CN107133222A - A kind of real-time language conversion equipment and conversion method based on heterogeneous framework - Google Patents
A kind of real-time language conversion equipment and conversion method based on heterogeneous framework Download PDFInfo
- Publication number
- CN107133222A CN107133222A CN201710248675.1A CN201710248675A CN107133222A CN 107133222 A CN107133222 A CN 107133222A CN 201710248675 A CN201710248675 A CN 201710248675A CN 107133222 A CN107133222 A CN 107133222A
- Authority
- CN
- China
- Prior art keywords
- language
- framework
- real
- equipment
- user
- 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
- 238000006243 chemical reaction Methods 0.000 title claims abstract description 42
- 238000000034 method Methods 0.000 title claims abstract description 20
- 238000013519 translation Methods 0.000 claims abstract description 19
- 230000015572 biosynthetic process Effects 0.000 claims abstract description 7
- 238000003786 synthesis reaction Methods 0.000 claims abstract description 7
- 238000012937 correction Methods 0.000 claims abstract description 4
- 230000007935 neutral effect Effects 0.000 claims description 21
- 210000002569 neuron Anatomy 0.000 claims description 19
- 238000013528 artificial neural network Methods 0.000 claims description 17
- 238000012545 processing Methods 0.000 claims description 9
- 230000004044 response Effects 0.000 claims description 6
- 238000004891 communication Methods 0.000 claims description 4
- 230000005611 electricity Effects 0.000 claims description 4
- 230000003139 buffering effect Effects 0.000 claims description 3
- 230000005284 excitation Effects 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 230000006870 function Effects 0.000 claims description 3
- 239000000203 mixture Substances 0.000 claims description 3
- 230000000694 effects Effects 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000012790 confirmation Methods 0.000 description 2
- 230000001133 acceleration Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 239000000686 essence Substances 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000004218 nerve net Anatomy 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
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
- G06F40/42—Data-driven translation
- G06F40/47—Machine-assisted translation, e.g. using translation memory
-
- 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
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
- G10L15/00—Speech recognition
- G10L15/08—Speech classification or search
- G10L15/16—Speech classification or search using artificial neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Machine Translation (AREA)
Abstract
The invention discloses a kind of real-time language conversion equipment and conversion method based on heterogeneous framework, input language and the numbering of output language that language transfer method is specified including acquisition user, user inputs voice by equipment, equipment is identified, and be shown in above screen, user judges whether the recognition result of display is correct, to incorrect carry out manual correction, equipment is according to recognition result or the word of input, machine translation is carried out based on database, equipment includes the result of translation on screen, or based on synthesis speech database, synthesis voice is exported by voice-output device.Basic language translation database is placed in apparatus of the present invention by the present invention, selection according to user to required translation languages every time, required database is put into system, its heterogeneous framework can be based on, pass through the high-speed computation of processor, the characteristics of quickly being handled parallel by FPGA field programmable processors again, is rapidly completed the translation of voice.
Description
Technical field
The invention belongs to language data processing technology field, more particularly to a kind of real-time language conversion based on heterogeneous framework
Device and conversion method.
Background technology
As China's economic increasingly incorporates the development of global economic integration so that it is transnational, across language exchange it is more next
More, such as in international conference, people need to carry out increasingly frequently international exchange with various countries, are not only the friendship of English
Stream, in addition to Portuguese, German, French, the multilingual such as Russian, although major part people, which try one's best, at present switchs to English and remove ditch
It is logical, but communication does not reach preferable effect still, causes daily exchanging meeting to be had any problem.
For the deficiency in above-mentioned use, one kind is deployment simultaneous interpretation personnel, but be so only suitable for meeting at the scene
Close, can not be used freely exchanging occasion；Another way is to be based on cloud service, inputs speech data by mobile terminal, so
It will be noted afterwards on speech data in Cloud Server, the identification of language completed by Cloud Server, and pass through network data library lookup
Corresponding contents complete translation, this by way of cloud service is translated, and the quality of network produces considerable influence to using, when
During unstable networks, easily cause user wait delay, influence user's using effect, it is impossible to meet user individual demand and
Experience.
Technology in relatively existing real-time language processing, the real-time language conversion equipment can flexibly realize multilingual
Conversion in real time, while the demand to simultaneous interpretation personnel is broken away from, while the support without network cloud service, solves network quality
The influence to live effect brought, and realized using heterogeneous framework to speech data acceleration processing, it is ensured that language data
Low latency and in real time conversion.
In summary, the problem of prior art is present be：When unstable networks, easily cause user and wait delay, shadow
Ring user's using effect, it is impossible to meet demand and the experience of user individual.
The content of the invention
To solve the problem of prior art is present, the present invention provides a kind of real-time language conversion equipment based on heterogeneous framework
And conversion method.
The present invention is achieved in that a kind of real-time language conversion method based on heterogeneous framework, described to be based on heterogeneous frame
The real-time language conversion method of structure comprises the following steps：
Step one, equipment acquisition user specifies input language and the numbering of output language；
Step 2, user inputs voice by equipment, and equipment is identified, and is shown in above screen；
Step 3, user judges whether the recognition result of display is correct, to incorrect carry out manual correction；
Step 4, equipment carries out machine translation according to recognition result or the word of input based on database；
Step 5, equipment includes the result of translation on screen, or based on synthesis speech database, will synthesize voice
Exported by voice-output device.
Further, in step one, language conversion database is called in into internal memory after equipment acquisition instruction, and FPGA is carried out
Configuration, builds in programmable FPGA and completes polytype multilayer neural network structure, and the parameters such as wherein weights are carried out
Initialization；
Described build completes polytype multilayer neural network structure, specifically includes：
Many node layers, and the interconnecting relation between node can be built in programming device, while in memory chip
Middle storage weights, each layer includes several neurons, each neuron threshold θ comprising onej, for changing neuron
Activity；Camber line W in networkijRepresent the weights between preceding layer neuron and later layer neuron for hidden layer and output layer
The each node of input, such as following formula：
By carrying out convolution to upper layer data, and the response that excitation function f () carrys out simulative neural network signal is set up, can
The response to upper strata neurode is completed in programmed logic device.
Further, in step 2, the speech data of input is identified based on heterogeneous framework neutral net for equipment；
Described be identified specifically includes：
Directly basic neural network parameter is called in memory chip by CPU, internal memory is read by isomery framework
The default parameter of these in chip quickly carries out the Information Communication of neutral net, is rapidly completed by FPGA parallel processing capability
Convolution algorithm, constantly adjusts variable parameter, while which part weights variable storage is in outside by CPU operation some algorithms
Quickly read and update in memory chip and by PLD, finally in output layer output characteristic value, so as to realize
The feature recognition and extraction of language data；The convolution algorithm formula is：
Further, step 3, inputs the recognition result that speech data and user confirm, based on heterogeneous framework according to user
Neutral net inputs speech data to user and the recognition result of user's confirmation learns, and adjusting parameter, so that constantly
Adapt to personal characteristic voice.
It is described to be learnt, and adjusting parameter, specifically include：
The correct result that system confirms according to user obtains trained values Tj, to neutral net back propagation learning, so as to adjust
Whole neural network weight, each unit j of output layer error amount can be as follows,
Errj=Oj(1-Oj)*(Tj-Oj),
It is as follows for each unit j error amounts of hidden layer by last to first hidden layer
Errj=Oj(1-Oj)*∑kErrk*Wij,
Each weighed value adjusting amount and weighed value adjusting are as follows in network, pass through
ΔWij=(l) * Errj*Oj,
Wij=Wij+ΔWij。
Further, in step 4, user directly calls in text information by interface, and the text information includes text, electricity
Sub- mail, telecine caption information or browses info web.
Another object of the present invention is to provide a kind of real-time language conversion equipment based on heterogeneous framework, described based on heterogeneous
The real-time language conversion equipment of framework is used based on processor circuit, a kind of extensive the of field programmable logic composition of FPGA
In the heterogeneous framework of quick Processing Algorithm, and build neural network structure；Specifically include：
Processor, is connected with programmable logic chip, according to database, for updating programmable logic chip；
Programmable logic chip, is connected with processor, for building corresponding neutral net framework and parameter；
Memory chip, is connected with processor, for storing comprising multilingual identification storehouse, the grammatical change data of different language
The multilingual conversion library file in storehouse and multiple voice synthetic model storehouse；
First memory chip, is connected with processor, is used for, as the running space of processor program, while also will caching
Partial data in language identification storehouse and language grammar conversion database.
Second memory chip, is connected with programmable logic chip, the language data for buffering input, while importing language
Storehouse data are recognized, the storehouse can be with the parameter of each several part neutral net in rapid configuration PLD, while in the internal memory
The intermediate data of each layer in neutral net is stored in chip.
Language/word input module, is connected with processor, includes text document, text, Email, electricity for inputting
Radio movies caption information or the text information for browsing webpage；
Broadcasting/display module, is connected with processor, for receiving voice document or text file, and voice document is carried out
Recognize and shown；
Expansion interface, is connected with processor, for connecting external computer；
Further, broadcasting/display module includes player, is connected with processor, for by player by target language
Say that information is broadcasted.
Basic language translation database is placed in apparatus of the present invention memory chip by the present invention, every time according to user to institute
The selection of languages need to be translated, required database is put into the storage chip of system, the present invention can be based on its heterogeneous framework, lead to
The characteristics of crossing the high-speed computation of processor, then quickly handled parallel by FPGA field programmable processors, is rapidly completed voice
Translation.
Heterogeneous framework of the invention can support nerual network technique simultaneously, can constantly be optimized simultaneously according to the feedback of user
Language translation database is updated, the raising of accuracy rate is realized.
The present invention disclose one kind and can not handled by network cloud, can the dress changed of the multilingual real-time language of complete independently
Put, while the device is based on heterogeneous framework, disposal ability is strong, supports nerual network technique, can be improved constantly by self-teaching
Conversion accuracy.User directly can be expressed using mother tongue even place spoken language, and Ben Aming can change it in real time
To object language and display or voice output, while the present invention is flexible to operation, user can be also confirmed after identification by showing
Word it is whether correct, so by learning case, promote this equipment to be learnt by man-machine interaction, thus come adapt to use
The characteristic voice at family, such as spoken, word speed, the difference of intonation.
Brief description of the drawings
Fig. 1 is the real-time language conversion method flow chart provided in an embodiment of the present invention based on heterogeneous framework.
Fig. 2 is the real-time language conversion equipment schematic diagram provided in an embodiment of the present invention based on heterogeneous framework.
In figure：1st, processor；2nd, programmable logic chip；3rd, memory chip；4th, the first memory chip；5th, the second internal memory core
Piece；6th, language/word input module；7th, broadcasting/display module；8th, expansion interface.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, with reference to embodiments, to the present invention
It is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not used to
Limit the present invention.
The present invention is described in detail below in conjunction with the accompanying drawings.
As shown in figure 1, the real-time language conversion method provided in an embodiment of the present invention based on heterogeneous framework, including following step
Suddenly：
S101:Input language and the numbering of output language that equipment acquisition user specifies；
S102:User inputs voice by equipment, and equipment is identified, and is shown in above screen；
S103:User judges whether the recognition result of display is correct, to incorrect carry out manual correction；
S104:Equipment carries out machine translation according to recognition result or the word of input based on database；
S105:Equipment includes the result of translation on screen, or based on synthesis speech database, synthesis voice is led to
Cross voice-output device output.
In step one, language conversion database is called in internal memory by equipment after obtaining instruction, and FPGA is configured,
Built in programmable FPGA and complete polytype multilayer neural network structure, and the parameters such as wherein weights are initialized；
Described build completes polytype multilayer neural network structure, specifically includes：
Many node layers, and the interconnecting relation between node can be built in programming device, while in memory chip
Middle storage weights, each layer includes several neurons, each neuron threshold θ comprising onej, for changing neuron
Activity；Camber line W in networkijRepresent the weights between preceding layer neuron and later layer neuron for hidden layer and output layer
The each node of input, such as following formula：
By carrying out convolution to upper layer data, and the response that excitation function f () carrys out simulative neural network signal is set up, can
The response to upper strata neurode is completed in programmed logic device.
In step 2, the speech data of input is identified based on heterogeneous framework neutral net for equipment；It is described to be known
Do not specifically include：
Directly basic neural network parameter is called in memory chip by CPU, internal memory is read by isomery framework
The default parameter of these in chip quickly carries out the Information Communication of neutral net, is rapidly completed by FPGA parallel processing capability
Convolution algorithm, constantly adjusts variable parameter, while which part weights variable storage is in outside by CPU operation some algorithms
Quickly read and update in memory chip and by PLD, finally in output layer output characteristic value, so as to realize
The feature recognition and extraction of language data；The convolution algorithm formula is：
Step 3, inputs the recognition result that speech data and user confirm, based on heterogeneous framework nerve net according to user
Network inputs speech data to user and the recognition result of user's confirmation learns, and adjusting parameter, so as to constantly adapt to individual
People's characteristic voice.
In step 4, user directly calls in text information by interface, and the text information includes text, Email,
Telecine caption information browses info web.
As shown in Fig. 2 the real-time language conversion equipment provided in an embodiment of the present invention based on heterogeneous framework is based on processor
A kind of heterogeneous framework for quick Processing Algorithm of the extensive field programmable logic composition of circuit, FPGA, and build
Play neural network structure；Specifically include：
Processor 1, is connected with programmable logic chip, according to database, for updating programmable logic chip；
Programmable logic chip 2, is connected with processor, for building corresponding neutral net framework and parameter；
Memory chip 3, is connected with processor, for storing comprising multilingual identification storehouse, different language syntax conversion number
Library file is changed according to the multilingual in storehouse and multiple voice synthetic model storehouse；
First memory chip 4, is connected with processor, as the running space of processor program, while will also cache language
Recognize the partial data in storehouse and language grammar conversion database.
Second memory chip 5, is connected with PLD, the language data for buffering input, while importing language
Speech identification storehouse data, the storehouse can be with the parameter of each several part neutral net in rapid configuration PLD, while interior at this
Deposit the intermediate data that each layer in neutral net is stored in chip.
Language/word input module 6, is connected with processor, includes text document, text, Email, electricity for inputting
Radio movies caption information or the text information for browsing webpage；
Broadcasting/display module 7, is connected with processor, for receiving voice document or text file, and voice document is carried out
Recognize and shown；
Expansion interface 8, is connected with processor, for connecting external computer.
Broadcasting/display module includes player, is connected with processor, for being broadcast target-language information by player
Go out.
Basic language translation database is placed in apparatus of the present invention by the present invention, every time according to user to required translation language
The selection planted, required database is put into system, and the present invention can be based on its heterogeneous framework, and the high speed that pass through processor is transported
The characteristics of calculating, then quickly handled parallel by FPGA field programmable processors, is rapidly completed the translation of voice.
Heterogeneous framework of the invention can support nerual network technique simultaneously, can constantly be optimized simultaneously according to the feedback of user
Language translation database is updated, the raising of accuracy rate is realized.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all essences in the present invention
Any modifications, equivalent substitutions and improvements made within refreshing and principle etc., should be included in the scope of the protection.
Claims (7)
1. a kind of real-time language conversion method based on heterogeneous framework, it is characterised in that the real-time language based on heterogeneous framework
Speech conversion method comprises the following steps：
Step one, user specifies the species of input language and output language to number by user interface；
Step 2, user inputs voice by equipment, and equipment is identified, and is shown in above screen；
Step 3, user judges whether the recognition result of display is correct, to incorrect carry out manual correction；
Step 4, equipment carries out machine translation according to recognition result or the word of input based on database；
Step 5, equipment includes the result of translation on screen, or based on synthesis speech database, synthesis voice is passed through
Voice-output device is exported.
2. the real-time language conversion method as claimed in claim 1 based on heterogeneous framework, it is characterised in that in step one,
Language conversion database is called in internal memory by equipment after obtaining instruction, and FPGA is configured, and has been built in programmable FPGA
Initialized into polytype multilayer neural network structure, and to parameters such as wherein weights；
Described build completes polytype multilayer neural network structure, specifically includes：
Many node layers, and the interconnecting relation between node can be built in programming device, while being deposited in memory chip
Weights are stored up, each layer includes several neurons, each neuron threshold θ comprising onej, for changing the work of neuron
Property；Camber line W in networkijThe weights between preceding layer neuron and later layer neuron are represented, for hidden layer and output layer
The each node of input, such as following formula：
By carrying out convolution to upper layer data, and the response that excitation function f () carrys out simulative neural network signal is set up, programmable
The response to upper strata neurode is completed in logical device.
3. the real-time language conversion method as claimed in claim 1 based on heterogeneous framework, it is characterised in that in step 2, if
It is standby that the speech data of input is identified based on heterogeneous framework neutral net.
Described be identified specifically includes：
Directly basic neural network parameter is called in memory chip by CPU, to form flexible by being configured to FPGA
Based on isomery framework neural network structure, such neutral net has complicated multilayer convolution framework, with the isomery frame postponed
Structure reads the Information Communication that these default parameters in memory chip quickly carry out neutral net, by FPGA parallel processing energy
Power is rapidly completed convolution algorithm, variable parameter is constantly adjusted by CPU operation some algorithms, while which part weights variable is deposited
Store up in outside memory chip, and quickly read and and the hidden layer of internal multilayer is carried out more by PLD
Newly, whereinThe weights between l layers of neuron and later layer neuron are represented,Represent what l layers of each neuron were included
Threshold value,L layers of neuron value are represented,The last characteristic value exported in output layer is represented, language can be realized by the value
The feature recognition and extraction of data；The convolution algorithm formula is：
4. the real-time language conversion method as claimed in claim 1 based on heterogeneous framework, it is characterised in that step 3, according to
User inputs the recognition result that speech data and user confirm, speech data is inputted to user based on heterogeneous framework neutral net
And the recognition result that user confirms is learnt, and adjusting parameter, so as to constantly adapt to personal characteristic voice.
It is described to be learnt, and adjusting parameter, specifically include：
The correct result that system confirms according to user obtains trained values TjAnd output valve Oj, to neutral net back propagation learning,
So as to adjust neural network weight, each unit j of output layer error value E rrjIt is as follows：
Errj=Oj(1-Oj)*(Tj-Oj),
It is as follows for each unit j error amounts of hidden layer by last to first hidden layer：
Errj=Oj(1-Oj)*∑kErrk*Wij,
Each weighed value adjusting amount Δ W in networkijIt is as follows：
ΔWij=(l) * Errj*Oj,
Result W is updated by weighed value adjustingijIt is as follows：
Wij=Wij+ΔWij。
5. the real-time language conversion method as claimed in claim 1 based on heterogeneous framework, it is characterised in that in step 4, is used
Text information is directly called in family by interface, and the text information includes text, Email, telecine caption information or clear
Look at info web.
6. a kind of real-time language based on heterogeneous framework of the real-time language conversion method based on heterogeneous framework as claimed in claim 1
Say conversion equipment, it is characterised in that it is big that the real-time language conversion equipment based on heterogeneous framework is based on processor circuit, FPGA
A kind of heterogeneous framework for quick Processing Algorithm of scale on-site PLD composition, and build neutral net knot
Structure；Specifically include：
Processor, is connected with programmable logic chip, according to database, for updating programmable logic chip；
Programmable logic chip, is connected with processor, for building corresponding neutral net framework and parameter；
Memory chip, is connected with processor, for store comprising multilingual identification storehouse, the grammatical conversion database of different language with
And the multilingual conversion library file in multiple voice synthetic model storehouse；
First memory chip, is connected with processor, as the running space of processor program, while also will caching language identification storehouse
With the partial data in language grammar conversion database.
Second memory chip, is connected with programmable logic chip, the language data for buffering input, while importing language identification
Storehouse data, the storehouse can be with the parameter of each several part neutral net in rapid configuration PLD, while in the memory chip
The intermediate data of each layer in middle storage neutral net.
Language/word input module, is connected with processor, includes text document, text, Email, TV electricity for inputting
Shadow caption information or the text information for browsing webpage；
Broadcasting/display module, is connected with processor, and for receiving voice document or text file, voice document is identified
And shown；
Expansion interface, is connected with processor, for connecting external computer.
7. the real-time language conversion equipment as claimed in claim 6 based on heterogeneous framework, it is characterised in that
Broadcasting/display module includes player, is connected with processor, for being broadcasted target-language information by player.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710248675.1A CN107133222A (en) | 2017-04-17 | 2017-04-17 | A kind of real-time language conversion equipment and conversion method based on heterogeneous framework |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710248675.1A CN107133222A (en) | 2017-04-17 | 2017-04-17 | A kind of real-time language conversion equipment and conversion method based on heterogeneous framework |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107133222A true CN107133222A (en) | 2017-09-05 |
Family
ID=59715826
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710248675.1A Pending CN107133222A (en) | 2017-04-17 | 2017-04-17 | A kind of real-time language conversion equipment and conversion method based on heterogeneous framework |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107133222A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108197123A (en) * | 2018-02-07 | 2018-06-22 | 云南衍那科技有限公司 | A kind of cloud translation system and method based on smartwatch |
| CN108735220A (en) * | 2018-04-11 | 2018-11-02 | 四川斐讯信息技术有限公司 | A kind of language learning intelligent earphone, intelligent interactive system and man-machine interaction method |
| CN109686363A (en) * | 2019-02-26 | 2019-04-26 | 深圳市合言信息科技有限公司 | A kind of on-the-spot meeting artificial intelligence simultaneous interpretation equipment |
| CN109997154A (en) * | 2017-10-30 | 2019-07-09 | 上海寒武纪信息科技有限公司 | Information processing method and terminal device |
| CN113178194A (en) * | 2020-01-08 | 2021-07-27 | 上海依图信息技术有限公司 | Voice recognition method and system for interactive hot word updating |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1553378A (en) * | 2003-05-26 | 2004-12-08 | 北京永邦专利技术开发有限责任公司 | Intelligent translating system with self-adaptation studying function |
| US20150193850A1 (en) * | 2014-01-06 | 2015-07-09 | Gil TAMIR | Network-based service matching |
-
2017
- 2017-04-17 CN CN201710248675.1A patent/CN107133222A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1553378A (en) * | 2003-05-26 | 2004-12-08 | 北京永邦专利技术开发有限责任公司 | Intelligent translating system with self-adaptation studying function |
| US20150193850A1 (en) * | 2014-01-06 | 2015-07-09 | Gil TAMIR | Network-based service matching |
Non-Patent Citations (2)
| Title |
|---|
| 臧婷 等: "《中国人工智能进展2007》", 31 December 2007, 北京邮电大学出版社 * |
| 贝特麦克哈贝尔: "《R语言数据挖掘》", 30 November 2016, 北京：机械工业出版社 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109997154A (en) * | 2017-10-30 | 2019-07-09 | 上海寒武纪信息科技有限公司 | Information processing method and terminal device |
| CN108197123A (en) * | 2018-02-07 | 2018-06-22 | 云南衍那科技有限公司 | A kind of cloud translation system and method based on smartwatch |
| CN108735220A (en) * | 2018-04-11 | 2018-11-02 | 四川斐讯信息技术有限公司 | A kind of language learning intelligent earphone, intelligent interactive system and man-machine interaction method |
| CN109686363A (en) * | 2019-02-26 | 2019-04-26 | 深圳市合言信息科技有限公司 | A kind of on-the-spot meeting artificial intelligence simultaneous interpretation equipment |
| CN113178194A (en) * | 2020-01-08 | 2021-07-27 | 上海依图信息技术有限公司 | Voice recognition method and system for interactive hot word updating |
| CN113178194B (en) * | 2020-01-08 | 2024-03-22 | 上海依图信息技术有限公司 | Voice recognition method and system for interactive hotword updating |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107133222A (en) | A kind of real-time language conversion equipment and conversion method based on heterogeneous framework | |
| CN207731321U (en) | Hardware accelerator engine | |
| CN207517054U (en) | Crossfire switchs | |
| Camgoz et al. | Neural sign language translation | |
| CN110263324A (en) | Text handling method, model training method and device | |
| CN111862140B (en) | Panoramic segmentation network and method based on collaborative module level search | |
| CN111583165A (en) | Image processing method, device, equipment and storage medium | |
| WO2021057056A1 (en) | Neural architecture search method, image processing method and device, and storage medium | |
| CN112347248A (en) | Aspect-level text emotion classification method and system | |
| CN110503097A (en) | Training method, device and the storage medium of image processing model | |
| CN108345890A (en) | Image processing method, device and relevant device | |
| CN106062786A (en) | Computing system for training neural networks | |
| CN107491534A (en) | Information processing method and device | |
| US11195098B2 (en) | Method for generating neural network and electronic device | |
| CN112883149B (en) | Natural language processing method and device | |
| CN109145314A (en) | Use the natural language processing of the integrated circuit based on CNN | |
| CN106560891A (en) | Speech Recognition Apparatus And Method With Acoustic Modelling | |
| CN109271644A (en) | A kind of translation model training method and device | |
| CN109918501A (en) | Method, apparatus, equipment and the storage medium of news article classification | |
| CN112232165B (en) | Data processing method, device, computer and readable storage medium | |
| CN105989067A (en) | Method for generating text abstract from image, user equipment and training server | |
| CN107993651A (en) | A kind of audio recognition method, device, electronic equipment and storage medium | |
| CN110968692A (en) | Text classification method and system | |
| CN117710534B (en) | Animation collaborative making method based on improved teaching and learning optimization algorithm | |
| JP2023549048A (en) | Speech recognition method and apparatus, computer device and computer program |
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 | ||
| CB02 | Change of applicant information |
Address after: 100040 Shijingshan Road, Shijingshan District, Beijing, No. 20, 16 layer 1601 Applicant after: Chinese translation language through Polytron Technologies Inc Address before: 100040 Shijingshan District railway building, Beijing, the 16 floor Applicant before: Mandarin Technology (Beijing) Co., Ltd. |
|
| CB02 | Change of applicant information | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170905 |
|
| RJ01 | Rejection of invention patent application after publication |