CN109448701A - A kind of intelligent sound recognizes the result statistical system and method for semantic understanding - Google Patents
A kind of intelligent sound recognizes the result statistical system and method for semantic understanding Download PDFInfo
- Publication number
- CN109448701A CN109448701A CN201811090843.XA CN201811090843A CN109448701A CN 109448701 A CN109448701 A CN 109448701A CN 201811090843 A CN201811090843 A CN 201811090843A CN 109448701 A CN109448701 A CN 109448701A
- Authority
- CN
- China
- Prior art keywords
- result
- test
- text
- voice
- tested
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 15
- 238000012360 testing method Methods 0.000 claims abstract description 104
- 230000004044 response Effects 0.000 claims abstract description 17
- 238000012986 modification Methods 0.000 claims description 21
- 230000004048 modification Effects 0.000 claims description 21
- 238000011156 evaluation Methods 0.000 claims description 14
- 238000007619 statistical method Methods 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 5
- 230000000052 comparative effect Effects 0.000 claims description 4
- 238000013095 identification testing Methods 0.000 claims description 4
- SBNFWQZLDJGRLK-UHFFFAOYSA-N phenothrin Chemical group CC1(C)C(C=C(C)C)C1C(=O)OCC1=CC=CC(OC=2C=CC=CC=2)=C1 SBNFWQZLDJGRLK-UHFFFAOYSA-N 0.000 claims description 4
- 238000002360 preparation method Methods 0.000 claims description 3
- 235000013399 edible fruits Nutrition 0.000 claims 1
- 230000002618 waking effect Effects 0.000 claims 1
- 230000000694 effects Effects 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 3
- 238000004321 preservation Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000012805 post-processing Methods 0.000 description 1
- 238000010998 test method Methods 0.000 description 1
Classifications
-
- 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
- G10L15/01—Assessment or evaluation of speech recognition systems
-
- 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
- G10L15/26—Speech to text systems
Landscapes
- Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Machine Translation (AREA)
Abstract
The invention discloses result statistical system and method that a kind of intelligent sound recognizes semantic understanding, system includes operating status acquiring unit, tested speech selection unit, response results acquiring unit, tests semantic acquiring unit and test cell.Using scheme of the present invention can be derived that one closer to the identification text with voice content, the rate of precision of semantic service is calculated after comparing, it is convenient and efficient and can automatically see report, it greatly improves the efficiency, the voice semantic test activity string isolated together.
Description
Technical field
The present invention relates to Computer Applied Technologies, and in particular to a kind of intelligent sound recognizes the result statistics of semantic understanding
System and method.
Background technique
It is increasingly mature with speech recognition technology, just there are more and more intelligent appliances in market, smart home is set
Standby, these smart machines are based on speech recognition technology, at present to the test and comparison trouble of these speech-sound intelligent equipment, main problem
It is, the accuracy rate test and the aptitude tests of semantic understanding of speech recognition can not be separated clearly, and final result is asked
When topic, the problem of can not fast and accurately summing up in the point that speech recognition problem or semantic understanding.
In short say from user to being replied, if answered dissatisfied, it is not known that go out where centre is baffled
Problem is showed.For example ask frog and tadpole, answer has sung a first small tadpole and has looked for mother；Or ask what you cry, machine returns
It is most beautiful to answer little Bai.When such issues that underscore, actually it or is that can not know speech recognition semantic reason out of joint
It solves out of joint.
When once speech recognition appearance and received text or model answer cannot match the case where, test semantic understanding does not have
Meaning, such as above example.
None " can correct " process of speech recognition for existing speech recognition and post-processing technology, artificial judgement
Process isolate out.And checking and modifying in the testing tool direct visualizztion of having no idea, is not also modified
Self-teaching afterwards, next time can also identify mistake in face of same voice.
The test of traditional speech recognition to semantic understanding can not perceive intermediate result, wonder which word identification is wrong
It is accidentally or correct invisible, and semantic understanding this step wonders that the details of context is difficult.
Summary of the invention
The purpose of the present invention is to provide result statistical system and method that a kind of intelligent sound recognizes semantic understanding, solutions
In certainly existing intelligent sound identification test process, the accuracy rate test result and semantic understanding for opening speech recognition cannot be distinguished
The problem of aptitude tests result.
In order to solve the above technical problems, the invention adopts the following technical scheme:
A kind of intelligent sound recognizes the result statistical system of semantic understanding,
Including the operating status acquiring unit for obtaining the current operating conditions being devices under；
Including for choosing the tested speech to be played from sound bank according to the current operating conditions being devices under
Tested speech selection unit；
Including for being devices under to described be devices under described in the tested speech and acquisition that broadcasting is chosen to described
The response results acquiring unit of the response results of tested speech；
Including carrying out semantic test according to the text of identification, the test semanteme that selection has been identified rear text obtains single
Member；
Including for obtaining voice and semantic test according to the response results as a result, the response results are rung with expected
Should result compare, obtain the test cell that final identification test result and semantic test result summarize.
Further, the operating status acquiring unit is particular by the unique identification being devices under to testing service
Device sends inquiry request, and acquisition is devices under the current operating conditions for being reported to testing service device.
Further, the response results acquiring unit is when choosing the tested speech to be played, if be devices under
Current operating conditions are then to call from sound bank to wake-up states and wake up audio；If the current operating conditions being devices under
For state to be identified, then identification audio is called from sound bank.
A kind of intelligent sound recognizes the result statistical method of semantic understanding, including following methods:
Test cell chooses voice and speech-recognition services API to be tested, inputs language to be tested to speech-recognition services API
Sound,
Intelligent recognition voice to be tested, and writing text is compiled as recognition result, by recognition result and language to be tested
Mark text in sound carries out identification comparison；
If recognition result is consistent with test text after comparison, the voice to be tested is as semantic test text；
Semantic test text input to semantic understanding is serviced into API, carries out semantic understanding test, compares semantic test point,
Obtain semantic understanding result；
Calculate the result of simultaneously statistics of speech recognition rate and semantic rate of precision.
Further technical solution is, if the recognition result of speech recognition and test text are inconsistent, carries out following
Processing mode:
Manually listen to voice to be tested, if manually listen to after approve of intelligent recognition voice to be tested as a result, if choose
The result of intelligent recognition；
Do not approved of after manually listening to intelligent recognition voice to be tested as a result, if modify mark text, and to modification
Mark text afterwards is evaluated；
If modified mark text evaluation is higher than the mark text before modification, after voice to be tested is determined as modification
Mark text；
If modified mark text evaluation is lower than the mark text before modification, before voice to be tested is determined as modification
Mark text.
Using determining mark text as semantic test text.
Further technical solution is that the recognition result of voice to be tested sets 10 points of full marks Ei as full marks；
Approve of after manually listened to intelligent recognition voice to be tested as a result, if intelligence test voice recognition result note
It is 10 points；
Do not approved of after manually listening to intelligent recognition voice to be tested as a result, if intelligence test voice be denoted as 8 points, when
Mark text evaluation after manual amendment is higher than 9 points or more, then uses modified mark text, mark after manual amendment
Text evaluation is lower than 7 points hereinafter, then using the mark text before modification.
Further technical solution is that the semantic understanding test is specifically:
The input condition that text is designed according to integrated SDK services API by semantic understanding and is sent directly into text text
Part is obtained a result；
According to the ways of writing of the semantic test point of Comparative result preparation, field, the corresponding return value of field；
According to result judgement of the field name after identical, the rate of precision of semantic understanding test is obtained.
Compared with prior art, the beneficial effects of the present invention are: can help to analyze " no using scheme of the present invention
What content voice clearly " has been identified as on earth, and artificial marking and machine marking are weighted, obtain one closer to
With the identification text of voice content, and machine is helped to do the training of speech recognition.Solve the problems, such as that machine talk identifies fallibility,
Test after carrying out re -training.And directly recognition result can be exported according to voice to be tested and semanteme returns the result, obtained
The checkpoint of semantic understanding out, calculates the rate of precision of semantic service after comparing, convenient and efficient and can automatically see report
It accuses, greatly improves the efficiency, the voice semantic test activity string isolated together.
Detailed description of the invention
It is system block diagram that Fig. 1, which is of the invention,.
Fig. 2 is flow diagram of the invention.
Fig. 3 is a kind of way of contrast schematic diagram of speech recognition result and mark text in the present invention.
Fig. 4 is speech recognition result and processing flow schematic diagram when different voice annotation text in the present invention.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
Fig. 1 shows following embodiment of the invention:
A kind of intelligent sound recognizes the result statistic device of semantic understanding,
Including the operating status acquiring unit for obtaining the current operating conditions being devices under；
Including for choosing the tested speech to be played from sound bank according to the current operating conditions being devices under
Tested speech selection unit；
Including for being devices under to described be devices under described in the tested speech and acquisition that broadcasting is chosen to described
The response results acquiring unit of the response results of tested speech；
Including carrying out semantic test according to the text of identification, the test semanteme that selection has been identified rear text obtains single
Member；
Including for obtaining voice and semantic test according to the response results as a result, the response results are rung with expected
Should result compare, obtain the test cell that final identification test result and semantic test result summarize.
According to above-described embodiment 1, preferably, the operating status acquiring unit is particular by being devices under
Unique identification to testing service device send inquiry request, acquisition be devices under the current operation shape for being reported to testing service device
State.
According to above-described embodiment 1, preferably, the response results acquiring unit is choosing the test language to be played
When sound, if the current operating conditions being devices under are to call from sound bank to wake-up states and wake up audio；If tested
The current operating conditions of equipment are state to be identified, then identification audio is called from sound bank.
Embodiment 2:
As shown in Fig. 2, a kind of intelligent sound recognizes the result statistical method of semantic understanding, including following methods:
Test cell chooses voice and speech-recognition services API to be tested, inputs language to be tested to speech-recognition services API
Sound,
Intelligent recognition voice to be tested, and writing text is compiled as recognition result, by recognition result and language to be tested
Mark text in sound carries out identification comparison, way of contrast as shown in Figure 3；
If recognition result is consistent with test text after comparison, the voice to be tested is as semantic test text；
Semantic test text input to semantic understanding is serviced into API, carries out semantic understanding test, compares semantic test point,
Obtain semantic understanding result；
Calculate the result of simultaneously statistics of speech recognition rate and semantic rate of precision.
Embodiment 3:
A kind of intelligent sound recognizes the result statistical method of semantic understanding, including following methods:
Voice to be tested and speech-recognition services API are selected, voice to be tested is inputted to speech-recognition services API,
Intelligent recognition voice to be tested, and writing text is compiled as recognition result, by recognition result and language to be tested
Mark text in sound carries out identification comparison；
If recognition result is consistent with test text after comparison, the voice to be tested is as semantic test text
(being semantic test text 1 in Fig. 2)；
If the recognition result of speech recognition is inconsistent with test text, carry out the following processing mode: manually listen to
Tested speech, approved of after manually listening to intelligent recognition voice to be tested as a result, if choose the result of intelligent recognition；If
It is not approving of intelligent recognition voice to be tested after manually listening to as a result, then modify mark text, and to modified mark text
It is evaluated；If modified mark text evaluation is higher than the mark text before modification, voice to be tested is determined as modifying
Mark text afterwards；If modified mark text evaluation is lower than the mark text before modification, voice to be tested is determined as
Mark text before modification；Using determining mark text as semantic test text (being semantic test text 2 in Fig. 2)；
Semantic test text input to semantic understanding is serviced into API, carries out semantic understanding test, compares semantic test point,
Obtain semantic understanding result；
Calculate the result of simultaneously statistics of speech recognition rate and semantic rate of precision.
What as shown in figure 4, manually hear 4.wav first, under artificial judgment, this voice document said on earth is what (if
What is heard is that Zhang San says, then saves according to the result heard；Machine answer is then modified if it is other results, this button is only
It is to listen to correct judgment to answer)；The result (machine marking is automatically set to 10 points) that " thumbing up " then approves of machine recognition is clicked, "
Three say " as this result preservation；" preservations " (machine marking be automatically set to 8 points) is clicked then into next interface: can be with
New result (wrongly write as marked text, or manually help the wrong voice of identification that can continue to test semanteme) is edited,
Text is inputted after reporting an error to save, and is clicked and is given a mark after saving, pops up two options (9 points or more, 6 points or less), select any one；
8 points of calculating average marks of comprehensive machine, using artificial as a result, using machine if if it is less than 7 points if average mark is greater than 8 points
As a result；It is thus saved according to the text for being most like the voice document that we hear, as semantic test text.
Embodiment 4:
A kind of intelligent sound recognizes the result statistical method of semantic understanding, including following methods:
Voice to be tested and speech-recognition services API are selected, voice to be tested is inputted to speech-recognition services API,
Intelligent recognition voice to be tested, and writing text is compiled as recognition result, by recognition result and language to be tested
Mark text in sound carries out identification comparison；
If recognition result is consistent with test text after comparison, the voice to be tested is as semantic test text；
If the recognition result of speech recognition is inconsistent with test text, carry out the following processing mode: manually listen to
Tested speech, approved of after manually listening to intelligent recognition voice to be tested as a result, if choose the result of intelligent recognition；If
It is not approving of intelligent recognition voice to be tested after manually listening to as a result, then modify mark text, and to modified mark text
It is evaluated；If modified mark text evaluation is higher than the mark text before modification, voice to be tested is determined as modifying
Mark text afterwards；If modified mark text evaluation is lower than the mark text before modification, voice to be tested is determined as
Mark text before modification；Using determining mark text as semantic test text.Such as: the recognition result of voice to be tested is set
Fixed 10 points are full marks；Approve of after manually listened to intelligent recognition voice to be tested as a result, if intelligence test voice identification
As a result 10 points are denoted as；Do not approved of after manually listening to intelligent recognition voice to be tested as a result, if intelligence test voice be denoted as 8
Point, mark text evaluation after manual amendment is higher than 9 points or more, then modified mark text is used, after manual amendment
Mark text evaluation lower than 7 points hereinafter, then using modification before mark text；
Semantic test text input to semantic understanding is serviced into API, carries out semantic understanding test, compares semantic test point,
Obtain semantic understanding result；
Calculate the result of simultaneously statistics of speech recognition rate and semantic rate of precision.
Embodiment 4:
A kind of intelligent sound recognizes the result statistical method of semantic understanding, including following methods:
Voice to be tested and speech-recognition services API are selected, voice to be tested is inputted to speech-recognition services API,
Intelligent recognition voice to be tested, and writing text is compiled as recognition result, by recognition result and language to be tested
Mark text in sound carries out identification comparison；
If recognition result is consistent with test text after comparison, the voice to be tested is as semantic test text；
Semantic test text input to semantic understanding is serviced into API, carries out semantic understanding test, compares semantic test point,
Obtain semantic understanding result；The specific test method is as follows: designing the input condition of text according to integrated SDK, passes through language
Reason and good sense solution service API is sent directly into text file, obtains a result；According to the writing side for the semantic test point that Comparative result prepares
The corresponding return value of formula, field, field；According to result judgement of the field name after identical, the rate of precision of semantic understanding test is obtained；
Calculate the result of simultaneously statistics of speech recognition rate and semantic rate of precision.
As the preferred of above-described embodiment:
The online disclosed semantic understanding service API (such as Baidu interrogates and flies equal SDK) of local Tool integration, according to integrated
SDK carrys out the input condition of the good text of decision design, can be sent directly into text file by interface API, obtain a result, and such as one
The parsing format of words.The text sent such as " Beijing weather today "；It returns the result as follows
How to be write according to the semantic test point that Comparative result prepares, if field is identical, if field is corresponding
Return value is identical
Text results expection such as preparation is as follows
According to result judgement of the field name after identical, the rate of precision of semantic understanding test is obtained；
Export test report.
Feature name in the application is explained:
Voice to be identified: the voice document of tested speech discrimination.
Voice annotation text: the text file marked correspondingly with voice to be identified is labelled with to be identified
Voice has said anything on earth.
Semantic text to be tested: will test the object to be tested of semantic understanding, and text-only file can be voice annotation text
Directly as semantic test text.
Semantic test point:, the text file that has marked one-to-one with semantic text to be tested.One text has multiple
Checkpoint will detect after testing whether how much whether checkpoint have covering/coverage rate.
Speech-recognition services API: the speech-recognition services interface of access inputs voice document, obtains recognition result, identifies
The result is that text.
Calculate phonetic recognization rate: being exported with speech recognition API, that is, the text results that identify and voice annotation text into
Discrimination is calculated after row comparison.
Semantic understanding services API: the semantic understanding service interface of access, inputs text file, obtains the multiple inspections of text
Point result.
It calculates the rate of precision of semantic understanding: being exported with semantic service API as a result, comparison semantic test point, calculates precisely
Rate.
Although reference be made herein to invention has been described for multiple explanatory embodiments of the invention, however, it is to be understood that
Those skilled in the art can be designed that a lot of other modification and implementations, these modifications and implementations will fall in this Shen
It please be within disclosed scope and spirit.More specifically, disclose in the application, drawings and claims in the range of, can
With the building block and/or a variety of variations and modifications of layout progress to theme combination layout.In addition to building block and/or layout
Outside the modification and improvement of progress, to those skilled in the art, other purposes also be will be apparent.
Claims (7)
1. the result statistical system that a kind of intelligent sound recognizes semantic understanding, it is characterised in that:
Including the operating status acquiring unit for obtaining the current operating conditions being devices under；
Including the test for choosing the tested speech to be played from sound bank according to the current operating conditions being devices under
Voice selection unit；
Including for being devices under to described be devices under described in the tested speech and acquisition that broadcasting is chosen to the test
The response results acquiring unit of the response results of voice；
Including carrying out semantic test according to the text of identification, selection has been identified the test semanteme acquiring unit of rear text；
Including for obtaining voice and semantic test as a result, by the response results and intended response knot according to the response results
Fruit compares, and obtains the test cell that final identification test result and semantic test result summarize.
2. the result statistical system that a kind of intelligent sound according to claim 1 recognizes semantic understanding, it is characterised in that:
The operating status acquiring unit sends inquiry request to testing service device particular by the unique identification being devices under, and obtains
Take the current operating conditions for being devices under and being reported to testing service device.
3. the result statistical system that a kind of intelligent sound according to claim 1 recognizes semantic understanding, it is characterised in that:
The response results acquiring unit is when choosing the tested speech to be played, if the current operating conditions being devices under are wait call out
The state of waking up then is called from sound bank and wakes up audio；If the current operating conditions being devices under are state to be identified, from language
Identification audio is called in sound library.
4. the result statistical method that a kind of intelligent sound recognizes semantic understanding, it is characterised in that including following methods:
Test cell chooses voice and speech-recognition services API to be tested, inputs voice to be tested to speech-recognition services API,
Intelligent recognition voice to be tested, and compile writing text as recognition result, will be in recognition result and voice to be tested
Mark text carry out identification comparison；
If recognition result is consistent with test text after comparison, the voice to be tested is as semantic test text；
Semantic test text input to semantic understanding is serviced into API, carries out semantic understanding test, semantic test point is compared, obtains
Semantic understanding result；
Calculate the result of simultaneously statistics of speech recognition rate and semantic rate of precision.
5. the result statistical method that a kind of intelligent sound according to claim 4 recognizes semantic understanding, it is characterised in that:
If the recognition result of speech recognition is inconsistent with test text, mode is carried out the following processing:
Manually listen to voice to be tested, if manually listen to after approve of intelligent recognition voice to be tested as a result, if choose intelligence
The result of identification；
Do not approved of after manually listening to intelligent recognition voice to be tested as a result, if modify mark text, and to modified
Mark text is evaluated；
If modified mark text evaluation is higher than the mark text before modification, voice to be tested is determined as modified mark
Explanatory notes sheet；
If modified mark text evaluation is lower than the mark text before modification, voice to be tested is determined as the mark before modification
Explanatory notes sheet；
Using determining mark text as semantic test text.
6. the result statistical method that a kind of intelligent sound according to claim 5 recognizes semantic understanding, it is characterised in that:
It is full marks that the recognition result of voice to be tested, which sets 10 points,；
Approve of after manually listened to intelligent recognition voice to be tested as a result, if the recognition result of intelligence test voice be denoted as 10
Point；
Do not approved of after manually listening to intelligent recognition voice to be tested as a result, if intelligence test voice be denoted as 8 points, when artificial
Modified mark text evaluation is higher than 9 points or more, then uses modified mark text, mark text after manual amendment
Evaluation is lower than 7 points hereinafter, then using the mark text before modification.
7. the result statistical method that a kind of intelligent sound according to claim 4 recognizes semantic understanding, it is characterised in that:
The semantic understanding test is specifically:
The input condition that text is designed according to integrated SDK services API by semantic understanding and is sent directly into text file,
It obtains a result；
According to the ways of writing of the semantic test point of Comparative result preparation, field, the corresponding return value of field；
According to result judgement of the field name after identical, the rate of precision of semantic understanding test is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811090843.XA CN109448701A (en) | 2018-09-19 | 2018-09-19 | A kind of intelligent sound recognizes the result statistical system and method for semantic understanding |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811090843.XA CN109448701A (en) | 2018-09-19 | 2018-09-19 | A kind of intelligent sound recognizes the result statistical system and method for semantic understanding |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109448701A true CN109448701A (en) | 2019-03-08 |
Family
ID=65530533
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811090843.XA Pending CN109448701A (en) | 2018-09-19 | 2018-09-19 | A kind of intelligent sound recognizes the result statistical system and method for semantic understanding |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109448701A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110417589A (en) * | 2019-07-23 | 2019-11-05 | 徐州工程学院 | A kind of vehicle-mounted voice cloud user experience quality road measuring method |
| CN111681642A (en) * | 2020-06-03 | 2020-09-18 | 北京字节跳动网络技术有限公司 | Speech recognition evaluation method, device, storage medium and equipment |
| CN111696523A (en) * | 2019-03-12 | 2020-09-22 | 大众问问(北京)信息科技有限公司 | Accuracy testing method and device of voice recognition engine and electronic equipment |
| CN111797368A (en) * | 2020-07-03 | 2020-10-20 | 全球能源互联网研究院有限公司 | Data watermark identification and analysis method and system |
| CN112417109A (en) * | 2020-10-26 | 2021-02-26 | 出门问问(苏州)信息科技有限公司 | Method and device for testing man-machine conversation system |
| CN110164474B (en) * | 2019-05-08 | 2021-09-14 | 北京百度网讯科技有限公司 | Voice wake-up automatic test method and system |
| CN113836010A (en) * | 2021-09-14 | 2021-12-24 | 招商银行股份有限公司 | Voice intelligent customer service automatic testing method, system and storage medium |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005227510A (en) * | 2004-02-12 | 2005-08-25 | Ntt Docomo Inc | Voice recognition device and voice recognition method |
| US7440895B1 (en) * | 2003-12-01 | 2008-10-21 | Lumenvox, Llc. | System and method for tuning and testing in a speech recognition system |
| CN104290097A (en) * | 2014-08-19 | 2015-01-21 | 白劲实 | Learning type intelligent home social contact robot system and method |
| CN104347081A (en) * | 2013-08-07 | 2015-02-11 | 腾讯科技（深圳）有限公司 | Method and device for testing scene statement coverage |
| CN104538031A (en) * | 2014-12-15 | 2015-04-22 | 北京云知声信息技术有限公司 | Intelligent voice service development cloud platform and method |
| CN104750779A (en) * | 2015-03-04 | 2015-07-01 | 华东师范大学 | Chinese multi-class word identification method based on conditional random field |
| CN106228986A (en) * | 2016-07-26 | 2016-12-14 | 北京奇虎科技有限公司 | The automated testing method of a kind of speech recognition engine, device and system |
| CN107516510A (en) * | 2017-07-05 | 2017-12-26 | 百度在线网络技术（北京）有限公司 | A kind of smart machine automated voice method of testing and device |
| CN107665708A (en) * | 2016-07-29 | 2018-02-06 | 科大讯飞股份有限公司 | Intelligent sound exchange method and system |
-
2018
- 2018-09-19 CN CN201811090843.XA patent/CN109448701A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7440895B1 (en) * | 2003-12-01 | 2008-10-21 | Lumenvox, Llc. | System and method for tuning and testing in a speech recognition system |
| JP2005227510A (en) * | 2004-02-12 | 2005-08-25 | Ntt Docomo Inc | Voice recognition device and voice recognition method |
| CN104347081A (en) * | 2013-08-07 | 2015-02-11 | 腾讯科技（深圳）有限公司 | Method and device for testing scene statement coverage |
| CN104290097A (en) * | 2014-08-19 | 2015-01-21 | 白劲实 | Learning type intelligent home social contact robot system and method |
| CN104538031A (en) * | 2014-12-15 | 2015-04-22 | 北京云知声信息技术有限公司 | Intelligent voice service development cloud platform and method |
| CN104750779A (en) * | 2015-03-04 | 2015-07-01 | 华东师范大学 | Chinese multi-class word identification method based on conditional random field |
| CN106228986A (en) * | 2016-07-26 | 2016-12-14 | 北京奇虎科技有限公司 | The automated testing method of a kind of speech recognition engine, device and system |
| CN107665708A (en) * | 2016-07-29 | 2018-02-06 | 科大讯飞股份有限公司 | Intelligent sound exchange method and system |
| CN107516510A (en) * | 2017-07-05 | 2017-12-26 | 百度在线网络技术（北京）有限公司 | A kind of smart machine automated voice method of testing and device |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111696523A (en) * | 2019-03-12 | 2020-09-22 | 大众问问(北京)信息科技有限公司 | Accuracy testing method and device of voice recognition engine and electronic equipment |
| CN111696523B (en) * | 2019-03-12 | 2024-03-01 | 大众问问(北京)信息科技有限公司 | Accuracy testing method and device of voice recognition engine and electronic equipment |
| CN110164474B (en) * | 2019-05-08 | 2021-09-14 | 北京百度网讯科技有限公司 | Voice wake-up automatic test method and system |
| CN110417589A (en) * | 2019-07-23 | 2019-11-05 | 徐州工程学院 | A kind of vehicle-mounted voice cloud user experience quality road measuring method |
| CN110417589B (en) * | 2019-07-23 | 2022-11-29 | 徐州工程学院 | Vehicle-mounted voice cloud user experience quality drive test method |
| CN111681642A (en) * | 2020-06-03 | 2020-09-18 | 北京字节跳动网络技术有限公司 | Speech recognition evaluation method, device, storage medium and equipment |
| CN111797368A (en) * | 2020-07-03 | 2020-10-20 | 全球能源互联网研究院有限公司 | Data watermark identification and analysis method and system |
| CN111797368B (en) * | 2020-07-03 | 2024-04-09 | 国网智能电网研究院有限公司 | Data watermark recognition analysis method and system |
| CN112417109A (en) * | 2020-10-26 | 2021-02-26 | 出门问问(苏州)信息科技有限公司 | Method and device for testing man-machine conversation system |
| CN112417109B (en) * | 2020-10-26 | 2023-08-01 | 问问智能信息科技有限公司 | Method and device for testing man-machine dialogue system |
| CN113836010A (en) * | 2021-09-14 | 2021-12-24 | 招商银行股份有限公司 | Voice intelligent customer service automatic testing method, system and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109448701A (en) | A kind of intelligent sound recognizes the result statistical system and method for semantic understanding | |
| CN107818798B (en) | Customer service quality evaluation method, device, equipment and storage medium | |
| CN108962282B (en) | Voice detection analysis method and device, computer equipment and storage medium | |
| US10699702B2 (en) | System and method for personalization of acoustic models for automatic speech recognition | |
| CN106328166B (en) | Human-computer dialogue abnormality detection system and method | |
| US20190318743A1 (en) | Metadata-based diarization of teleconferences | |
| CN106294774A (en) | User individual data processing method based on dialogue service and device | |
| CN112966082B (en) | Audio quality inspection method, device, equipment and storage medium | |
| US10217457B2 (en) | Learning from interactions for a spoken dialog system | |
| CN108428448A (en) | A kind of sound end detecting method and audio recognition method | |
| CN110232916A (en) | Method of speech processing, device, computer equipment and storage medium | |
| CN107767861A (en) | voice awakening method, system and intelligent terminal | |
| CN111833902B (en) | Awakening model training method, awakening word recognition device and electronic equipment | |
| CN105654949A (en) | Voice wake-up method and device | |
| CN108766441A (en) | A kind of sound control method and device based on offline Application on Voiceprint Recognition and speech recognition | |
| CN108694940A (en) | A kind of audio recognition method, device and electronic equipment | |
| CN108664269A (en) | A kind of feature attachment code peculiar smell detection method based on deep learning | |
| CN103077720A (en) | Speaker identification method and system | |
| CN109708256B (en) | Voice determination method and device, storage medium and air conditioner | |
| CN105931639B (en) | A kind of voice interactive method for supporting multistage order word | |
| Moritz et al. | All-in-one transformer: Unifying speech recognition, audio tagging, and event detection. | |
| CN109460558B (en) | Effect judging method of voice translation system | |
| CN107134277A (en) | A kind of voice-activation detecting method based on GMM model | |
| CN110176228A (en) | A kind of small corpus audio recognition method and system | |
| WO2024188277A1 (en) | Text semantic matching method and refrigeration device system |
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 | ||
| CB03 | Change of inventor or designer information | ||
| CB03 | Change of inventor or designer information |
Inventor after: Liao Zhiliang Inventor after: Tao Liang Inventor after: Wang Daoning Inventor after: Zhang Yadong Inventor after: Ma Limin Inventor before: Ma Yongfei Inventor before: Wang Daoning Inventor before: Ma Limin |
|
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190308 |