A (en) * | 2016-12-29 | 2017-05-24 | 北京奇虎科技有限公司 | Speech recognition error correction method and speech recognition error correction device used for intelligent hardware equipment |
CN107977356A (en) * | 2017-11-21 | 2018-05-01 | 新疆科大讯飞信息科技有限责任公司 | Method and device for correcting recognized text |
Non-Patent Citations (1)
Title |
---|
张锋 等: "语音识别在英语发音纠错中的应用", 《计算机应用与软件》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109360554A (en) * | 2018-12-10 | 2019-02-19 | 广东潮庭集团有限公司 | A kind of language identification method based on language deep neural network |
CN110880316A (en) * | 2019-10-16 | 2020-03-13 | 苏宁云计算有限公司 | Audio output method and system |
Also Published As
Publication number | Publication date |
---|---|
CN108647346B (en) | 2021-10-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Tahon et al. | Towards a small set of robust acoustic features for emotion recognition: challenges | |
US11322155B2 (en) | Method and apparatus for establishing voiceprint model, computer device, and storage medium | |
CN112750465B (en) | Cloud language ability evaluation system and wearable recording terminal | |
US9177558B2 (en) | Systems and methods for assessment of non-native spontaneous speech | |
US10020007B2 (en) | Conversation analysis device, conversation analysis method, and program | |
US20210272571A1 (en) | Systems and methods for audio processing | |
US20110213610A1 (en) | Processor Implemented Systems and Methods for Measuring Syntactic Complexity on Spontaneous Non-Native Speech Data by Using Structural Event Detection | |
CN109192194A (en) | Voice data mask method, device, computer equipment and storage medium | |
JP7230806B2 (en) | Information processing device and information processing method | |
Levitan et al. | Combining Acoustic-Prosodic, Lexical, and Phonotactic Features for Automatic Deception Detection. | |
Quintas et al. | Automatic Prediction of Speech Intelligibility Based on X-Vectors in the Context of Head and Neck Cancer. | |
US9691389B2 (en) | Spoken word generation method and system for speech recognition and computer readable medium thereof | |
US20200090681A1 (en) | Mental health diagnostics using audio data | |
CN110136721A (en) | A kind of scoring generation method, device, storage medium and electronic equipment | |
Benkı́ | Quantitative evaluation of lexical status, word frequency, and neighborhood density as context effects in spoken word recognition | |
CN114121006A (en) | Image output method, device, equipment and storage medium of virtual character | |
CN116343824A (en) | Comprehensive evaluation and solution method, system, device and medium for talent expression capability | |
CN110503941B (en) | Language ability evaluation method, device, system, computer equipment and storage medium | |
Wang et al. | Speech emotion verification using emotion variance modeling and discriminant scale-frequency maps | |
CN108647346A (en) | A kind of the elderly's voice interactive method and system for wearable electronic | |
US20190103110A1 (en) | Information processing device, information processing method, and program | |
US20240023858A1 (en) | Systems and methods for human-machine partnered ptsd prediction | |
Chen et al. | Automated empathy detection for oncology encounters | |
CN113593523A (en) | Speech detection method and device based on artificial intelligence and electronic equipment | |
AU2021306718B2 (en) | System to confirm identity of candidates |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right | ||
TR01 | Transfer of patent right |
Effective date of registration: 20230414 Address after: Rooms 604 and 605, Dingsheng Ginza Business Building, No. 2518 Huayuan Road, Songling Town, Wujiang District, Suzhou City, Jiangsu Province, 215200 Patentee after: Suzhou Jiujiu Chunhui Pension Service Co.,Ltd. Address before: 215000 east side of Chang'an Road, Wujiang Economic and Technological Development Zone, Suzhou City, Jiangsu Province Patentee before: SUZHOU DONGWEI NETWORK TECHNOLOGY Co.,Ltd. |