CN106919559A - Machine translation method and machine translation system - Google Patents

Machine translation method and machine translation system Download PDF

Info

Publication number
CN106919559A
CN106919559A CN201610903217.2A CN201610903217A CN106919559A CN 106919559 A CN106919559 A CN 106919559A CN 201610903217 A CN201610903217 A CN 201610903217A CN 106919559 A CN106919559 A CN 106919559A
Authority
CN
China
Prior art keywords
information
text information
translation
particular text
text
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
Application number
CN201610903217.2A
Other languages
Chinese (zh)
Inventor
持田哲司
秦务
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Panasonic Intellectual Property Management Co Ltd
Original Assignee
Panasonic Intellectual Property Management Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Panasonic Intellectual Property Management Co Ltd filed Critical Panasonic Intellectual Property Management Co Ltd
Publication of CN106919559A publication Critical patent/CN106919559A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/40—Processing or translation of natural language
    • G06F40/58—Use of machine translation, e.g. for multi-lingual retrieval, for server-side translation for client devices or for real-time translation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/40—Processing or translation of natural language
    • G06F40/55—Rule-based translation

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Machine Translation (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

In order to improve the function of machine translation system, it is necessary to further improve.In consideration of it, the disclosure provides machine translation method and machine translation system.The machine translation method includes：Acquisition step (S1), text message, the input voice that the first speech data expression first language is said before the translation that the first speech data is converted to text message and generated by acquirement；Determination step (S2), whether text message contains the first particular text information for being stored in storage part before judging the translation obtained in acquisition step (S1)；And output step (S3), in the case of being judged to that translating preceding text message contains the first particular text information in determination step (S2), output is stored at least one party in storage part and the second particular text information and the second speech data of the second particular text information that are associated with the first particular text information.

Description

Machine translation method and machine translation system
Technical field
The present invention relates to machine translation method and machine translation system.
Background technology
In recent years, turned over so as to translate into machine that is multi-lingual and carrying out voice output by that can be spoken with mother tongue The system of translating attracts attention.This machine translation system is expected to break the aphasis in the world, undertakes the promotion to globalization exchange.
In addition, in recent years, the scope of application of this machine translation system from towards individual using expanding to towards commercial profit With used as the media of communication with the visiting foreigner, the utilization in communal facility, shop etc. is investigated.
In the case where being incorporated into this machine translation system towards commercial affairs utilization, purposes is utilized by each, there is frequency It is required speak text or the dialogue text translated numerously.Thus, for example in patent document 1, disclose make continually by It is required that the technology that the article (sentence) translated is associated with code-phase of passing on a message and registered in advance.Thus, user is by profit Used time specifies desired code of passing on a message, and can recall the article associated with code-phase of passing on a message.
Citation
Patent document 1：Japanese Unexamined Patent Publication 9-139969 publications
The content of the invention
Invent problem to be solved
However, the technology disclosed in patent document 1 can not generally mitigate speaker to be continually required to be turned over The burden of translate, content aspect such as different text of speaking according to the utilization purposes in commercial affairs.
Accordingly, with respect to the text of speaking for being continually required to be translated, in order to mitigate the burden of speaker, and realization makes The function that the translation time for speaking text shortens this machine translation system is improved, it is necessary to further function improves.
The present invention be in view of said circumstances and make, can further improve machine translation system its object is to provide Function machine translation method and machine translation system.
Technical teaching for solving the problem was
To achieve these goals, during the machine translation method that a mode of the disclosure is related to is machine translation system Machine translation method, in the storage part of the machine translation system, be stored with the first particular text information in association, and the At least one party in the second speech data of two particular text information and the second particular text information, first particular text Certain words or specific text that information expression is expressed with first language, the second particular text information represent pre-prepd The word of the second language expression of the language different from the first language or the sizing text of sentence are used as, described second is special Text message is determined with the certain words or the specific text without translation equivalence (translation Equivalence), the machine translation method includes：Acquisition step, is obtained the first speech data is converted to text message and Text message before the translation of generation, first speech data represents the input voice said with the first language；Judge step Suddenly, judge before the translation obtained in the acquisition step text message whether containing being stored in described in the storage part First particular text information；And output step, text message contains before the translation is judged in the determination step In the case of the first particular text information, output be stored in the storage part and with the first particular text information phase At least one party in the second particular text information of association and the second speech data of the second particular text information.
In addition, to achieve these goals, the machine translation system that a mode of the disclosure is related to possesses：Storage part, Be stored with the first particular text information in association, and the second particular text information and the second particular text information the second language At least one party in sound data, the certain words or specific that the first particular text information expression is expressed with first language Text, the second particular text information represents the pre-prepd second language for being used as the language different from the first language The word of expression or the sizing text of sentence, the second particular text information is with the certain words or the specific text not With translation equivalence；Obtaining section, text message before the translation that the first speech data is converted to text message and generated by acquirement, First speech data represents the input voice said with the first language；Detection unit, judgement takes in the obtaining section Whether text message contains the first particular text information for being stored in the storage part before the translation for obtaining；And output Portion, in the case where text message before being judged to the translation by the detection unit contains the first particular text information, Output be stored in the storage part and the second particular text information for being associated with the first particular text information and At least one party in the second speech data of the second particular text information.
Additionally, these master or specific technical schemes both can be by system, method, integrated circuit, computer The recording mediums such as the CD-ROM of program or embodied on computer readable are realized, it is also possible to by system, method, integrated circuit, calculating Any combination of machine program and recording medium is realized.
Invention effect
According to the machine translation method and machine translation system of the disclosure, raising machine translation system has been can be implemented as Function further improvement.
Brief description of the drawings
Fig. 1 is the figure of of the functional structure for representing existing machine translation system.
Fig. 2A is the figure of the overall picture for representing the service that the information management system in the disclosure is provided.
Fig. 2 B are the figures of the variation of the part-structure for representing the information providing system in the disclosure.
Fig. 2 C are the figures of the variation of the part-structure for representing the information providing system in the disclosure.
Fig. 3 is the block diagram of of the structure for representing the machine translation system in implementation method 1.
Fig. 4 is the block diagram of of the detailed construction for representing the translation determination processing unit shown in Fig. 3.
Fig. 5 is the block diagram of the another way of the structure for representing the machine translation system in implementation method 1.
Fig. 6 is the flow chart of the work summary for representing the machine translation system in implementation method 1.
Fig. 7 is the flow chart of the concrete mode of the work for representing the machine translation system in implementation method 1.
Fig. 8 is the figure for representing the specific text for being stored in storage part in association and literary one that shapes in implementation method 1.
Fig. 9 is the figure of of the usage scenario for representing the machine translation system for possessing display.
Figure 10 is the figure of of the structure for representing the machine translation system that the variation of implementation method 1 is related to.
Figure 11 is the figure of of the structure for representing the information terminal that the variation of implementation method 1 is related to.
Figure 12 is the timing diagram of the work of the machine translation system in the variation for represent implementation method 1.
Figure 13 is the timing diagram of another of the work of the machine translation system in the variation for represent implementation method 1.
Figure 14 is the flow chart of the concrete mode of the work for representing the machine translation system in implementation method 2.
Figure 15 is the flow chart of the concrete mode of the work for representing the machine translation system in implementation method 3.
Figure 16 is to represent in implementation method 3 to be stored in being associated with certain words and order of speaking for storage part in association Sizing text the figure of.
Figure 17 is that the information management system in the Class1 (our company data-centric cloud service) of the service that represents is provided The figure of the overall picture of service.
Figure 18 is the service that the information management system in the type 2 (IaaS utilizes type cloud service) of the service that represents is provided The figure of overall picture.
Figure 19 is the service that the information management system in the type 3 (PaaS utilizes type cloud service) of the service that represents is provided The figure of overall picture.
Figure 20 is the service that the information management system in the type 4 (SaaS utilizes type cloud service) of the service that represents is provided The figure of overall picture.
Label declaration
10、10A、90：Machine translation system
11、91：Voice input section
12、92：Speech recognition section
13、93：Translation Service
14、94：Phonetic synthesis portion
15、95：Voice output portion
16、16A：Translation determination processing unit
17、23：Display
20：Information terminal
21：Communication unit
22、163：Storage part
30：Communication unit
41、42、43、44：Server
50：Speaker (speaker)
51：Hearer (listener)
110：Voice recognition processing portion
130：Translation processing unit
140：Phonetic synthesis processing unit
161：Obtaining section
162、162A：Detection unit
164：Output section
Specific embodiment
(turning into the opinion on basis of the invention)
Machine translation was come across before and after nineteen ninety.However, the actual conditions of the translation precision originally of machine translation are, English Day translation up to 60% or so, day English translation up to 50% or so.Accordingly, it would be desirable to the artificial phase to being produced by machine translation When the mistake of quantity is corrected, technology of the complete machine translation in being at that time dream.However, in recent years, by by deeply The machine learning techniques of the height such as degree study (Deep Learning) are used for machine translation, make translation precision obtain drastically carrying It is high.Therefore, machine translation in PC applications, Web applications, smart mobile phone application etc. as the translation system of simple possible gradually It is utilized.
On the other hand, the development of the speech recognition various technologies that also reason statistical method brings, realizes identification The raising of rate.Therefore, speech recognition is applied not only to the content text of speaking of user, but also conduct is realized by speaking Equipment control etc., the speech control interface of electric equipment and be gradually utilized.
Under this background, as the aphasis for breaking the world, the technology of the promotion to globalization exchange is undertaken, in recent years Come, spoken with mother tongue, translate into the multi-lingual machine translation system for carrying out voice output and attract attention.
Fig. 1 is the figure of of the functional structure for representing existing machine translation system.Machine translation system shown in Fig. 1 90 possess voice input section 91, speech recognition section 92, Translation Service 93, phonetic synthesis portion 94 and voice output portion 95.
Voice input section 91 is transfused to the voice that speaker is said with first language.The language that voice input section 91 will be input into Sound is converted to speech data and exports to speech recognition section 92.The speech data that speech recognition section 92 pairs is received performs voice to be known Other places are managed, the treatment that speech data is changed in execution to the text data of first language.Speech recognition section 92 will be changed out First language text data output to Translation Service.The text data of 93 pairs of first languages for receiving of Translation Service is performed to The translation treatment that two language are translated, generates the text data of second language.The second language that Translation Service 93 will be generated Phonetic synthesis portion 94 is arrived in text data output.The text data of the second language that phonetic synthesis portion 94 will receive is converted to second The speech data of language, and export to voice output portion 95.The speech data of the second language that voice output portion 95 will receive Exported (say) as the voice of second language.
So, the content of speaking of 90 couples of speakers spoken with first language of machine translation system is received, and is translated Into after second language, voice output is carried out to hearer as the voice of second language.Thereby, it is possible to realize using language different Speaker between exchange.
This machine translation system is used in tourist famous-city, SNS (social networking service) of individual etc., mainly quilt Individual is utilized, but with precision of identifying speech in recent years, translate the raising of precision, it is exchanged as with the visiting foreigner Instrument, the utilization towards commercial affairs in communal facility, shop etc. is also gradually studied.
However, in the case where machine translation system being incorporated into towards commercial affairs utilization, it is desirable to which specific surface is to individual using more Plus accurate translation, translation precision i.e. higher.On the other hand, for example in hotel and/or travel companies, communication, inquire The aspect such as utilize in the interior commercial affairs such as place, medical field, shop, purposes utilized by each, exist exclusive session, word, The special dialogue text such as expression.And, as long as there is machine translation system does not use machine learning techniques etc. for example in advance to spy Different dialogue text is learnt, and just cannot correctly carry out speech recognition and machine translation this problem.
Furthermore, according to the utilization purposes in commercial affairs, there is speak text or the dialogue text for being continually required to be translated. That is, there is speaker (ISP) continually being said to the other side (service acceptor) that mother tongue is different language Words or the sentence (speak text or dialogue text) talked with it.Therefore, in order that machine translation system is by these sentence translations Giving other side (service acceptor), speaker (ISP) needs repeatedly to say these sentences with identical content.This turns into says The burden of words person (ISP).
Especially in the case where the length of these sentences is long, not only can further increase the burden of speaker, but also Can cause in machine translation system, it is impossible to disposably correctly carry out speech recognition and the machine translation of these sentences.That is, In the case where the length of these sentences is long, it will cause the noise to a large amount of environments of speaking of machine translation system input, Therefore, also result in the voice recognition processing of these sentences (speak text or dialogue text), the Frequency of misrecognition increases Greatly, the translation text of output error.As a result, speaker has to speak again again, therefore, the further of speaker can be turned into Burden.
As the countermeasure to this problem, for example Patent Document 1 discloses：The message of message of being passed on a message in making sizing In data sink, make continually to be required that the article translated is associated with code-phase of passing on a message and registered in advance.
More specifically, in patent document 1, the word or article to being frequently utilized have been respectively allocated and have been passed on a message Corresponding table obtained from code (No) is pre-stored in sizing and passes on a message in memory (33).When connecing for being received in receiving circuit (22) Receive in data containing passing on a message in the case of code, based on the corresponding table that sizing is passed on a message in memory (33) is stored in, extracts and receive Contained code of passing on a message is corresponding in data recalls content (word, article).And, enter by with the code of passing on a message received contained by data Row is replaced, and generation receives the message of signal.So, speaker specifies desired code of passing on a message by when utilizing, and can recall The article associated with code-phase of passing on a message.Thereby, it is possible to only easily produce message long by using code is passed on a message, use can be mitigated The burden at family.
However, in the technology disclosed in patent document 1, the word or article that are registered in advance in corresponding table and conduct Pass on a message code nonsensical numerical value as number be associated.Therefore, user needs to remember word or article in correspondence Which it is associated with number in table, the burden weight of user.Especially, the word or article of registration in advance are more, and user's is negative Load can turn into bigger problem.
That is, in the technology disclosed in patent document 1, being continually required to be turned over to mitigate speaker According to the utilization purposes in commercial affairs, the technical task of the burden of the aspect such as different text of speaking is not solved translate, content Certainly.
Therefore, mitigate speaker for the burden of text of speaking that is continually required to be translated and contract to improve The function of the machine translation system that the translation time of the short text of speaking shortens is, it is necessary to further function improves.
To achieve these goals, the machine translation method that a technical scheme of the disclosure is related to is machine translation system In machine translation method, in the storage part of the machine translation system, be stored with the first particular text information in association, With at least one party in the second speech data of the second particular text information and the second particular text information, described first is specific Certain words or specific text that text message expression is expressed with first language, the second particular text information represent advance standard The word of the standby second language expression for being used as the language different from the first language or the sizing text of sentence, described the Two particular text information do not have translation equivalence, the machine translation method bag with the certain words or the specific text Include：Acquisition step, text message, first language before the translation that the first speech data is converted to text message and generated by acquirement Sound data represent the input voice said with the first language；Determination step, judges the institute obtained in the acquisition step Whether text message contains the first particular text information for being stored in the storage part before stating translation；And output step, It is defeated in the case that text message contains the first particular text information before being judged to the translation in the determination step Go out to be stored in the storage part and the second particular text information for being associated with the first particular text information and should At least one party in the second speech data of the second particular text information.
Thus, speaker does not say the full text of the text of speaking of usage frequency first language high, and only says comprising the The certain words of one language or specific text (the first particular text information), it becomes possible to the second language that output is continually utilized Sizing literary (the second particular text information) and its a side of speech data, therefore, the burden of speaking of speaker can be mitigated.In addition, Second language sizing text (the second particular text information) and its speech data with the second language is briefly described with first language The word of sizing text of speech etc. is associated.That is, the second particular text information and its speech data of implementation method 1 are not It is associated with nonsensical numeral etc..Thus, it is also possible to mitigate user and have to remember the negative of relevance in advance or separately Load.
In addition, for example, it is also possible to be, in the determination step, judge before the translation text message whether with storage It is consistent in the first particular text information of the storage part, in the output step, sentence when in the determination step Be set to text message before the translation it is consistent with the first particular text information in the case of, output is stored in the storage part And the second particular text information for being associated with the first particular text information and the second particular text information At least one party in the second speech data.
Thereby, it is possible to only in the content (text message before translation) for speaking literary of speaker and the first particular text information one In the case of cause, export a side of the second particular text information and its speech data.
I.e. it is capable in the case where speaker has only said the first particular text information, make the second particular text One side of information and its speech data exports as translation result.On the other hand, can have been said in speaker special comprising first In the case of determining the words and phrases beyond text message, translation text is (after translation obtained from making to translate described content in itself Text message) output.Thus, speaker can be subject to profit using the first particular text information as word or article in dialogue With.
In addition, for example, it is also possible to being in the storage part, the second particular text information to be distinguished in association Be stored with more than two first particular text information and order information, and the order information represents the institute more than described two The order that the first particular text information should occur in the text respectively is stated, in the determination step, is judged in the acquirement step Before the translation obtained in rapid text message whether containing be stored in the storage part it is described two more than it is first specific Whether according to the sequence text message and the order represented by information contain the first particular text letter more than described two Breath, in the output step, before the translation is judged in the determination step text message contain be stored in it is described Storage part it is described two more than the first particular text information and be order according to the sequence represented by information containing In the case of stating more than two first particular text information, export and described two the first particular text information above And the second particular text information and second voice of the second particular text information of the order information association At least one party in data.
Thereby, it is possible to appear in the content of the text of speaking of speaker in a predetermined order in multiple first particular text information In the case of in (translation before text message), make the second particular text information being associated with multiple first particular text information and The one side output of its speech data.That is, speaker can be properly selected according to tongue be make with comprising first The one side output of the speak corresponding second particular text information of text and its speech data of particular text information, still makes to described Content translated in itself obtained from translate text output.
In addition, for example, it is also possible to be, in the storage part, the second particular text information described to one, in association Be stored with more than one different described first particular text information, more than one different the first particular text information Contain identical certain words and the different specific text of expression respectively.
Thereby, it is possible to the multiple expression ways and/or different multiple different to a second particular text information setting Keyword.Thus, in the case where user wants the second particular text information of output, can by multiple expression ways and/or Some in multiple keywords makes the second particular text information output.
In addition, for example, it is also possible to be, in the output step, when being judged to the translation in the determination step Not comprising in the case of the first particular text information, text message before the translation is translated into institute to preceding text message by output State text message after translation obtained from second language.
Thus, in the case that text message contains the first particular text information before the translation, omitting will text envelope before translation The translation translated to second language is ceased to process, and text message does not include the situation of the first particular text information before translation Under, perform text message before translating and processed to the translation that second language is translated.It is high thereby, it is possible to be directed to usage frequency Speak or exclusive session, word, expression way, realize translation reduction in processing time.
In addition, for example, it is also possible to be, in the output step, when being judged to the translation in the determination step In the case that preceding text message contains the first particular text information for being stored in the storage part, after the translation is not exported Text message.
So, in the case that text message contains the first particular text information before the translation, not performing will text before translation Information is processed to the translation that second language is translated.Thereby, it is possible to simplify the translation treatment in machine translation processing system, and And, translation disposal ability can be processed as other, the function therefore, it is possible to seek machine translation system is improved.
In addition, for example, it is also possible to be also make the 3rd particular text information at least with the first particular text information and institute State the second particular text information in the second particular text information and be stored in the storage part in association, the described 3rd is special It is that the second particular text information is translated into particular text information obtained from the first language to determine text message, in institute State in output step, exporting the second speech data of the second particular text information and the second particular text information In at least one party in the case of, also export the 3rd particular text information.
So, in sizing text of the output as the second language of the second particular text information, also export and second language The article of the sizing first language of the text in translation relation of speech.Thus, speaker is able to confirm that based on speaking described in oneself Text can convey to the information of what content of hearer.
In addition, for example, it is also possible to being that the 3rd particular text presentation of information exported in the output step is in aobvious Show device.
Thus, speaker can visually confirm that hearer can be conveyed in what based on the text of speaking described in oneself The information of appearance.
In addition, for example, it is also possible to being that the machine translation system is by predetermined communication unit and the letter with display Breath terminal connection, in the output step, by the predetermined communication unit, the second particular text information and will be somebody's turn to do Described information terminal is arrived at least one party's output in the second speech data of the second particular text information.
It is special when outputing described second in the output step in described information terminal in addition, for example, it is also possible to be In the case of determining text message, phonetic synthesis treatment is carried out to the second particular text information, generate second speech data, it is right Voice represented by the second speech data of generation is exported.
In addition, for example, it is also possible to being that the machine translation method is used to say the people of the first language and says described second In predetermined state between the people of language.
In addition, to achieve these goals, the machine translation system that a technical scheme of the disclosure is related to possesses：Storage Portion, be stored with the first particular text information in association, and the second particular text information and the second particular text information the At least one party in two speech datas, certain words or spy that the first particular text information expression is expressed with first language Fixed text, the second particular text information represents pre-prepd the second language for being used as the language different from the first language Say the word of expression or the sizing text of sentence, the second particular text information and the certain words or the specific text Without translation equivalence；Obtaining section, text envelope before the translation that the first speech data is converted to text message and generated by acquirement Breath, first speech data represents the input voice said with the first language；Detection unit, judges in the obtaining section Whether text message contains the first particular text information for being stored in the storage part before the translation for obtaining；And it is defeated Go out portion, text message contains the situation of the first particular text information before the translation is judged to by the detection unit Under, output is stored in the storage part and the second particular text information being associated with the first particular text information With at least one party in the second speech data of the second particular text information.
Additionally, these master or specific technical schemes both can be by system, method, integrated circuit, computer The recording mediums such as the CD-ROM of program or embodied on computer readable are realized, it is also possible to by system, method, integrated circuit, calculating Any combination of machine program and recording medium is realized.
Hereinafter, referring to the drawings, machine translation method that specifically a technical scheme of this disclosure is related to etc. is said It is bright.Additionally, embodiments described below represents a concrete example of the invention.The number for representing in the following embodiments Value, shape, inscape, step, order of step etc. are one, it is not intended that limit the present invention.In addition, for following reality Inscape in the inscape in mode, not being documented in the independent claims for representing upper concept is applied, is made For arbitrary inscape is illustrated.In addition, in all of implementation method, it is also possible to combine respective content.
(overall picture of the service of offer)
First, the overall picture of the service for being provided as information management system the machine translation system in present embodiment enters Row explanation.
Fig. 2A is the figure of the overall picture for representing the service that the information management system in the disclosure is provided.Fig. 2 B are to represent this public affairs The figure of the variation of the part-structure of the information providing system in opening.Fig. 2 C are the portions for representing the information providing system in the disclosure The figure of the variation of separation structure.Information management system shown in Fig. 2A possess group 11000, data center's operator 11100 with And service provider 11200.
Group 11000 is, for example, enterprise, group or family etc., and its scale is not limited.Group 11000 possess including the first equipment and The multiple equipment 11010 and home gateway 11020 of the second equipment.Multiple equipment 11010 includes what can be connected with internet Equipment (such as smart mobile phone, personal computer (PC) or television set etc.) and its own equipment that can not be connected with internet (such as illumination, washing machine or refrigerator etc.).Although multiple equipment 11010 can not can also be connected including its own with internet But the equipment that can be connected with internet via home gateway 11020.In addition, user 10100 uses the multiple equipment in group 11000 11010。
Data center's operator 11100 possesses Cloud Server 11110.Cloud Server 11110 be via internet with it is each The virtual server of the equipment of kind cooperation.Cloud Server 11110 is main to being difficult to by treatment such as common database management tools Mass data (big data) etc. is managed.Data center's operator 11100 performs management, the Cloud Server 11110 of data Management and the operations of data center etc. to carrying out these management.Carried out on data center's operator 11100 The details of work, will be described hereinafter.
Here, data center's operator 11100 is not limited to only to carry out management or the Cloud Server 11110 of data The company of management.For example, as shown in Figure 2 B, an equipment system for equipment among developing or having manufactured multiple equipment 11010 In the case of making business and having carried out the management of data or the management of Cloud Server 11110 etc., equipment manufacturers are equivalent in data Heart operator 11100.In addition, data center's operator 11100 is not limited to a company.For example, as shown in Figure 2 C, Equipment manufacturers and management company are common or share the situation of the management for carrying out data or the management of Cloud Server 11110 Under, both sides or one party are equivalent to data center's operator 11100.
Service provider 11200 possesses server 11210.Server mentioned here 11210, its scale is not limited, for example Also memory in PC etc. is used including individual.In addition, also there is service provider 11200 not possess the situation of server 11210.
Additionally, in above-mentioned information management system, home gateway 11020 is not required.For example, in Cloud Server When 11110 carry out all of data management, it is not necessary to home gateway 11020.In addition, also being set just like all in family The standby situation for being all connected to internet is such, in the absence of the situation of its own equipment that can not be connected with internet.
Next, the transmission of the information in the above-mentioned information management system of explanation.
First, each log information is sent to data center's operation by first equipment or the second equipment of group 11000 respectively The Cloud Server 11110 of company 11100.Cloud Server 11110 accumulates log information (Fig. 2A of the first equipment or the second equipment Arrow 11310).Here, log information refers to such as operation conditions or the date time for representing multiple equipment 11010 Deng information.For example, log information includes audiovisual record, the video recording reservation information of video recorder, the washing machine operation day of television set Opening and closing number of times of time phase, the amount of washings, the opening and closing date-time of refrigerator or refrigerator etc., but be not restricted to that these are believed Breath, it is also possible to including the various information that can be obtained from various equipment.Additionally, log information can also be via internet from multiple Equipment 11010 is directly provided to Cloud Server 11110 in itself.In addition, log information can also be temporarily tired from multiple equipment 11010 Product is to home gateway 11020, then is provided to Cloud Server 11110 from home gateway 11020.
Next, the log information that the Cloud Server 11110 of data center's operator 11100 will be accumulated is with certain Unit is supplied to service provider 11200.Here, so-called certain unit, both can be can be to data Center Inter company 11100 information accumulated are arranged unit, or the service provider to be supplied to service provider 11200 Unit required by 11200.In addition, though being set to be provided with certain unit, but certain unit is may not be, Can also be that the information content of offer changes according to situation.Log information is stored in service provider 11200 and is protected as needed In some servers 11210 (arrow 11320 of Fig. 2A).
And, it is the information mutually suitable with the service for providing a user with that service provider 11200 arranges log information, and is carried Supply user.The user for being provided information both can be user 10100, or the outside for using multiple equipment 11010 User 10200.Method as information is provided to user 10100, user 10200, for example can also be from service provider 11200 Directly information (arrow 11330,11340 of Fig. 2A) is provided to user 10100, user 10200.In addition, as to user 10100 The method of offer information, for example can also be again via the Cloud Server 11110 of data center's operator 11100 to user 10100 provide information (arrow 11350,11360 of Fig. 2A).Alternatively, it is also possible to be the cloud of data center's operator 11100 It is the information mutually suitable with the service for providing a user with that server 11110 arranges log information, and is supplied to service provider 11200。
Additionally, user 10100 both can be different, or identical from user 10200.
(implementation method 1)
Hereinafter, the machine translation system in the disclosure is illustrated.
[structure of machine translation system]
Fig. 3 is the block diagram of of the structure for representing the machine translation system 10 in implementation method 1.
Machine translation system 10 is made in predetermined situation between the people for saying the people of first language and saying second language .More specifically, machine translation system 10 as described above for example in hotel and/or travel companies, communication, inquire In the interior commercial affairs such as place, medical field, shop (predetermined situation), it is used as the media of communication with the visiting foreigner.Such as Fig. 3 institutes Show, it is defeated that the machine translation system 10 possesses voice input section 11, speech recognition section 12, Translation Service 13, phonetic synthesis portion 14, voice Go out portion 15 and translation determination processing unit 16.They can also for example be connected by LSI (large scale integrated circuit) internal bus.
The ＞ of ＜ voice input sections 11
Voice input section 11 is transfused to the voice that speaker is said with first language.The language that voice input section 11 will be input into Sound is converted to speech data (being recited as below " speech data before translation ") and exports to speech recognition section 12.In present embodiment In, voice input section 11 is for example made up of microphone.
The ＞ of ＜ speech recognition sections 12
Speech data performs voice recognition processing before translation acquired by speech recognition section 12 pairs, and performing will the preceding voice of translation The voice recognition processing that data are changed to the text data (being recited as below " text message before translation ") of first language.Language Before the translation that sound identification part 12 will be changed out text information output to translation determination processing unit 16.
Additionally, speech recognition section 12 is realized by the computer being made up of CPU, memory etc., speech recognition section 12 can Voice recognition processing is performed with oneself, but is not limited to this.Speech recognition section 12 can also have communication function and memory Function, is communicated by the predetermined communication unit such as internet with cloud (server zone).In this case, speech recognition section 12 Speech data before acquired translation can also be sent to cloud, the voice performed for speech data before the translation is obtained from cloud Result obtained from identifying processing.
The ＞ of ＜ Translation Service 13
Translation Service 13 turns in the case of text message before achieving translation from translation determination processing unit 16 to acquired Translate preceding text message and perform the translation treatment translated to second language, the text data for generating second language (is recorded below It is " text message after translation ").Text information output is to phonetic synthesis portion 14 after the translation that Translation Service 13 will be generated.
In addition, although after setting the translation that Translation Service 13 will be generated text information output to phonetic synthesis portion 14, but not It is limited to this, or, text information output translates determination processing unit to translation determination processing unit 16 after the translation that will be generated 16 are exported to phonetic synthesis portion 14.That is, after the translation that will can also be generated of Translation Service 13 text message via Phonetic synthesis portion 14 is arrived in the output of translation determination processing unit 16.
In addition, Translation Service 13 is realized by the computer being made up of CPU, memory etc., Translation Service 13 can be held with oneself Row translation is processed, but is not limited to this.Translation Service 13 can also have communication function and memory function, by internet etc. Predetermined communication unit is communicated with cloud.
In this case, text message before acquired translation can also be sent to cloud by Translation Service 13, obtained from cloud and held The hand-manipulating of needle is to result obtained from the translation treatment of text message before translation.
The ＞ of ＜ phonetic syntheses portion 14
Phonetic synthesis portion 14 is carried out according to text message generation after translation the after translation is obtained in the case of text message The phonetic synthesis treatment of the speech data (being also recited as below " speech data after translation ") of two language.In addition, phonetic synthesis portion 14 in the case of the second particular text information that obtains, and generated according to the second particular text information the voice number of second language Processed according to the phonetic synthesis of (being also recited as below " speech data after translation ").And, phonetic synthesis portion 14 is by turning over for being generated Translate rear speech data output to voice output portion 15.
Additionally, phonetic synthesis portion 14 is realized by the computer being made up of CPU, memory etc., phonetic synthesis portion 14 can Processed with oneself performing phonetic synthesis, but be not limited to this.Phonetic synthesis portion 14 can also have communication function and memory Function, is communicated by the predetermined communication unit such as internet with cloud, by text data or second after acquired translation Particular text information is sent to cloud.And execution can also be obtained for text data or second after transmitted translation from cloud Result obtained from the voice recognition processing of particular text information.
The ＞ of ＜ voice outputs portion 15
Voice output portion 15 enter the second language for being about to receive speech data carried out as the voice of second language it is defeated The voice output treatment for going out (say).In the present embodiment, voice output portion 15 is for example made up of loudspeaker.
＜ translates the ＞ of determination processing unit 16
Fig. 4 is the block diagram of of the detailed construction for representing the translation determination processing unit 16 shown in Fig. 3.
Translation determination processing unit 16 is for example as shown in figure 4, possess obtaining section 161, detection unit 162, storage part 163 and defeated Go out portion 164.
Storage part 163 is stored the first particular text information with the second particular text information in association, and described Certain words or specific text that one particular text information expression is expressed with first language, the second particular text information are represented The word or sentence of the pre-prepd second language expression for being used as the language different from first language are shaped text, second Particular text information does not have translation equivalence with certain words or specific text.Here, the first particular text information is, for example, The short sentence comprising keywords (certain words) such as " cigarette ", " translaters " such as " explanation of cigarette " and/or " explanation of translater " Sub (specific text), but it is also possible to be only keyword (certain words).In addition, so-called without translation equivalence, it is meant that translation It is asymmetrical.Namely it is meant that for " explanation of translater " this specific text, not with the sentence of particular text body And/or comprising particular text body sentence correspondence, but with " welcome you.This is translater.Please with brief English after serge sound Speech.Japanese will be translated into for you." such sizing text correspondence.
In addition it is also possible to be in storage part 163, to a second particular text information, to be stored with association one The first different particular text information of the above, more than one the first different particular text information is specific containing identical respectively Word and the different specific text of expression.That is, the second particular text information can also be from different multiple expression ways And/or different multiple keywords are associated.Thereby, it is possible to by different multiple expression ways and/or different multiple passes In keyword some come recall as sizing text the second particular text information.
Obtaining section 161 obtains text message before the translation that the first speech data is converted to text message and is generated, described The input voice that first speech data expression first language is said.
Whether text message is containing being stored in storage part 163 before the translation that detection unit 162 judges acquired by obtaining section 161 First particular text information.
Detection unit 162 makes output section in the case where being judged to that translating preceding text message contains the first particular text information 164 outputs are stored in storage part 163 and the second particular text information being associated with the first particular text information.Here, sentencing Determine portion 162 in the case where being judged to that translating preceding text message contains the first particular text information for being stored in storage part 163, no Make text message before the translation acquired by the output of output section 164 obtaining section 161.
In addition, detection unit 162 makes in the case where being judged to translate preceding text message not comprising the first particular text information Text message before the output translation of output section 164.
Output section 164 exports text envelope before the second particular text information or translation according to the result of determination of detection unit 162 Breath.In the present embodiment, output section 164 is according to the result of determination of detection unit 162, by the second particular text information output to language Sound combining unit 14, by text information output before translation to Translation Service 13.
Additionally, translation determination processing unit 16 is for example not limited to the example shown in Fig. 4.Hereinafter, the situation is entered using Fig. 5 Row explanation.Fig. 5 is the block diagram of the another way of the structure for representing the machine translation system 10 in implementation method 1.Pair with Fig. 3 and Fig. 4 same key element addition of identical label, and omit detailed description.
Machine translation system 10A shown in Fig. 5 translates determination processing compared with the machine translation system 10 shown in Fig. 3 The structure of portion 16A is different.Determination processing unit 16A is compared with the translation determination processing unit 16 shown in Fig. 4 for translation, does not have Obtaining section 161 and output section 164, and display 17 is added with, and the structure of detection unit 162A is different.
The institute that detection unit 162A possesses above-mentioned detection unit 162 is functional, also comprising obtaining section 161 and output section 164 Function.That is, detection unit 162A obtains text message before the translation that speech recognition section 12 is exported, acquired turning over is judged Translate whether preceding text message contains the first particular text information for being stored in storage part 163.Detection unit 162A is turned in acquired Preceding text message is translated not comprising in the case of the first particular text information, text message before translation is carried out to Translation Service 13 defeated Go out.
On the other hand, detection unit 162A text messages before acquired translation contain and are stored in the first of storage part 163 In the case of particular text information, not to text message before the output translation of Translation Service 13.Detection unit 162A is carried from storage part 163 The second particular text information corresponding with the first particular text information for being judged as being contained in text message before translation is taken, and it is defeated Go out to phonetic synthesis portion 14.
Display 17 can also for example show the second particular text information for being output to phonetic synthesis portion 14.
In addition it is also possible to be also make the 3rd particular text information at least with the first particular text information and the second specific text The second particular text information in this information is stored in storage part 163 in association, and the 3rd particular text information is by Two particular text information translate into particular text information obtained from first language.
In this case, or, detection unit 162A with the first particular text information for determining or will be extracted The corresponding 3rd particular text information output of the second particular text information to display 17, the 3rd particular text information of output It is shown in display 17.Furthermore, it is also possible in the voice number of the second language generated according to the second particular text information According to output before and after, separately output generated according to the 3rd particular text information corresponding with the second particular text information the The speech data of one language.
Thus, speaker can visually or acoustically confirm to be conveyed to based on the text of speaking described in oneself obedient The information of people what content.
[work of machine translation system 10]
Work summary to the machine translation system 10 according to composition as described above is illustrated.
Fig. 6 is the flow chart of the work summary for representing the machine translation system 10 in implementation method 1.
First, machine translation system 10 carries out acquirement treatment, and the acquirement treatment is obtained and for the first speech data to be converted to text This information and text message, the input voice (S1) that the first speech data expression first language is said before the translation that generates.
Next, machine translation system 10 carries out determination processing, the determination processing judges the translation obtained in S1 above Whether this information contains the first particular text information (S2) for being stored in storage part 163.
Next, machine translation system 10 is worked as is judged to that translating preceding text message contains the first particular text information in S2 In the case of, output treatment is carried out, output treatment output is stored in storage part 163 and related to the first particular text information Second particular text information (S3) of connection.
Next, using Fig. 7 and Fig. 8, the concrete mode to the work of machine translation system 10 is illustrated.
Fig. 7 is the flow chart of the concrete mode of the work for representing the machine translation system 10 in implementation method 1.
As shown in fig. 7, first, the identification speaker's word of machine translation system 10, i.e. when voice input section 11 is transfused to (S11 during voice described in speaker：It is), the voice that will enter into voice input section 11 is converted to speech data and defeated before translation Go out to speech recognition section 12.
Next, speech data performs voice recognition processing (S12) before translation acquired by speech recognition section 12 pairs, will turn over Translate preceding speech data and be converted to text message before the translation of first language.Text envelope before the translation that speech recognition section 12 will be changed out Translation determination processing unit 16 is arrived in breath output.
Next, translation determination processing unit 16 carries out determination processing.That is, translation determination processing unit 16 is judged acquired Before translation among text message, if contain the certain words or specific text (S13) that pre-register in storage part 163.
When being judged to translate (S13 in the case that preceding text message contains certain words or specific text in S13：It is), Extraction is stored in storage part 163 and the second particular text information (S14) being associated with the first particular text information.And, Translation the second particular text information output that will extract of determination processing unit 16 is to phonetic synthesis portion 14.Additionally, at translation judgement Reason portion 16 can also be by the 3rd particular text information output corresponding with the second particular text information for extracting to machine translation The display that system 10 has.
On the other hand, when being judged to translate situation of the preceding text message not comprising certain words or specific text in S13 Under (S13：It is no), to text message before the output translation of Translation Service 13.Text message is performed before translation acquired by Translation Service 13 pairs The translation translated to second language processes (S15), text data after generation translation.After the translation that Translation Service 13 will be generated Phonetic synthesis portion 14 is arrived in text message output.
Next, phonetic synthesis portion 14 carries out phonetic synthesis treatment, second spy of the phonetic synthesis treatment according to acquired by Determine text message after text message or translation, generate speech data (S16) after the translation of second language.Phonetic synthesis portion 14 will Voice output portion 15 is arrived in speech data output after the translation for being generated.
Next, voice output portion 15 carries out voice output treatment, the voice output is processed language after acquired translation Sound data are exported (say) (S17) as the voice of second language.
Fig. 8 is to represent the specific text for being stored in storage part 163 in association and literary one that shapes in implementation method 1 Figure.
In fig. 8, as one of the first particular text information, the specific text of Japanese is illustrated, as the second specific text One of this information, illustrates the sizing text of English.In addition, as one of the 3rd particular text information, illustrating English Sizing text translate into obtained by Japanese sizing text translation text.
For example, (being translated when speaker says " Fan Breakdown Machine explanation The ま The with Japanese to machine translation system 10 The explanation of device) " in the case of, in machine translation system 10, by voice recognition processing by the language of the voice described in speaker Sound data are converted to text message before the translation of Japanese.Then, in machine translation system 10, entered by translating determination processing unit 16 Row determination processing, be judged to before translation in text message containing be stored in storage part 163 as it is specific text " turn over Breakdown Machine Illustrate (explanation of translater) ".In this case, in machine translation system 10, Japanese described in speaker is not entered to be about to Article " Fan Breakdown Machine explanations The ま The (carrying out the explanation of translater) " translate into the translation treatment of English.Instead It is that extraction is stored in storage part 163 and illustrates (explanation of translater) with as the " Fan Breakdown Machine of specific text " corresponding English Sizing text " the Welcome.This is a translation machine.Please speak with a short of language English after the sound a beep.It translated into Japanese. (welcome you.This is translation Device.Please talked with brief English after serge sound.Japanese will be translated into for you.) " as the second particular text information, and Phonetic synthesis treatment and voice output treatment are performed, so as to export the voice of the sizing text of above-mentioned English.
Here, explanation uses word " Fan Breakdown Machine (translater) in speaker " say words unlike those described above and turned over by machine Translate the example in the case that the translation of system 10 is processed.For example, saying " Fan Breakdown Machine to machine translation system 10 with Japanese in speaker In the case of は あ Chi ら In Trafficking Sales て お り ま The (translater is on sale thereat) ", in machine translation system 10, pass through The speech data of the voice described in speaker is converted to voice recognition processing text message before the translation of Japanese.Then, in machine In device translation system 10, determination processing is carried out by translating determination processing unit 16, be judged to not include in text message before translation It is stored in the specific text of storage part 163.In this case, in machine translation system 10, enter to be about to the Japanese described in speaker Article " Fan Breakdown Machine は あ Chi ら In Trafficking Sales て お り ま The (translater is on sale thereat) " translate at the translation of English Reason, for example, translating into the article of " We sell a translation machine over there. " this English, and holds The treatment of row phonetic synthesis and voice output treatment, so as to export the voice of above-mentioned english article.
Additionally, speaker has said to machine translation system 10 being also using the situation if word " タ バ Us (cigarette) " Likewise, therefore, in this description will be omitted.
But, in fig. 8, for be stored in storage part 163, comprising the different of same word " タ バ Us (cigarette) " Specific text is " タ バ Us Approximately (limitation of cigarette) " and " タ バ Us illustrate (explanation of cigarette) ", is associated with certain Type text.So, or, it is literary (the second particular text information) for same sizing, it is associated with multiple different specific texts (the One particular text information).Alternatively, it is also possible to for, it is literary (the second particular text information) for same sizing, be associated with multiple differences Certain words (the first particular text information).Thus, in the case where user wants the second particular text information of output, can Second particular text information output is made by some in multiple expression ways and/or multiple keywords.
In addition, as described above, machine translation system 10 can also possess display.In this case, machine translation system 10 are carrying out determination processing and to the hearer output with English as mother tongue are being stored in storage part 163 and corresponding with specific text English sizing text when, sizing text translation literary (the 3rd particular text information) corresponding with sizing text is shown in the display Give the speaker that Japanese is mother tongue.Thus, it is different but associated from the specific text of the first language described in speaker in output Second language sizing text (the second particular text information) voice in the case of, can to speaker notify what be outputed The voice of the sizing text of sample content.Even if in addition, when speaker does not remember the content of sizing text, it is also possible on the spot with sizing text The literary content of sizing of the translation text (the sizing text of first language) to confirm second language, therefore, the dialogue with hearer also can It is enough successfully to carry out.
[effect etc.]
As described above, according to implementation method 1, speaker (speaker) is only said comprising the first language to machine translation system 10 The certain words of speech or it is specific text (the first particular text information), it becomes possible to make by each using purposes exclusive session, Word, expression etc. continually using second language sizing text (the second particular text information) exported.Thus, can mitigate The burden of speaking of speaker.
In addition, second language sizing text (the second particular text information) with the second language is briefly described with first language The word of sizing text of speech etc. is associated.That is, the second particular text information of implementation method 1 not with it is nonsensical Numeral etc. is associated.Thus, it is also possible to mitigate user and have to remember in advance or separately the burden of relevance.
Additionally, as described above, the machine translation system 10 in implementation method 1 can also possess display.Fig. 9 is to represent tool The figure of of the usage scenario (use-case) of the machine translation system 10 of standby display.Speaker 50 is the people for saying first language, Equivalent to ISP.Hearer 51 is the people for saying second language, equivalent to service beneficiaries.Illustrate in fig .9 and speak People 50 is during taking in guests etc. to the multilingual scene of hearer 51.
In this case, machine translation system 10 is carrying out determination processing and to the hearer with second language as mother tongue During corresponding with the first particular text information the second particular text information of 51 voice outputs, it is also possible to make the second particular text information First language translation text, i.e. the 3rd particular text presentation of information is in display.Thereby, it is possible to notify speaker 50 to export Second particular text information of what content.Even if in addition, not remembeing in the second particular text information in speaker 50 Rong Shi, it is also possible to which the 3rd particular text information that translates to first language on the spot is confirmed in the second particular text information Hold, therefore, the dialogue with hearer 51 also can be carried out successfully.
Additionally, in above-mentioned implementation method 1, illustrating to use first language institute to the input speaker of machine translation system 10 The voice said, the situation of the voice of the output second language of machine translation system 10, but it is not limited to this.Can also be, to passing through The voice described in information terminal input speaker's first language that predetermined communication unit is connected with machine translation system, the letter Breath terminal exports the voice of second language.
(variation)
Each composition part of the machine translation system 10 in implementation method 1 shown in illustrated, such as Fig. 3 can also disperse It is present in information terminal and multiple servers.Hereinafter, illustrated by variation of example in this case.
[structure of machine translation system 10A]
Figure 10 is the figure of of the structure for representing the machine translation system 10A that the variation of implementation method 1 is related to.Machine Translation system 10A is connected by predetermined communication unit with the information terminal 20 with display, by predetermined communication unit By the second particular text information output to information terminal 20.As shown in Figure 10, machine translation system 10A possess information terminal 20, Server 41, server 42, server 43 and server 44.Information terminal 20, server 41, server 42, server 43 And server 44 is connected by communication unit 30 respectively.
The ＞ of ＜ communication units 30
Communication unit 30 is, for example, the wired or wireless network that internet is connected to by light path, ADSL etc..In the feelings Under condition, information terminal 20 for example can also be special-purpose terminal, server 41, server 42, server 43 and server 44 It such as can also be Cloud Server.
Additionally, communication unit 30 is not limited to the mobile telephone network of this, or 3G, 4G, 5G circuit etc..In the situation Under, information terminal 20 for example can also be special-purpose terminal, and server 41, server 42, server 43 and server 44 are for example It can also be Cloud Server.
In addition, communication unit 30 can also be Bluetooth, Ibeacon, IrDA, Wi-Fi, TransferJet, specific The short-range radios such as low-power radio.In this case, information terminal 20 be, for example, terminal, server 41, server 42, Server 43 and server 44 for example can also be the Special clothes of dedicated local server and/or indoor (on-premises) Business device.
In addition, communication unit 30 can also be 1 pair of special circuit of N connections.In this case, information terminal 20 is, for example, Terminal, server 41, server 42, server 43 and server 44 for example can also be dedicated local server and/or family Interior private server.
In addition, communication unit 30 can also be DCM (Data Communication Module, data communication module) etc. Fast wireless network.In this case, information terminal 20 for example can also be car-mounted terminal, server 41, server 42, service Device 43 and server 44 for example can also be Cloud Servers.
The ＞ of ＜ information terminals 20
Figure 11 is the figure of of the structure for representing the information terminal 20 that the variation of implementation method 1 is related to.It is pair same with Fig. 3 The key element of sample addition of identical label, and omit detailed description.
Information terminal 20 is transfused to the voice described in speaker's first language, and is based on what is be transfused to hearer output Represented by voice speak text the second particular text information or by this speak text translated to second language obtained from Translate the voice of text.So, information terminal 20 plays the effect as the user interface in machine translation system 10A.
As shown in figure 11, information terminal 20 possesses voice input section 11, voice output portion 15, communication unit 21, storage part 22 And display 23.Additionally, the composition part not necessarily of display 23.Communication unit 21 is by being made up of CPU, memory etc. Computer realizes, between server 41, server 42, server 43 and server 44 enters line number by communication unit 30 According to transmitting-receiving.The storage communication unit 21 of storage part 22 is communicated with server 41, server 42, server 43 and server 44 Data that are acquired or being exported etc..Display 23 for example obtains the second particular text information or the 3rd in communication unit 21 In the case of particular text information, the second particular text information or the 3rd particular text information are shown.
In this variation, information terminal 20 by speaker input represent first language the voice spoken when, by institute The voice of input is converted to speech data before translation, and server 41 is sent to via communication unit 30.Information terminal 20 is from service Device 41 receives text message before the translation of the first language changed out by voice recognition processing.
In addition, text message is transmitted to server 42 before the translation that will be received from server 41 of information terminal 20. And, information terminal 20 from server 42 receive the second particular text information or translation before text message.
Information terminal 20 is sent out in the case of text message before receiving translation from server 42 to server 43 Send.And, information terminal 20 receives text message after translating from server 43.Additionally, text message can also be without before translation By information terminal 20, and directly sent to server 42 from server 41.
In addition, information terminal 20 from server 42 in the case where the second particular text information is received, it is specific by second Text message is transmitted to server 44.On the other hand, the text envelope after translation is received from server 43 of information terminal 20 In the case of breath, text message after translation is transmitted to server 44.In addition, the second particular text information can also be without By information terminal 20, and server 44 is sent directly to by server 42.And, information terminal 20 receives second from server 44 Speech data after the translation of text message after particular text information or translation.
Information terminal 20 receive the second particular text information or translation after text message translation after speech data when, Voice output is carried out in voice output portion 15.
Additionally, information terminal 20 can also have phonetic synthesis processing unit 140.In this case, information terminal 20 also may be used It is raw when the second particular text information is received from server 42, phonetic synthesis treatment is carried out to the second particular text information Into second speech data, and export the voice represented by generated second speech data.
The ＞ of 41~server of ＜ servers 44
Server 41 possesses communication unit (not shown) and voice recognition processing portion 110.Server 41 is directed to from information terminal Speech data before 20 translations for sending, performs voice recognition processing in voice recognition processing portion 110, and performing will the preceding voice of translation The voice recognition processing that data are changed to text message before the translation of first language.In addition, server 41 is by first language Translation before text message be sent to information terminal 20.
Additionally, server 41 directly can also send the preceding text message of translation not via information terminal 20 to server 42.
Server 42 possesses communication unit (not shown) and translation determination processing unit 16A (not shown), communication i.e. (not shown) Portion, detection unit 162 and storage part 163.On detection unit 162 and storage part 163, as described above, therefore, omitting herein Describe in detail.
Text message contains first before server 42 works as the translation for being judged to send in detection unit 162 from information terminal 20 In the case of particular text information, will be stored in storage part 163 and second be associated with the first particular text information is specific Text message is sent to information terminal 20.On the other hand, server 42 is worked as and is judged to translate preceding text message in detection unit 162 In the case of not comprising the first particular text information, text message before translation is sent to information terminal 20.
Additionally, the second particular text information can also be sent directly to service by server 42 not via information terminal 20 Device 44, or text message before translation is sent directly to server 43.
Server 43 possesses communication unit (not shown) and translation processing unit 130.Server 43 is directed to and is sent out from information terminal 20 Text message before the translation sent, the translation translated to second language is performed in processing unit 130 is translated and is processed, generation second Text message after the translation of language.Text message is sent to information terminal 20 after the translation that server 43 will be generated.
Additionally, server 43 can also directly send text message after translating not via information terminal 20 to server 44.
Server 44 possesses communication unit (not shown) and phonetic synthesis processing unit 140.Server 44 is from information terminal 20 Send after translating in the case of text data, text data next life after carrying out according to translation in phonetic synthesis processing unit 140 The phonetic synthesis treatment of speech data after into the translation of second language.In addition, server 44 is sending from information terminal 20 In the case of two particular text information, carry out generating according to the second particular text information in phonetic synthesis processing unit 140 The phonetic synthesis treatment of speech data after the translation of two language.And, the second particular text information that server 44 will be generated Or speech data is sent to information terminal 20 after the translation of text message after translation.
Additionally, in Fig. 10, illustrate machine translation system 10A the dispersion of each composition part be present in information terminal and The example of multiple servers, but it is not limited to this.For example, both each part that constitutes can be scattered in than the server shown in Figure 10 The few server of quantity, it is also possible to each part that constitutes is come together in into a server.
[work of machine translation system 10A]
Next, the work to the machine translation system 10 according to composition as described above is illustrated.Below, figure is used 12, exchanging for the data between information terminal 20 and each server is illustrated.
Figure 12 is the timing diagram of of the work of the machine translation system 10A in the variation for represent implementation method 1.
As shown in figure 12, first, when information terminal 20 represents the voice spoken of first language by speaker's input (S101) speech data is sent to server 41 (S102) via communication unit 30 before, changing translation obtained from the voice.
Next, server 41 performs the voice recognition processing to speech data before the translation that receives, will language before translation Sound data are converted to text message before the translation of first language.And, text message hair before the translation that server 41 will be changed out It is sent to information terminal 20 (S103).
Next, text message is transmitted to server 42 before the translation that will be received from server 41 of information terminal 20 (S104).Server 42 performs determination processing, and text message contains the before the translation for being judged to be received from information terminal 20 In the case of one particular text information, the second particular text information being associated with the first particular text information is sent to information Terminal 20 (S105).Additionally, server 42 before the translation for being judged to be received from information terminal 20 text message not comprising the In the case of one particular text information, as without item is met, text message before translation is sent to information terminal 20 (S105).This Outward, it is also possible to only will be sent to information terminal 20 without item this meaning is met.
Next, information terminal 20 text message or nothing before translation is received from server 42 meet item this meaning In the case of, text message before translation is transmitted (S106) to server 43.Then, server 43 performs translation treatment, According to text message before the translation received from information terminal 20 come text message after the translation for generating second language.And, clothes Text message is sent to information terminal 20 (S107) after the translation that business device 43 will be generated.
Next, information terminal 20 is in the case of text message after receiving translation from server 43, will translate hereinafter This information is transmitted (S108) to server 44.
On the other hand, in S105, information terminal 20 is receiving the situation of the second particular text information from server 42 Under, translation treatment is skipped, the second particular text information is transmitted (S108) to server 44.
Next, server 44 performs phonetic synthesis treatment, text envelope after the second particular text information of generation or translation Speech data after the translation of breath.And, text message after the second particular text information that server 44 will be generated or translation Translation after speech data be transmitted (S109) to information terminal 20.
Finally, information terminal 20 performs text after the second particular text information or the translation to being received from server 44 The voice output (S110) of speech data after the translation of information.
[effect etc.]
As described above, according to this variation, speaker (speaker) is only said comprising the first language to machine translation system 10 The certain words of speech or it is specific text (the first particular text information), it becomes possible to make by each using purposes exclusive session, Word, expression etc. continually using second language sizing text (the second particular text information) exported.Thus, can mitigate The burden of speaking of speaker.
In addition, although be set to be stored with the storage part 163 that server 42 has related to the first particular text information Second particular text information of connection, but it is not limited to this.In storage part 163, it is also possible to be stored with and believe with the first particular text Speech data after the translation of the second particular text information of manner of breathing association.Hereinafter, in this case, information terminal 20 and each clothes The exchange of the data between business device is illustrated.
Figure 13 is the timing diagram of another of the work of the machine translation system 10A in the variation for represent implementation method 1. Pair key element same with Figure 12 addition of identical label, and omit detailed description.Timing diagram and Figure 12 institutes shown in Figure 13 The timing diagram for showing compares, and the treatment of S105a and S108a is different, therefore, only difference is illustrated below.
That is, in S105a, server 42 performs determination processing, before the translation for being judged to be received from information terminal 20 In the case that text message contains the first particular text information, the second particular text being associated with the first particular text information Speech data is sent to information terminal 20 after the translation of information.
In addition, in S108a, the language after the translation that the second particular text information is received from server 42 of information terminal 20 In the case of sound data, any content is not sent to server 44, that is to say, that skip phonetic synthesis treatment.
Therefore, in S110, information terminal 20 performs turning over to the second particular text information for being received from server 42 Translate the voice output of rear speech data.
(implementation method 2)
On the machine translation system 10 that implementation method 1 is related to, illustrate to be judged before translation be in text message The situation of the no determination processing containing the first particular text information for being stored in storage part 163, but it is not limited to this.Can also enter Whether row is judged at the judgement completely the same with the first particular text information of storage part 163 is stored in of the preceding text message of translation Reason.For the situation, below as implementation method 2, by with the difference of implementation method 1 centered on illustrate.
[structure of machine translation system 10]
Compared with the machine translation system 10 that the machine translation system 10 that implementation method 2 is related to is related to implementation method 1, turn over It is different to translate determination processing unit 16.In addition the machine translation system 10 that structure is related to implementation method 1 be likewise, Therefore omit the description.
＜ translates the ＞ of determination processing unit 16
The translation determination processing unit 16 that implementation method 2 is related to is compared with the translation determination processing unit 16 that implementation method 1 is related to Compared with the work of detection unit 162 is different.In addition the translation determination processing unit 16 that structure is related to implementation method 1 is Likewise, therefore omitting the description.
In the present embodiment, detection unit 162 judge obtaining section 161 acquired by translation before text message be stored in Whether the first particular text information in storage portion 163 is consistent.Detection unit 162 is being judged to translate preceding text message and the first specific text In the case that this information is consistent, exports output section 164 and be stored in storage part 163 and be associated with the first particular text information The second particular text information.Other work as in the first embodiment described, therefore, omit explanation in this.
[work of machine translation system 10]
Concrete mode to the work of the machine translation system 10 in the implementation method 2 according to composition as described above is said It is bright.
Figure 14 is the flow chart of the concrete mode of the work for representing the machine translation system in implementation method 2.It is pair same with Fig. 7 The key element of sample addition of identical label, and omit detailed description.That is, the place in S11, S12, S14 shown in Figure 14~S17 Manage as in the first embodiment described, therefore omit the description.Hereinafter, it is pair different from implementation method 1 including including S23 Determination processing is illustrated.
In S23, translation determination processing unit 16 carries out determination processing.That is, translation determination processing unit 16 judges acquired Translation before text message with pre-register in storage part 163 certain words or it is specific text whether consistent (S23).And, when It is judged to translate preceding text message in S23 and pre-registers in the consistent feelings of the certain words or specific text of storage part 163 (S23 under condition：It is), extraction is stored in storage part 163 and the second particular text letter being associated with the first particular text information Breath (S14).
Here, using Fig. 8, more specifically illustrating.
For example, speaker says " Fan Breakdown Machine to the machine translation system 10 in implementation method 2 with Japanese illustrates The ま The (carrying out the explanation of translater) ".As shown in figure 8, in storage part 163, the spy that is stored with as the first particular text information Fixed text " Fan Breakdown Machine explanations (explanation of translater) ", but the content of speaking of speaker is except " Fan Breakdown Machine illustrate (translater Explanation) " beyond this words and phrases, also contain " The ま The (carrying out) " this words and phrases.In this case, in S23, embodiment party Machine translation system 10 in formula 2 is judged to that the content of speaking of speaker is not fully consistent with the first particular text information (S23：It is no), advance to S15.Additionally, the machine translation system 10 in implementation method 2 performs translation treatment in S15, by " I Text message exports language as a result after the translation of explain this translator machine. " this second language Sound combining unit 14.
On the other hand, if speaker says " Fan Breakdown Machine explanations to the machine translation system 10 in implementation method 2 with Japanese (explanation of translater) ".As shown in figure 8, in storage part 163, as the first particular text information, the specific text that is stored with " is turned over Breakdown Machine illustrate (explanation of translater) ".In this case, in S23, the machine translation system 10 in implementation method 2 judges It is speak content and the first completely the same (S23 of particular text information of speaker：It is), advance to S14.
[effect etc.]
As described above, the machine translation system 10 in implementation method 2 text message and first particular text before translation are believed In the case that breath is consistent, the second particular text information is exported.That is, the machine translation system 10 is speaker (speaker) In the case of only having said the first particular text information, exported the second particular text information as translation result, in speaker In the case that (speaker) has been said comprising the words and phrases beyond the first particular text information, output is carried out in itself to described content Text is translated obtained from translation.
Thus, following effects can be obtained, i.e., can be by the first particular text using the speaker of the machine translation system 10 Certain words or specific text represented by information are contained in dialogue and are used.
In other words, there is speaker to want to illustrate that (translater is said using on such as " Fan Breakdown Machine to other side's (hearer) It is bright) " etc. registered specific text sizing is literary, i.e. the literary situation of the usage frequency of registration sizing high in advance.In this case, Speaker can export the machine translation system 10 by only saying registered certain words (the first particular text information) Sizing is literary (the second particular text information).
On the other hand, content of sizing text (the second particular text information) of prior registration etc. is not understood that in other side sometimes In the case of, speaker wants to illustrate hearer using the speak expression different from the sizing text of prior registration.Now, when this Machine translation system 10 is in content of speaking (translation before text message) containing for example " turning over Breakdown Machine and illustrate that (translater is said It is bright) " etc. registered specific text (the first particular text information) when, output is corresponding with specific literary (the first particular text information) Sizing text in the case of, speaker can be by can not comprising the first particular text information, this be limited in content of speaking.Cause Whether this, in implementation method 2, uses text message before translation and consistent this decision condition of the first particular text information.By This, speaker is free to say comprising such as " Fan Breakdown Machine explanations (explanation of translater) " etc. registered specific text If the content of (the first particular text information), the machine translation system 10 is set to carry out translation treatment.So, speaker being capable of spirit Select illustration method livingly, i.e., be to hearer to export sizing text in the way of illustrate, or entered using the language of oneself Row explanation.
In addition, for example, it is assumed that in the example for carrying out the explanation on translater, utilized before hearer has and turned over In the case of translating the experience of device, speaker carries out " Fan Breakdown Machine explanations は such as to hearer should not In ょ う か(whether not Need the explanation of translater) " such confirmation.Now, work as in content of speaking containing for example in the machine translation system 10 When " explanation of translater " etc. is registered specific literary in the case of output text of shaping corresponding with the specific text, it is impossible to be somebody's turn to do Confirm.On the other hand, in the present embodiment, は can will be illustrated not such as " Fan Breakdown Machine by the machine translation system 10 output Want In ょ う か(explanation of translater whether is not needed) " such content of speaking translated obtained from the second language The translation text of speech, the confirmation is carried out to hearer.So, the machine translation system 10 according to present embodiment, speaker can also Enough natural dialogues are confirmed whether to need the explanation of translater, are prevented from customer's output of the explanation to not needing translater As the explanation of translater of sizing text, this situation occurs.
Furthermore, the machine translation system 10 according to present embodiment, text message is specific with as first only before translation In the case that the certain words of text message or specific text are completely the same, the second particular text information is exported.Therefore, also can use Following effects are obtained, that is, is eliminated speaker and is such as taken notice of the content of speaking of speaker oneself with so that not comprising spy in common dialogue Order word or the such unnecessary worry of specific text.
Additionally, machine translation system 10 and the grade of implementation method 1 are it is equally possible that possess display.In this case, machine Device translation system 10 carry out determination processing and to English as mother tongue hearer output be stored in storage part 163 and with During the sizing text of specific literary corresponding English, in the display will literary (the 3rd specific text of sizing text translation corresponding with sizing text This information) it is shown to speaker with Japanese as mother tongue.Thus, the specific text of the first language in output and described in speaker is not In the case of the voice of sizing text (the second particular text information) of same but associated second language, speaker can be notified Output the voice of the sizing text of what content.Even if in addition, when speaker does not remember the content of sizing text, it is also possible to when Field confirms the content of the sizing text of second language with sizing text translation text (sizing of first language is literary), therefore, with hearer Dialogue also can successfully carry out.
(variation)
Implementation method 1 and implementation method 2 are illustrated as different implementation methods respectively above, but it is also possible to They are combined and is implemented.One of the situation is illustrated using Fig. 8.
In fig. 8, it is provided with project " classification ".For example the specific of " 1 " is set on the value represented by " classification " Text, can contain the feelings of the specific text (the first particular text information) in the content of speaking (text message before translation) of speaker Under condition, machine translation system 10 is set to export shape literary (second particular text information) corresponding with the specific text.On the other hand, close Be set to the specific text of " 2 " in the attribute represented by " classification ", can only speaker content of speaking (translation before text envelope Breath) with this it is specific text (first particular text information) it is completely the same in the case of, make machine translation system 10 export it is specific with this The corresponding sizing of text is literary (the second particular text information).
Thereby, it is possible to carry out determination processing by following other rule：On the specific text being rarely employed, if be contained in Somewhere in the text of speaking of speaker, then replace with the second particular text information registered in advance, high on usage frequency Certain words and/or specific text, then only in the case of completely the same, replace with the second particular text letter registered in advance Breath.I.e. it is capable to carry out determination processing to each specific text by individual other rule.This case is it is meant that being capable of basis Usage frequency of the certain words, particular text to be registered in dialogue usually, individually to set wieldy rule. I.e. it is capable to make the convenience of speaker (user) further improve.
(implementation method 3)
It is pair different from illustrated determination processing in implementation method 1 and implementation method 2 to sentence in implementation method 3 The example of fixed treatment is illustrated.
[structure of machine translation system 10]
Compared with the machine translation system 10 that the machine translation system 10 that implementation method 3 is related to is related to implementation method 1, turn over It is different to translate determination processing unit 16.In addition the machine translation system 10 that structure is related to implementation method 1 be likewise, Therefore omit the description.
＜ translates the ＞ of determination processing unit 16
The translation determination processing unit 16 that implementation method 3 is related to is compared with the translation determination processing unit 16 that implementation method 1 is related to Compared with the work of the content and detection unit 162 that are stored in storage part 163 is different.In addition structure and embodiment party The translation determination processing unit 16 that formula 1 is related to is likewise, therefore omitting the description.
In the present embodiment, storage part 163 is stored with the first particular text information and the second particular text in association Information, certain words or specific text that the first particular text information expression is expressed with first language, described second is specific Text message represents the word or sentence of the pre-prepd second language expression for being used as the language different from first language Shape literary, the second particular text information does not have translation equivalence with certain words or specific text.
Furthermore, 163 pairs of the second particular text information of storage part are stored with more than two first specific in association respectively Text message and order information, the order information represent that more than two first particular text information should appear in text respectively In order.
Detection unit 162 judge before translation in text message whether containing be stored in storage part 163 this is more than two Whether in order the first particular text information and the order represented by information contain more than two first particular text letters Breath.
Here, set detection unit 162 be judged as translating preceding text message contain be stored in storage part 163 this is more than two First particular text information and be that order in the order represented by information contains more than two first particular texts letters Breath.In this case, detection unit 162 make output section 164 export with more than two first particular text information and should Second particular text information of order information association.
Additionally, on other structures, as in the first embodiment described, therefore, omit explanation herein.
So, the translation determination processing unit 16 in implementation method 3 judge speaker content of speaking (translation before text envelope Breath) in whether contain the certain words that pre-register or specific text (the first particular text by the order of speaking registered in advance Information).And, when in the content of speaking of speaker by registered in advance speak order the situation containing the certain words etc. Under, translation treatment is not carried out, and (the second particular text is believed to export sizing text corresponding with the certain words etc., pre-registering Breath).
[work of machine translation system 10]
Using Figure 15 and Figure 16, to the machine translation system 10 that is related to according to the implementation method 3 of composition as described above The concrete mode of work is illustrated.
Figure 15 is the flow chart of the concrete mode of the work for representing the machine translation system 10 in implementation method 3.Pair and Fig. 7 Same key element addition of identical label, and omit detailed description.That is, in S11, S12, S14 shown in Figure 15~S17 Treatment as in the first embodiment described, therefore, it is pair different from implementation method 1 below, including sentencing including S33, S34 Fixed treatment is illustrated.
In S33 and S34, translation determination processing unit 16 carries out determination processing.That is, translation determination processing unit 16 judges institute Whether text message contains the multiple certain words or multiple specific texts pre-registered in storage part 163 before the translation of acquirement (S33)。
Determination processing unit 16 is translated in S33 and is judged to translate that preceding text message contains the plurality of certain words or this is more (S33 in the case of individual specific text：It is), determine whether the order as represented by the order information pre-registered in storage part 163 Contain the plurality of certain words or the plurality of specific text (S34).That is, in S34, translation determination processing unit 16 judges The plurality of certain words or the plurality of specific text are included in the preceding text message of translation in what order.It is more specific and Speech, in S34, translation determination processing unit 16 judge the plurality of certain words or it is the plurality of it is specific text order whether with advance It is registered in the sequence consensus represented by the order information of storage part 163.
Determination processing unit 16 is translated in the S34 it is judged to that to be that order in the order represented by information contains the plurality of specific (S34 in the case of word or the plurality of specific text：Be), extraction be stored in storage part 163 and with the plurality of certain words Or the second particular text information (S14) that the plurality of specific text is associated.Additionally, work as to be judged to translate preceding text in S33 Information do not include the plurality of certain words or the plurality of specific text, or be judged in S34 the plurality of certain words or In the case that order of the plurality of specific text not in the order represented by information is contained in the preceding text message of translation, advance to S15。
Figure 16 is to represent in implementation method 3 to be stored in being associated with certain words and order of speaking for storage part in association Sizing text the figure of.
In figure 16, as one of the first particular text information, multiple certain words of Japanese are illustrated, as expression Multiple certain words should occur one of the order information of order in the text, illustrate order of speaking.In addition, in Figure 16 In, as one of the second particular text information, illustrate the sizing text of English.Furthermore, in figure 16, as the 3rd specific text One of this information, illustrates the sizing text translation text translated into the sizing text of English obtained by Japanese.
More specifically, in figure 16, this is suitable with " (1) " to illustrate such as " (guide) in case " this certain words Sequence information is associated, " East Jing Station (Tokyo station) " this certain words stores and depositing in association with " (2) " this order information This case in storage portion 163.
For example, when speaker says " case Nei, East Jing Station (guide, Tokyo station) " to machine translation system 10 with Japanese, In machine translation system 10, the speech data of the voice described in speaker is converted to by turning over for Japanese by voice recognition processing Translate preceding text message.Then, in machine translation system 10, determination processing is carried out by translating determination processing unit 16, is judged to Contain multiple certain words i.e. " (guide) in case " for being stored in storage part 163 and " East capital Station (east before translation in text message Capital station) ".And then, in machine translation system 10, judge whether the plurality of certain words press what is be associated with each certain words Order shown in order information is contained in text message before translation.That is, in storage part 163, this is special with " (guide) in case " Order represented by the associated order information of order word is " (1) ", to " East Jing Station (Tokyo station) " this certain words is related Order represented by the order information of connection is " (2) ".And, in machine translation system 10, it is judged to the text envelope before translation In breath with this sequentially contain " (guide) in case " and " East Jing Station (Tokyo station) " as certain words.That is, machine Translation system 10 is judged to order and the sequence consensus for determining represented by order information.Therefore, machine translation system 10 is carried Take and multiple certain words i.e. " (guide) in case " and " East Jing Station (Tokyo station) " (the first particular text information) determining of associating Type text (the second particular text information) is " We will guide you to Tokyo Station.First, please turn left out from this building and go as straight 100m.And then,you will find it On the right hand side. (now navigate to you and go to Tokyo station.Turn left from this building first, keep straight on 100m. Then, the right hand can be shown in Tokyo station.) ", and phonetic synthesis treatment and voice output treatment are performed, export determining for above-mentioned English The voice of type text.
On the other hand, if speaker to hearer in order to be confirmed whether to need the guide to Tokyo station, with Japanese to machine Translation system 10 says The い ま ょ う か in " East Jing Station へ ご cases(allow me to fugle with you and go to Tokyo station ) " like this.In this case, in machine translation system 10, by voice recognition processing by the language described in speaker The speech data of sound is converted to text message before the translation of Japanese.Then, in machine translation system 10, by translation determination processing Portion 16 carries out determination processing, is judged to that translating preceding text message contains multiple certain words i.e. " the East capital for being stored in storage part 163 Station (Tokyo station) " and " (guide) in case ".And then, in machine translation system 10, judge whether the plurality of certain words press The order shown in order information being associated with each certain words is contained in text message before translation.That is, in storage part 163 In, Yu " East Jing Station (Tokyo station) " order represented by the order information that is associated of this certain words is " (2) ", and " in case Order represented by the associated order information of this certain words of (guide) " is " (1) ".And, in machine translation system 10 In, it is judged to before translation in text message not sequentially contain " guide " and certain words as " Tokyo station " with this. That is, machine translation system 10 is judged to that the order represented by order information is inconsistent with the order for determining.Therefore, machine Device translation system 10 enters to be about to the article of the Japanese described in speaker " The い ま ょ う か in the Station へ ご cases of East capital (allow me to fugle with you and go to Tokyo station) " translate into English translation treatment, be for example translated as " Would you wish a guide to Tokyo Station" this English article, and perform phonetic synthesis treatment and voice output treatment, it is defeated Go out the voice of the article of above-mentioned English.
Additionally, speaker says to machine translation system 10 has used multiple certain words " explanation (explanation) " and " チ ェ ッ Situation if Network ア ウ ト (checking out (checkout)) " is also likewise, therefore, omitting explanation herein.
In addition, machine translation system 10 and the grade of implementation method 1 are it is equally possible that possess display.In this case, machine Device translation system 10 carry out determination processing and to English as mother tongue hearer output be stored in storage part 163 and with During the sizing text of specific literary corresponding English, in the display will literary (the 3rd specific text of sizing text translation corresponding with sizing text This information) it is shown to speaker with Japanese as mother tongue.Thus, the specific text of the first language in output and described in speaker is not In the case of the voice of sizing text (the second particular text information) of same but associated second language, speaker can be notified Output the voice of the sizing text of what content.Even if in addition, when speaker does not remember the content of sizing text, it is also possible to when Field confirms the content of the sizing text of second language with sizing text translation text (sizing of first language is literary), therefore, with hearer Dialogue also can successfully carry out.
[effect etc.]
As described above, according to implementation method 3, speaker (speaker) according to whether to note multiple certain words or many Whether the mode of order of speaking of individual specific text is spoken to machine translation system 10, can easily switch using being registered in advance Usage frequency it is high sizing text be the second particular text information.
In addition, for example in the case where first language is Japanese, as shown in figure 16, effectively using the usual dialogue in Japanese Middle verb is contained in last this phraseological feature of sentence, and the certain words or specific text setting to showing " verb " are suitable Sequence information " (1) ", certain words or specific text setting order information to performance " subject ", " object " in addition etc. " (2) ".Thereby, it is possible to prevent when in text of speaking not so as to for the purpose of the second particular text information output but contain Certain words or it is specific literary when export the second particular text information.In addition, speaker can suitably enter according to tongue The following selections of row, i.e., be the second particular text of output as translation result for the text of speaking comprising the first particular text information Information, or output described content is translated in itself obtained from translate text.
That is, the machine translation system in implementation method 3, can be readily available following effects：By intentionally with The different mode of common grammer is spoken, and the translation of the usual dialogue of speaker can not be influenceed to process, and it is specific to export second Text message.In addition, the machine translation system in implementation method 3, the second of storage part 163 is stored in even not knowing The speaker of the output condition of particular text information, as long as being spoken according to common grammer, it is also possible to prevent from unconsciously exporting Second particular text information.
In addition it is also possible to " verb " and " other parts of speech " is paired advance as multiple certain words or multiple specific texts It is registered in storage part 163.For example, to " verb " setting " guide " this certain words, and " (1) " conduct is set in association For the order information of the certain words.On the other hand, the specific list of multiple " other parts of speech " matched with it can be set Word.For example, " Osaka station ", " Tokyo sky tree ", " lavatory " etc..Now, as the order information for " other parts of speech ", also may be used To be all set as in association " (2) ".So, multiple certain words or multiple specific are made by way of being associated with 1 couple of n Being stored in storage part 163 to (pair) for text, can mechanically supplement the second particular text information, moreover it is possible to realize storage part The ease for maintenance of 163 additional, renewal, repeated entries avoidance etc..
More than, the machine translation method that one or more technical scheme of the disclosure is related to is illustrated based on implementation method And machine translation system, but the disclosure is not limited to the implementation method.Without departing from the objective of the disclosure, by ability The various modifications that field technique personnel expect are applied to present embodiment and/or the inscape in different embodiments are combined into structure The mode built can also be included in the range of one or more technical scheme of the disclosure.
For example, in the machine translation method and machine translation system of the disclosure, as illustrated with Figure 10, also may be used So that voice recognition processing, translation determination processing, translation treatment, phonetic synthesis are performed by different independent servers respectively Treatment.It is of course also possible to pass through the part treatment during same server performs these multiple treatment, it is same it can in addition contain pass through One server performs all for the treatment of.Whether any mode, can obtain same effect.
In addition, being not necessarily required to perform above-mentioned multiple treatment on the server via communication units such as networks, it is also possible in letter In breath terminal a part therein is performed in execution, the functional structure i.e. in the information terminal connected via internal bus.Separately Outward, it is also possible to perform a part therein by being directly connected in the ancillary equipment of information terminal.
In addition, in the machine translation method and machine translation system of the disclosure, the content translation that voice is spoken into It is multilingual when conveying to other side, as described above, it is also possible to other side is not only conveyed to by multilingual voice output, also lead to Cross and show multilingual text in the display and convey to other side.Its reason is that can obtain same effect.
Additionally, technology illustrated in the above-mentioned technical solutions, such as can realize in the type of following cloud service. But, realize that the type of the cloud service of technology illustrated in the above-mentioned technical solutions is not limited to these.
(the Class1 of service：Our company data-centric cloud service)
Figure 17 is that the information management system in the Class1 (our company data-centric cloud service) of the service that represents is provided The figure of the overall picture of service.In this type, service provider 11200 obtains information from group 11000, provides a user with service. In this type, service provider 11200 has the function of data center's operator.That is, service provider 11200 possesses management The Cloud Server 11110 of big data (big data).Therefore, in the absence of data center's operator.
In this type, service provider 11200 is runed and is managed to data center's (Cloud Server) 12030.Separately Outward, service provider 11200 manages operating system (OS) 12020 and applies 12010.Service provider 11200 uses service OS12020 and service (arrow 12040) is provided using 12010 that provider 11200 is managed.
(the type 2 of service：IaaS utilizes type cloud service)
Figure 18 is the service that the information management system in the type 2 (IaaS utilizes type cloud service) of the service that represents is provided The figure of overall picture.Here, IaaS is infrastructure service the abbreviation of (Infrastructure as a Service), be by with The cloud service itself provided as the service via internet in the basis for building and running computer system provides model.
In this type, data center's operator 11100 is runed and managed to data center's (Cloud Server) 12030 Reason.In addition, the management OS12020 of service provider 11200 and applying 12010.Service provider 11200 uses service provider 11200 OS12020 for being managed and service (arrow 12040) is provided using 12010.
(the type 3 of service：PaaS utilizes type cloud service)
Figure 19 is the service that the information management system in the type 3 (PaaS utilizes type cloud service) of the service that represents is provided The figure of overall picture.Here, PaaS is platform servicing the abbreviation of (Platform as a Service), will be turned into for building And the cloud service that the platform of the foundation of runs software is provided as the service via internet provides model.
In this type, the management OS12020 of data center's operator 11100, to data center's (Cloud Server) 12030 Runed and managed.In addition, the management of service provider 11200 applies 12010.Service provider 11200 uses data center Application 12010 that the OS12020 and service provider 11200 that operator 11100 is managed are managed provides service (arrow It is first 12040).
(the type 4 of service：SaaS utilizes type cloud service)
Figure 20 is the service that the information management system in the type 4 (SaaS utilizes type cloud service) of the service that represents is provided The figure of overall picture.Here, SaaS is software servicing the abbreviation of (Software as a Service).SaaS utilizes type cloud service The e.g. cloud service with following function provides model, that is, do not possess the profits such as company or the individual of data center's (Cloud Server) The application that the platform provider that user can possess data center's (Cloud Server) via Web vector graphics such as internets is provided.
In this type, the management of data center's operator 11100 applies 12010, OS12020 is managed, to data center (Cloud Server) 12030 is runed and managed.In addition, service provider 11200 uses the institute of data center's operator 11100 The OS12020 of management and service (arrow 12040) is provided using 12010.
More than, no matter in the type of any cloud service, service provider 11200 provides service.In addition, for example, Service provider or data center's operator both can with self-developing OS, using or big data database, in addition Can also be ordered to third party.
Industrial applicability
The disclosure can be used in machine translation method and machine translation system, enable in particular to be used in PC applications, Web Using the machine translation method in the machine translation system and machine translation system of the simple possibles such as, smart mobile phone application.

Claims (12)

1. a kind of machine translation method, is the machine translation method in machine translation system,
In the storage part of the machine translation system, be stored with the first particular text information in association, and the second specific text At least one party in the second speech data of this information and the second particular text information, the first particular text information is represented With first language express certain words or specific text, the second particular text information represent it is pre-prepd be used as with The word of the second language expression of the different language of the first language or the sizing text of sentence, the second particular text letter Breath does not have translation equivalence with the certain words or the specific text,
The machine translation method includes：
Acquisition step, text message, described first before the translation that the first speech data is converted to text message and generated by acquirement Speech data represents the input voice said with the first language；
Whether determination step, judge before the translation obtained in the acquisition step text message containing being stored in described depositing The first particular text information in storage portion；And
Output step, text message contains the first particular text letter before the translation is judged in the determination step In the case of breath, output be stored in the storage part and be associated with the first particular text information it is described second specific At least one party in the second speech data of text message and the second particular text information.
2. machine translation method according to claim 1,
In the determination step, whether text message is special be stored in the storage part described first before judging the translation Determine text message consistent,
In the output step, text message is specific with described first before the translation is judged in the determination step In the case that text message is consistent, output is stored in the storage part and the institute being associated with the first particular text information State at least one party in the second speech data of the second particular text information and the second particular text information.
3. machine translation method according to claim 1,
In the storage part, more than two described first are stored with association respectively to the second particular text information Particular text information and order information, the order information represent that the first particular text information more than described two respectively should Appearance order in the text,
In the determination step, whether text message contains storage before judging the translation obtained in the acquisition step In the storage part it is described two more than the first particular text information and whether according to the sequence suitable represented by information Sequence contains the first particular text information more than described two,
In the output step, before the translation is judged in the determination step text message contain be stored in it is described Storage part it is described two more than the first particular text information and be order according to the sequence represented by information containing In the case of stating more than two first particular text information, export and described two the first particular text information above And the second particular text information and second voice of the second particular text information of the order information association At least one party in data.
4. machine translation method according to claim 1 and 2,
In the storage part, the second particular text information described to one, be stored with more than one different institute in association The first particular text information is stated, more than one different the first particular text information contains the specific list of identical respectively Word and the different specific text of expression.
5. machine translation method according to any one of claim 1 to 4,
In the output step, text message does not include described first before the translation is judged in the determination step In the case of particular text information, after text message before the translation is translated into translation obtained from the second language by output Text message.
6. machine translation method according to claim 5,
In the output step, before the translation is judged in the determination step text message contain be stored in it is described In the case of the first particular text information of storage part, text message after the translation is not exported.
7. machine translation method according to any one of claim 1 to 6,
Also make the 3rd particular text information at least with the first particular text information and the second particular text information in The second particular text information is stored in the storage part in association, and the 3rd particular text information is by described second Particular text information translates into particular text information obtained from the first language,
In the output step, described the second of the second particular text information and the second particular text information is being exported In the case of at least one party in speech data, the 3rd particular text information is also exported.
8. machine translation method according to claim 7,
The 3rd particular text presentation of information exported in the output step is in display.
9. machine translation method according to any one of claim 1 to 8,
The machine translation system is connected by predetermined communication unit with the information terminal with display,
In the output step, by the predetermined communication unit, by the second particular text information and second spy Described information terminal is arrived at least one party's output determined in the second speech data of text message.
10. machine translation method according to claim 9,
In described information terminal, in the case of the second particular text information is outputed in the output step, to institute Stating the second particular text information carries out phonetic synthesis treatment, generates second speech data, to the second speech data for generating Represented voice is exported.
11. machine translation methods according to any one of claim 1 to 10,
The predetermined state that the machine translation method is used between the people for saying the people of the first language and saying the second language In.
A kind of 12. machine translation systems, possess：
Storage part, be stored with the first particular text information in association, and the second particular text information and second particular text At least one party in the second speech data of information, the specific list that the first particular text information expression is expressed with first language Word or specific text, the second particular text information represent pre-prepd and are used as the language different from the first language Second language expression word or the sizing text of sentence, the second particular text information and the certain words or institute Specific text is stated without translation equivalence；
Obtaining section, text message, first language before the translation that the first speech data is converted to text message and generated by acquirement Sound data represent the input voice said with the first language；
Whether detection unit, judge before the translation obtained in the obtaining section text message containing being stored in the storage part The first particular text information；And
Output section, text message contains the first particular text information before the translation is judged to by the detection unit In the case of, output is stored in the storage part and second particular text being associated with the first particular text information At least one party in the second speech data of information and the second particular text information.
CN201610903217.2A 2015-12-25 2016-10-17 Machine translation method and machine translation system Pending CN106919559A (en)

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
JP2015253935 2015-12-25
JP2015-253935 2015-12-25
JP2016145755A JP2017120616A (en) 2015-12-25 2016-07-25 Machine translation method and machine translation system
JP2016-145755 2016-07-25

Publications (1)

Publication Number Publication Date
CN106919559A true CN106919559A (en) 2017-07-04

Family

ID=59272110

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610903217.2A Pending CN106919559A (en) 2015-12-25 2016-10-17 Machine translation method and machine translation system

Country Status (2)

Country Link
JP (1) JP2017120616A (en)
CN (1) CN106919559A (en)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109426669A (en) * 2017-08-25 2019-03-05 松下电器（美国）知识产权公司 Information processing method, information processing unit and the recording medium having program recorded thereon
CN110021290A (en) * 2018-01-08 2019-07-16 上海西门子医疗器械有限公司 Medical system and real-time language conversion method for medical system
CN110322881A (en) * 2018-03-29 2019-10-11 松下电器产业株式会社 Speech translation apparatus, voice translation method and its storage medium
CN111785258A (en) * 2020-07-13 2020-10-16 四川长虹电器股份有限公司 Personalized voice translation method and device based on speaker characteristics
CN112055876A (en) * 2018-04-27 2020-12-08 语享路有限责任公司 Multi-party dialogue recording/outputting method using voice recognition technology and apparatus therefor
CN115278331A (en) * 2022-06-23 2022-11-01 中译语通科技股份有限公司 Multi-language subtitle display method, system, device and medium based on machine translation

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH05197744A (en) * 1992-01-17 1993-08-06 Nippon Hoso Kyokai <Nhk> Word processor for foreign language
US20030115059A1 (en) * 2001-12-17 2003-06-19 Neville Jayaratne Real time translator and method of performing real time translation of a plurality of spoken languages
US20040002848A1 (en) * 2002-06-28 2004-01-01 Ming Zhou Example based machine translation system
US20040102956A1 (en) * 2002-11-22 2004-05-27 Levin Robert E. Language translation system and method
CN101008942A (en) * 2006-01-25 2007-08-01 北京金远见电脑技术有限公司 Machine translation device and method thereof
US20080133245A1 (en) * 2006-12-04 2008-06-05 Sehda, Inc. Methods for speech-to-speech translation
WO2009038209A1 (en) * 2007-09-20 2009-03-26 Nec Corporation Machine translation system, machine translation method, and machine translation program
US20110238407A1 (en) * 2009-08-31 2011-09-29 O3 Technologies, Llc Systems and methods for speech-to-speech translation
CN103299361A (en) * 2010-08-05 2013-09-11 谷歌公司 Translating languages

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH05197744A (en) * 1992-01-17 1993-08-06 Nippon Hoso Kyokai <Nhk> Word processor for foreign language
US20030115059A1 (en) * 2001-12-17 2003-06-19 Neville Jayaratne Real time translator and method of performing real time translation of a plurality of spoken languages
US20040002848A1 (en) * 2002-06-28 2004-01-01 Ming Zhou Example based machine translation system
US20040102956A1 (en) * 2002-11-22 2004-05-27 Levin Robert E. Language translation system and method
CN101008942A (en) * 2006-01-25 2007-08-01 北京金远见电脑技术有限公司 Machine translation device and method thereof
US20080133245A1 (en) * 2006-12-04 2008-06-05 Sehda, Inc. Methods for speech-to-speech translation
WO2009038209A1 (en) * 2007-09-20 2009-03-26 Nec Corporation Machine translation system, machine translation method, and machine translation program
US20110238407A1 (en) * 2009-08-31 2011-09-29 O3 Technologies, Llc Systems and methods for speech-to-speech translation
CN103299361A (en) * 2010-08-05 2013-09-11 谷歌公司 Translating languages

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109426669A (en) * 2017-08-25 2019-03-05 松下电器（美国）知识产权公司 Information processing method, information processing unit and the recording medium having program recorded thereon
CN109426669B (en) * 2017-08-25 2023-03-24 松下电器（美国）知识产权公司 Information processing method, information processing apparatus, and recording medium having program recorded thereon
CN110021290A (en) * 2018-01-08 2019-07-16 上海西门子医疗器械有限公司 Medical system and real-time language conversion method for medical system
CN110322881A (en) * 2018-03-29 2019-10-11 松下电器产业株式会社 Speech translation apparatus, voice translation method and its storage medium
CN112055876A (en) * 2018-04-27 2020-12-08 语享路有限责任公司 Multi-party dialogue recording/outputting method using voice recognition technology and apparatus therefor
CN111785258A (en) * 2020-07-13 2020-10-16 四川长虹电器股份有限公司 Personalized voice translation method and device based on speaker characteristics
CN111785258B (en) * 2020-07-13 2022-02-01 四川长虹电器股份有限公司 Personalized voice translation method and device based on speaker characteristics
CN115278331A (en) * 2022-06-23 2022-11-01 中译语通科技股份有限公司 Multi-language subtitle display method, system, device and medium based on machine translation
CN115278331B (en) * 2022-06-23 2023-10-20 中译语通科技股份有限公司 Multi-language subtitle display method, system, equipment and medium based on machine translation

Also Published As

Publication number Publication date
JP2017120616A (en) 2017-07-06

Similar Documents

Publication Publication Date Title
CN106919559A (en) Machine translation method and machine translation system
US9317501B2 (en) Data security system for natural language translation
US9047868B1 (en) Language model data collection
KR20190100512A (en) Electronic device and method for communicating with chatbot
US11030400B2 (en) System and method for identifying and replacing slots with variable slots
US20160225372A1 (en) Smart home connected device contextual learning using audio commands
CN108010523A (en) Information processing method and recording medium
KR102445519B1 (en) System and method for manufacturing conversational intelligence service providing chatbot
US11915693B2 (en) System and method for rule based modifications to variable slots based on context
EP3866160A1 (en) Electronic device and control method thereof
CN111009233A (en) Voice processing method and device, electronic equipment and storage medium
JP6927318B2 (en) Information processing equipment, information processing methods, and programs
US20210350784A1 (en) Correct pronunciation of names in text-to-speech synthesis
WO2016203805A1 (en) Information processing device, information processing system, information processing method, and program
JP2018120202A (en) Information processing method and program
ES2751375T3 (en) Linguistic analysis based on a selection of words and linguistic analysis device
CN106713111A (en) Processing method for adding friends, terminal and server
CN110175323A (en) Method and device for generating message abstract
EP1402518B1 (en) Pattern processing system specific to a user group
CN108268602A (en) Analyze method, apparatus, equipment and the computer storage media of text topic point
CN107870899A (en) Information processing method, message processing device and program product
KR102226427B1 (en) Apparatus for determining title of user, system including the same, terminal and method for the same
CN110740212B (en) Call answering method and device based on intelligent voice technology and electronic equipment
US20220292261A1 (en) Methods for Emotion Classification in Text
CN111556096B (en) Information pushing method, device, medium and electronic equipment

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
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20170704

WD01 Invention patent application deemed withdrawn after publication