CN101409074A - Method for updating embedded speech identification system function - Google Patents
Method for updating embedded speech identification system function Download PDFInfo
- Publication number
- CN101409074A CN101409074A CNA2008101624045A CN200810162404A CN101409074A CN 101409074 A CN101409074 A CN 101409074A CN A2008101624045 A CNA2008101624045 A CN A2008101624045A CN 200810162404 A CN200810162404 A CN 200810162404A CN 101409074 A CN101409074 A CN 101409074A
- Authority
- CN
- China
- Prior art keywords
- function
- renewal
- speech recognition
- data
- file
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a method used for updating the functions of an embedded speech recognition system. The invention comprises the steps as follows: a function updating request is initiated by the system; the present speech recognition program is halted and the present data is backed up; updating data are obtained by the network or a local file system; the obtained updating data are checked; speech recognition training is carried out according to the updating data and the training result is stored; and the system with updated functions is operated. By adopting the method used for updating the functions of the embedded speech recognition system of the invention, updating can be carried out on the functions of the system by the user without changing the speech recognition program maintained in the embedded system, thereby improving user experience of the product and raising value of the product.
Description
Technical field
The present invention relates to the method for updating embedded speech identification system function, relate in particular to and a kind ofly need not upgrade speech recognition program and realize the systemic-function method for updating.
Background technology
Speech recognition technology is since the 1950's development, up to now, comparative maturity in theory research, however in application, be subjected to the restriction of many-sided factors such as hardware performance, cost and the market demand, be not widely used for a long time.Along with constantly developing of the raising of chip integration and performance, embedded system, these unfavorable factors are eased gradually and eliminate, and have greatly promoted the development of speech recognition product.In built-in speech recognition system, generally adopt little vocabulary speech recognition system to meet the demands, this type systematic can be discerned the voice of tens keywords usually.The method of a kind of updating embedded speech identification system function proposed by the invention is promptly at this type systematic.
In built-in speech recognition system, usually the function of system is relatively-stationary, promptly to certain specific Embedded Speech Recognition System product, keyword that it can be discerned and quantity thereof, to the response of different recognition results, processing mode etc. when system makes mistakes, in the whole use of product, remain unchanged substantially.For the fixing voice product of this systemic-function, the user tends to feel shortcomings such as product function is stiff, the man machine language interface is unfriendly in long-term use, even produces and be weary of psychology, is unfavorable for the expansion of voice produce market.
Along with the widespread use of speech recognition technology on embedded system,, can not only improve user experience, obtain user's favorable comment, and can bring added value to product if many products can allow the user that product is carried out renewal on the function.For example, in education product, toy for children,, then can bring dirigibility in the use to the user if can the audio file (as sound story, juvenile song, text recording etc.) of system be upgraded; If can the keyword that system can discern be upgraded, then the user can be provided with new voice command according to the preference of oneself, even adopts dialect to carry out man-machine interaction; If can upgrade the various prompt tones of system (as startup music, background music etc. when reading aloud text), then can allow the user that product is had personalized setting.Simultaneously, because the strong interest that the user upgrades systemic-function will stimulate and improve the association area network resources demand, therefore can provide speech data relevant service on the internet.
Therefore, how can allow the user that the function of built-in speech recognition system is upgraded, become the topic that the people in the industry is concerned about.
Summary of the invention
The object of the present invention is to provide a kind of method of updating embedded speech identification system function, utilize the method, can keep allowing the user that systemic-function is upgraded under the constant situation of speech recognition program, and then improve the user experience of product, the value of lifting product.
The present invention mainly towards application be the embedded system that has embedded OS, possesses speech identifying function, this system by external memory interface and (or) mode of network realizes function renewal method provided by the invention.
In order to achieve the above object, technical scheme of the present invention is as follows:
A kind of method of updating embedded speech identification system function comprises step:
1) starts speech recognition system, and, initiate the function renewal request according to system's setting or user's request;
2) before startup is upgraded, stop speech recognition program earlier, back up all data that speech recognition program uses then, and obtain more new data by internet or local file system;
3) the more new data that obtains is carried out after the data correctness verification decision and whether adopt more new data,, then declare the function renewal failure if more new data makes a mistake, and according to the original function of Backup Data recovery system; Otherwise, enter next step;
4) after Data Update was finished, system carried out speech recognition training according to new data more, preserved training result and more after the new configuration file, system promptly obtains new function, declares the function renewal success simultaneously.
Described step 1) adopts including but not limited to following several modes or its array configuration and initiates the function renewal request: the user initiates function renewal request mode immediately by configuration file set-up function update mode, the automatic configuration feature update mode of system, user.
Described step 2) purpose is to obtain more new data, specifically comprises step:
21) stop all examples of speech recognition program, or only stop wherein the program of conflicting and the operation of module being arranged with function renewal;
21) stop all examples of speech recognition program, or only stop wherein the program of conflicting and the operation of module being arranged with function renewal;
22) back up all data that speech recognition program uses, or all data that in the function renewal process, may be modified, delete;
22) select according to the user or the update mode of configuration file, search current available function renewal, and by the user select or according to configuration file automatically decision wait the function deleting, upgrade or increase;
23) if adopt the mode of network to upgrade, then send file transfer requests to server by network; If adopt the mode of local file system to upgrade, then send the file access request to the local operation system; Net result is to obtain the required file of function renewal.
The purpose of described step 3) is to guarantee that the more new data packets of being obtained is correct, in order to avoid occur unusually in the renewal process of back, its detailed process comprises step:
31) calculate the proof test value of all updating files, and compare with the proof test value of downloading with updating file, if the inconsistent situation of generation is arranged, then declare the function renewal failure and use Backup Data to restore the system to the preceding state of renewal, deletion Backup Data after recovering to finish;
32) check whether the content of all updating files and form be legal,, then declare the function renewal failure and use Backup Data to restore the system to the preceding state of renewal, deletion Backup Data after recovering to finish if the inconsistent situation of generation is arranged;
33) according to new data more, revise configuration file, make speech recognition program when starting, continue to finish the Status Flag of function renewal comprising modification.
Described step 4) is to finish the final step of function renewal, its objective is to carry out speech recognition training and upgrade the relevant configuration file, guarantees the function renewal success, and detailed process comprises step:
41) start speech recognition program, or restart owing to the conflict program that is stopped and the operation of module being arranged, and read the configuration file of in step 33, having revised with function renewal;
42) according to configuration file, whether check system is in the state that carries out function renewal, if not, then moves original function; Otherwise, enter next step, continue to finish function renewal;
43) information that provides according to configuration file uses more that new data carries out speech recognition training, and the result who preserves training is to local file system;
44) according to more new data and training result, revise configuration file, show that comprising modification systemic-function upgrades completed Status Flag;
45) Backup Data of creating in the deletion step 22, the speech recognition system software after operation function upgrades then.
The present invention compares with background technology, and the beneficial effect that has is:
The present invention is a kind of method of updating embedded speech identification system function, and this method is used for increasing the renewable ability of function on existing built-in speech recognition system.Utilize this method, can under the constant situation of speech recognition program, be aided with, the function of system is upgraded (comprise and increase function, delete function and modify feature) as external data interfaces such as USB memory device and networks.At present, the system that on built-in speech recognition system, possesses the function renewal characteristics that yet there are no this paper and proposed.Be different from existing speech recognition system, this method has been considered the functional expansionary of built-in speech recognition system effectively, by function renewal packet and configuration file, the existing systems function is made amendment.This method is supported the various systems' (as voice toy, learning machine) that possess speech identifying function, and can be extended to the speech recognition system that possesses multiple function, can satisfy the needs of user in different phase.
Description of drawings
Fig. 1 is the overall flow figure of the updating embedded speech identification system function method of the preferred embodiment of the present invention.
Fig. 2 is the form synoptic diagram of new data packets more in the preferred embodiment of the present invention, and the file of the available xml form of this figure is described.
Fig. 3 is the internal logic synoptic diagram of speech recognition program in the preferred embodiment of the present invention.
Embodiment
The method of the updating embedded speech identification system function that the present invention proposes, obtain more new data by network or other approach such as local file system etc., the information that provides according to new data mainly relies on the voice training functional module of speech recognition program to come prior function is upgraded then.Function renewal as described herein is meant, by the mode of deleting, revising or cover original function is changed, and present embodiment adopts the mode that covers to carry out function renewal.Other modes can be by that analogy.Speech recognition system is divided into continuous speech recognition system and isolated word recognition system according to the continuity of input voice, and isolated word identification is also referred to as key word recognition.Key word recognition is meant, the information result such as proper vector (code book) that comprise each crucial voice that given one group of keyword and precondition thereof are good, when a bit of voice of input, calculate the similarity degree between each feature of this section voice and code book, and judge the most approaching keyword that changes the input voice, this crucial keyword is recognition result.For guaranteeing the reliability of recognition result, often adding some parameters and threshold value in system waits and reduces because the influence that situations such as environmental background noise, speaker's asophia are brought, at this moment the recognition result of importing voice may comprise error message, i.e. this section phonetic entry is not considered to a legal keyword.If the voice of input are considered to some in the keyword set, corresponding process is carried out in predefined response according to this keyword so, and typical operation is that one section voice is play in action according to predefine.Because in man machine language's reciprocal process, therefore the corresponding different action of same keyword possibility, after identifying certain keyword, has necessity recognition result and response condition is carried out record, so that can utilize this information in follow-up identifying.For example, in possessing function children speech toys such as " telling a story ", " singing ", when the double requirement of user " singing ", the potential hope of user is, the music that system play should not be identical, and plays the music of not playing recently as far as possible.Be necessary to write down the music of at every turn being play this moment.
Difference and the contact of Fig. 1 and Fig. 3 are: that Fig. 1 represents only is the overall flow figure of function renewal process in the speech recognition system, and what Fig. 3 represented is the schematic flow diagram of the internal logic flow process of whole speech recognition system; The content of Fig. 1 is the branch of the concrete function of Fig. 3.Because the function renewal process relates to the flow process shown in above two figure, and because its inherent contiguity will alternately be mentioned to these two figure so that the mentioned function renewal method that arrives of the present invention to be described better in embodiment.
Fig. 1 is the overall flow figure of updating embedded speech identification system function method, and behind starting of embedded operation system, the user starts speech recognition system as required or starts this software automatically by system, specifically depends on user interface.Usually, the system that only has speech recognition does not purely need graphic user interface, and the speech recognition software system starts with operating system.If other situations, the user manually boots speech recognition software as required.
After speech recognition system starts, at first check the setting of relevant function renewal in the configuration file, this will have multiple result.The one, function renewal is disabled, directly ignores renewal process this moment and carries out the function of present speech recognition software system.The 2nd, function renewal enable and the current process that is not in function renewal in, judge whether available renewal source according to actual conditions this moment.If embedded system possesses network function, speech recognition system can seek whether available function renewal is arranged at present by the resource on the web search.The 3rd, function renewal enable and the current process that is in function renewal in, promptly must continue uncompleted function renewal process now.For the two kinds of situations in back, its overall procedure is seen shown in Figure 3, and present embodiment will continue to describe in detail to this in the back.
The configuration file that the present invention mentioned is meant the file of preserving with the relevant configuration information of speech recognition system software, and this document has been described the major parameter that speech recognition software uses in the course of the work.The form of configuration file can be xml form or other user-defined formats.For simplicity, present embodiment adopts uses xml form more widely, and other configuration file format should realize same function.To the function renewal of speech recognition, configuration file mainly provides following entry record state: whether function renewal is enabled; The state of function renewal (renewal process finishes, renewal process carry out in); The mode of function renewal (manually, automatically, reaching other relevant settings).
Find that when speech recognition program starts the back residing state is not in the function renewal process, and show that according to relevant the setting can start function renewal the time, system is written into also operation function update module.After the function renewal module started, at first the mode of audit function renewal generally had two kinds of selections: network renewal and local file system renewal.When selecting network to upgrade in the configuration file, update module is checked the renewal that each the renewal source in the configuration file can be used successively, and compares with the version number of current speech recognition system etc., and judges the current available more new data packets that whether has.If there is not more new data packets, then the speech recognition program that before had been stopped is carried out in hold function update module and continuation.When finding available renewal is arranged, allow the user select (method of selection can be the simple and fixing speech recognition process of graphic user interface, hardware user interface or function).If the function renewal mode that adopts is undertaken by local file system, then according to the path at the specified place, renewal source of configuration file or path (generally being applicable to the embedded system that the has graphical user interface) search of the interim appointment of user whether available renewal is arranged, process afterwards is the same.
After detecting available function renewal and user and confirming to enter renewal process, function renewal module check also stops all and speech recognition process or the relevant example of thread.This can realize by following process: module that storage in advance is relevant with functional modules such as speech recognition, function renewals in the configuration file of speech recognition system and corresponding file; Searching conflicts module with function renewal when, search one by one according to these records in the configuration file.
For normally finishing of upgrading of assurance function and when the function renewal process makes a mistake, can in time revert to virgin state, need relevant file be backed up.Need the file of backup to mainly contain: speech recognition program and function renewal module itself, the All Files (as audio file) relevant and the configuration file of speech recognition system with original function.A kind of reliable but inefficient methodology is all files of backup, but does not adopt the method in the present embodiment.After the backup, should in configuration file, add the information of relevant backup file, comprise path, proof test value, the BACKUP TIME point at backup file place, so that can state before upgrading take place correctly to return to when unusual in renewal process.
After obtaining update package, need carry out the data correctness verification to new data packets more.Data correctness has two layers of meaning, and the one, the correctness of data transmission promptly obtains mistake on more whether new data packets is sent out transmission course from network or other positions; The 2nd, the correctness of new packet format and content more, promptly whether each file meets predetermined form.
As shown in Figure 2, be a kind of organizational form of new data packets more, each more new data packets form by the files of a plurality of use xml forms.After obtaining more new data packets, at first open data check file wherein, this document has write down the filename and the corresponding proof test value of each file in the new data packets more, and wherein each file is carried out the verification of data transmission correctness.If in the process of verification mistake takes place, think that then the more new data packets of being obtained is wrong, should stop immediately upgrading.As shown in Figure 1, after judging the renewal error in data, the information (position of Backup Data, proof test value etc.) of the relevant Backup Data that use Backup Data and configuration file provide is recovered original function, and deletion Backup Data, afterwards more in the new configuration file update mode when next speech recognition program starts, normally move so that allow for finishing.
Verification relative complex to more new packet format and content correctness is a little, and its complexity depends on the scale of system and the complicacy of function.As shown in Figure 2, upgrade supporting paper at response mode, more in the file such as new sequences description document and code name definition list, the organizational form of data all has certain regulation.These regulations have constituted the more agreement of new data packets.The content of agreement is formulated according to different application needs, but formulation can not be changed after finishing.This method dirigibility is poor.A kind of better method is that protocol description is put into more in the new data packets, the agreement checking data in the new data packets more when the correctness of each checking data like this, and its cost is the complicacy that has increased the function renewal module.What present embodiment adopted is fixed protocol, does not promptly more comprise the protocol description file in the new data packets, and the equal implicit expression of all agreements is included in the function renewal module.If verification is found the data packet format and content and makes a mistake that then corresponding processing mode is the same.
After confirming that more new data packets correctly, more new data packets is under certain catalogue for decompress(ion), and more the information of new data packets is put in the configuration file then, can carry out voice training according to these information after making.The information that need write configuration file mainly contains: the more state of path, place of All Files and uses thereof identification code, function renewal in the new data packets (the modification state is for just in renewal process), the path at voice training program place and the input parameter of program.
Afterwards, function renewal program start speech recognition program and stop self.As shown in Figure 3, be in the function renewal state this moment, enters the "Yes" conditional branching.At this moment,, therefore no longer repeat the correctness verification, directly enter the voice training stage owing to new data packets has more been carried out verification before the function renewal module.In general, in the performance history of speech recognition system, these two functions of voice training and speech recognition are developed respectively, and form independently module of function.Some functions can be upgraded as not need keyword is changed, and then can skip speech recognition steps.If function renewal requires to adjust key word (comprise interpolation, deletion, revise), then need affected keyword set is carried out voice training.
After voice training is finished, should be with the file storage training result of certain form.Arrive this, the process of voice training finishes.Next step, must create (function covers and upgrades) or revise (funtion part renewal) function response file description list in strict accordance with the requirement that response mode in the new data packets more upgrades supporting paper.For example, after increasing a keyword, should take which kind of type action and specifically which type of action after should specifying this keyword identified, these actions are represented (seeing that response mode upgrades supporting paper among Fig. 2) with respective rule sign indicating number and respective rule descriptor.
Example file form among Fig. 2 has related to multiple code names such as group ID under the keyword ID, keyword, respond style sign indicating number, rule of response descriptor code, action type sign indicating number, operational code, and these code names are relevant with the system realization, are included in the function renewal agreement.More the new sequences description list has indicated order and step that renewal process should be taked; Response mode upgrades instruction card and has indicated after the keyword training is finished, and which renewal the configuration relevant with the keyword response mode carries out in the configuration file of function renewal program reply speech recognition system.For example, during keyword " singing " corresponding response was described, its respond style remained unchanged, and promptly is the broadcast a piece of music, but can change take place for the quantity of the music of playing, and then should be able to reflect these renewals in the configuration file of speech recognition system.
The final step of function renewal is to revise configuration file, update mode is set for finishing, and deletes institute's backed up data before function renewal.During design was used, Backup Data can remain according to user's requirement, for example stores under certain catalogue, so that can return to old function later on.At this moment, the function renewal process ends, and restarts speech recognition system and can obtain new function.
Illustrate and described these embodiment for the purpose that function of the present invention and structural principle are shown, and under the situation that does not break away from such principle, can make amendment it.Therefore, the present invention includes the interior all modifications of spirit and scope that is encompassed in claims.
Claims (5)
1. the method for a updating embedded speech identification system function is characterized in that, comprises step:
1) starts speech recognition system, and, initiate the function renewal request according to system's setting or user's request;
2) before startup is upgraded, stop speech recognition program earlier, back up all data that speech recognition program uses then, and obtain more new data by internet or local file system;
3) the more new data that obtains is carried out after the data correctness verification decision and whether adopt more new data,, then declare the function renewal failure if more new data makes a mistake, and according to the original function of Backup Data recovery system; Otherwise, enter next step;
4) after Data Update was finished, system carried out speech recognition training according to new data more, preserved training result and more after the new configuration file, system promptly obtains new function, declares the function renewal success simultaneously.
2. the method for a kind of updating embedded speech identification system function according to claim 1 is characterized in that: described step 1) adopts including but not limited to following several modes or its array configuration and initiates the function renewal request: the user initiates function renewal request mode immediately by configuration file set-up function update mode, the automatic configuration feature update mode of system, user.
3. the method for a kind of updating embedded speech identification system function according to claim 1 is characterized in that, described step 2) comprising:
21) stop all examples of speech recognition program, or only stop wherein the program of conflicting and the operation of module being arranged with function renewal;
22) back up all data that speech recognition program uses, or all data that in the function renewal process, may be modified, delete;
22) select according to the user or the update mode of configuration file, search current available function renewal, and by the user select or according to configuration file automatically decision wait the function deleting, upgrade or increase;
23) if adopt the mode of network to upgrade, then send file transfer requests to server by network; If adopt the mode of local file system to upgrade, then send the file access request to the local operation system; Net result is to obtain the required file of function renewal.
4. the method for a kind of updating embedded speech identification system function according to claim 1 is characterized in that, described step 3) comprises:
31) calculate the proof test value of all updating files, and compare with the proof test value of downloading with updating file, if the inconsistent situation of generation is arranged, then declare the function renewal failure and use Backup Data to restore the system to the preceding state of renewal, deletion Backup Data after recovering to finish;
32) check whether the content of all updating files and form be legal,, then declare the function renewal failure and use Backup Data to restore the system to the preceding state of renewal, deletion Backup Data after recovering to finish if the inconsistent situation of generation is arranged;
33) according to new data more, revise configuration file, make speech recognition program when starting, continue to finish the Status Flag of function renewal comprising modification.
5. the method for a kind of updating embedded speech identification system function according to claim 1 is characterized in that, described step 4) comprises:
41) start speech recognition program, or restart owing to the conflict program that is stopped and the operation of module being arranged, and read the configuration file of in step 33, having revised with function renewal;
42) according to configuration file, whether check system is in the state that carries out function renewal, if not, then moves original function; Otherwise, enter next step, continue to finish function renewal;
43) information that provides according to configuration file uses more that new data carries out speech recognition training, and the result who preserves training is to local file system;
44) according to more new data and training result, revise configuration file, show that comprising modification systemic-function upgrades completed Status Flag;
45) Backup Data of creating in the deletion step 22, the speech recognition system software after operation function upgrades then.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2008101624045A CN101409074A (en) | 2008-11-24 | 2008-11-24 | Method for updating embedded speech identification system function |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2008101624045A CN101409074A (en) | 2008-11-24 | 2008-11-24 | Method for updating embedded speech identification system function |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101409074A true CN101409074A (en) | 2009-04-15 |
Family
ID=40572054
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA2008101624045A Pending CN101409074A (en) | 2008-11-24 | 2008-11-24 | Method for updating embedded speech identification system function |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101409074A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2010142254A1 (en) * | 2009-06-12 | 2010-12-16 | 华为技术有限公司 | Speech identification training processing, controlling method and device |
| US8504024B2 (en) | 2009-05-27 | 2013-08-06 | Huawei Technologies Co., Ltd. | Method for implementing an intelligent service and communications system |
| CN103472994A (en) * | 2013-09-06 | 2013-12-25 | 乐得科技有限公司 | Operation control achieving method, device and system based on voice |
| CN104346239A (en) * | 2014-06-27 | 2015-02-11 | 国家电网公司 | Method and device for recovering anomaly of application program in embedded system |
| CN105307195A (en) * | 2015-09-22 | 2016-02-03 | 广东欧珀移动通信有限公司 | Method and device for updating access point name (APN) parameter |
| US9535579B2 (en) | 2012-10-09 | 2017-01-03 | International Business Machines Corporation | Keyword-based user interface in electronic device |
| CN106484486A (en) * | 2016-10-31 | 2017-03-08 | 乐视控股（北京）有限公司 | A kind of application program update processing method and processing device |
| CN106796786A (en) * | 2014-09-30 | 2017-05-31 | 三菱电机株式会社 | Speech recognition system |
| CN108509215A (en) * | 2018-03-22 | 2018-09-07 | 广州视源电子科技股份有限公司 | System software replacing method and device, terminal equipment and storage medium |
| CN109754786A (en) * | 2017-11-06 | 2019-05-14 | 奥迪股份公司 | Phonetic controller for vehicle |
-
2008
- 2008-11-24 CN CNA2008101624045A patent/CN101409074A/en active Pending
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8504024B2 (en) | 2009-05-27 | 2013-08-06 | Huawei Technologies Co., Ltd. | Method for implementing an intelligent service and communications system |
| US8909533B2 (en) | 2009-06-12 | 2014-12-09 | Huawei Technologies Co., Ltd. | Method and apparatus for performing and controlling speech recognition and enrollment |
| WO2010142254A1 (en) * | 2009-06-12 | 2010-12-16 | 华为技术有限公司 | Speech identification training processing, controlling method and device |
| US9582159B2 (en) | 2012-10-09 | 2017-02-28 | International Business Machines Corporation | Keyword-based user interface in electronic device |
| US10365806B2 (en) | 2012-10-09 | 2019-07-30 | International Business Machines Corporation | Keyword-based user interface in electronic device |
| US9535579B2 (en) | 2012-10-09 | 2017-01-03 | International Business Machines Corporation | Keyword-based user interface in electronic device |
| CN103472994A (en) * | 2013-09-06 | 2013-12-25 | 乐得科技有限公司 | Operation control achieving method, device and system based on voice |
| CN103472994B (en) * | 2013-09-06 | 2017-02-08 | 网易乐得科技有限公司 | Operation control achieving method, device and system based on voice |
| CN104346239A (en) * | 2014-06-27 | 2015-02-11 | 国家电网公司 | Method and device for recovering anomaly of application program in embedded system |
| CN104346239B (en) * | 2014-06-27 | 2018-06-05 | 国家电网公司 | The abnormal restoring method and device of application program in embedded system |
| CN106796786A (en) * | 2014-09-30 | 2017-05-31 | 三菱电机株式会社 | Speech recognition system |
| CN105307195B (en) * | 2015-09-22 | 2019-02-19 | Oppo广东移动通信有限公司 | A kind of method and device updating access point name parameter |
| CN105307195A (en) * | 2015-09-22 | 2016-02-03 | 广东欧珀移动通信有限公司 | Method and device for updating access point name (APN) parameter |
| CN106484486A (en) * | 2016-10-31 | 2017-03-08 | 乐视控股（北京）有限公司 | A kind of application program update processing method and processing device |
| CN109754786A (en) * | 2017-11-06 | 2019-05-14 | 奥迪股份公司 | Phonetic controller for vehicle |
| CN108509215A (en) * | 2018-03-22 | 2018-09-07 | 广州视源电子科技股份有限公司 | System software replacing method and device, terminal equipment and storage medium |
| CN108509215B (en) * | 2018-03-22 | 2022-03-29 | 广州视源电子科技股份有限公司 | System software replacing method and device, terminal equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101409074A (en) | Method for updating embedded speech identification system function | |
| CN109983460B (en) | Services for developing dialog-driven applications | |
| US7899673B2 (en) | Automatic pruning of grammars in a multi-application speech recognition interface | |
| US7734461B2 (en) | Apparatus for providing voice dialogue service and method of operating the same | |
| RU2525440C2 (en) | Markup language-based selection and utilisation of recognisers for utterance processing | |
| KR101418163B1 (en) | Speech recognition repair using contextual information | |
| JP2019159309A (en) | Method and apparatus for determining speech interaction satisfaction | |
| US10957314B2 (en) | Developer platform for providing automated assistant in new domains | |
| JP2008514983A (en) | Interactive conversational conversations of cognitively overloaded users of devices | |
| CN111312230B (en) | Voice interaction monitoring method and device for voice conversation platform | |
| US11315547B2 (en) | Method and system for generating speech recognition training data | |
| CN113421561B (en) | Voice control method, voice control device, server, and storage medium | |
| US11615788B2 (en) | Method for executing function based on voice and electronic device supporting the same | |
| US8315874B2 (en) | Voice user interface authoring tool | |
| CN108831444B (en) | Semantic resource training method and system for voice conversation platform | |
| CN114678027A (en) | Error correction method and device for voice recognition result, terminal equipment and storage medium | |
| Lee | Voice user interface projects: build voice-enabled applications using dialogflow for google home and Alexa skills kit for Amazon Echo | |
| KR102175539B1 (en) | Conversational service apparatus and method based on user utterance | |
| JP2021108095A (en) | Method for outputting information on analysis abnormality in speech comprehension | |
| CN110447026B (en) | Developer platform for providing automated assistant in new domain | |
| CN112905869B (en) | Self-adaptive training method, device, storage medium and equipment for language model | |
| CN116108918A (en) | Training method and related device for dialogue pre-training model | |
| JP7448240B2 (en) | Efficient dialogue structure | |
| CN113407689A (en) | Method and device for model training and business execution | |
| CN115171672A (en) | Voice processing method, device and computer storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Open date: 20090415 |