CN113488061B - Distribution network dispatcher identity verification method and system based on improved Synth2Aug - Google Patents
Distribution network dispatcher identity verification method and system based on improved Synth2Aug Download PDFInfo
- Publication number
- CN113488061B CN113488061B CN202110896813.3A CN202110896813A CN113488061B CN 113488061 B CN113488061 B CN 113488061B CN 202110896813 A CN202110896813 A CN 202110896813A CN 113488061 B CN113488061 B CN 113488061B
- Authority
- CN
- China
- Prior art keywords
- dispatcher
- voice
- instruction
- model
- training
- 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.)
- Active
Links
- 238000012795 verification Methods 0.000 title claims abstract description 28
- 238000000034 method Methods 0.000 title claims abstract description 20
- 238000012549 training Methods 0.000 claims abstract description 58
- 238000012360 testing method Methods 0.000 claims abstract description 30
- 238000005070 sampling Methods 0.000 claims abstract description 17
- 238000006243 chemical reaction Methods 0.000 claims abstract description 14
- 238000012986 modification Methods 0.000 claims description 9
- 230000004048 modification Effects 0.000 claims description 9
- 230000002457 bidirectional effect Effects 0.000 claims description 5
- 230000000295 complement effect Effects 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000003607 modifier Substances 0.000 description 2
- 240000005369 Alstonia scholaris Species 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L17/00—Speaker identification or verification
- G10L17/06—Decision making techniques; Pattern matching strategies
- G10L17/08—Use of distortion metrics or a particular distance between probe pattern and reference templates
-
- G—PHYSICS
- G10—MUSICAL INSTRUMENTS; ACOUSTICS
- G10L—SPEECH ANALYSIS OR SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING; SPEECH OR AUDIO CODING OR DECODING
- G10L17/00—Speaker identification or verification
- G10L17/04—Training, enrolment or model building
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Abstract
The invention discloses a distribution network dispatcher identity verification method and system based on improved Synth2Aug, which form different dispatching text instructions by utilizing different MTTG models according to given dispatching instruction keywords; converting the dispatching text instruction into virtual dispatching instruction voice by using a text voice conversion module; dividing an original recording sampling sample of a dispatcher into dispatcher training voice and dispatcher testing voice according to a certain proportion; mixing the virtual dispatching instruction voice and the dispatcher training voice to form a voiceprint recognition model training sample, and training the dispatcher voiceprint recognition model; and testing the model by using a dispatcher test voice sample, and outputting a dispatcher identity authentication result. According to the invention, a plurality of dispatcher virtual instruction voices are generated under the condition that the dispatcher voice samples are fewer, the voice print model training is realized together with the real original instruction voices, the principle that the main meanings are consistent and the expression modes are different is met among the plurality of virtual instruction voice samples, and the voice print model training method is the same as the scene of the on-site working condition.
Description
Technical Field
The invention relates to the field of distribution network dispatching, in particular to a distribution network dispatcher identity verification method and system based on improved Synth2 Aug.
Background
With the continuous expansion of the power grid scale, more and more power equipment are managed by a dispatching mechanism, and dispatching personnel change and handover frequently, so that the identity of the dispatching personnel needs to be verified in order to ensure the power grid dispatching safety. At present, the personnel of both sides of the power grid dispatching realize identity authentication in a mode of mutually notifying names, so that potential safety hazards of impersonation and replacement of external personnel are easily caused. In order to ensure the safety of a dispatching system, the prior research literature proposes a strategy for identifying voiceprints of two dispatching parties by using a deep learning model. However, in the power grid dispatching work, the dispatching personnel of the provincial dispatching department or the urban dispatching department are fewer, and each person can provide limited sampling samples, so that the requirement of the deep learning model can not be met.
Aiming at the problems, the invention provides a distribution network dispatcher identity verification method based on improved Synth2Aug based on the existing Synth2Aug technology, and the voice print recognition of the distribution network dispatcher is carried out, so that the accuracy of the dispatcher identity verification is improved.
Disclosure of Invention
In order to solve the defects in the prior art, the invention aims to provide a distribution network dispatcher identity verification method and system based on improved Synth2Aug, which automatically generate dispatcher virtual instruction voice according to dispatching instruction keywords, realize training of a voiceprint model together with real original voice, and carry out voiceprint recognition of distribution network dispatcher, so that the accuracy of dispatcher identity authentication is improved.
In order to achieve the purpose of the invention, the technical scheme adopted by the invention is as follows:
a distribution network dispatcher identity verification method based on an improved Synth2Aug, the method comprising the steps of:
(1) Giving a group of scheduling instruction keywords;
(2) According to a given scheduling instruction keyword, different MTTG models are utilized to form different scheduling text instructions;
(3) Converting the dispatching text instruction obtained in the step (2) into virtual dispatching instruction voice by utilizing a text voice conversion module;
(4) Dividing an original recording sampling sample of a dispatcher into dispatcher training voice and dispatcher testing voice according to a certain proportion;
(5) Mixing the virtual dispatching instruction voice and the dispatcher training voice to form an integral training sample of the voiceprint recognition model, and training the dispatcher voiceprint recognition model;
(6) After the model is successfully trained, testing the model by using a dispatcher to test a voice sample; and outputting a dispatcher identity authentication result, and determining whether the dispatcher identity authentication is successful.
Further, in the step (2), the mttg_1 model uses a bidirectional LSTM as a core module, generates a scheduling text instruction, and uses fewer modification components; the MTTG_2 model uses GRU as a core module, generates a dispatch text instruction, and uses corresponding modifier components.
Further, in the step (3), the text-to-speech conversion module is implemented by adopting a Tacotron2 model, takes the dispatching text instruction as the input of the model, and outputs the dispatching personnel virtual dispatching instruction speech.
Further, in the step (4), the original recording sample of the dispatcher is divided into dispatcher training voice and dispatcher testing voice according to the ratio of 7:3.
Further, in the step (5), the virtual scheduling instruction speech and the scheduler training speech are respectively divided into N subsets with different sizes, namely a virtual scheduling instruction speech subset (vaudio_k, k=1, 2,..n) and an original scheduler training speech subset (audio_train_k, k=1, 2,..n), and subsets with equal k are respectively taken out from the virtual scheduling instruction speech subset and the original scheduler training speech subset to be mixed, so as to form N mixed subsets, and finally the N mixed subsets are input into a voiceprint recognition training model to perform model training.
Further, N may be 10.
The system comprises a dispatching text instruction generation module, a text voice conversion module, a dispatcher original record sampling module, a dispatcher voiceprint recognition model training module and a dispatching network dispatcher identity verification module;
the dispatch text instruction generation module forms different dispatch text instructions by using different MTTG models according to given dispatch instruction keywords;
the text-to-speech conversion module is used for converting the obtained dispatching text instruction into virtual dispatching instruction speech;
the dispatcher original recording sampling module is used for carrying out dispatcher original recording sampling and dividing the sampling sample into dispatcher training voice and dispatcher testing voice according to a certain proportion;
the dispatcher voiceprint recognition model training module mixes the virtual dispatching instruction voice with dispatcher training voice to form an integral training sample of the voiceprint recognition model, and trains the dispatcher voiceprint recognition model;
and the distribution network dispatcher identity verification module is used for testing the model by using a dispatcher test voice sample, outputting a dispatcher identity verification result and determining whether the dispatcher identity verification is successful.
Compared with the prior art, the voice recognition method has the advantages that virtual instruction voice of the dispatcher can be generated under the condition that the voice samples of the dispatcher are fewer, and further training of the voiceprint model is achieved together with the real original instruction voice. The virtual instruction voice is automatically generated according to the dispatching instruction keywords, and the principle that the main meanings are consistent and the expression modes are different is met among a plurality of virtual instruction voice samples, and the virtual instruction voice is identical to the scene of the dispatching field working condition.
Drawings
Fig. 1 is a flow chart of a method for authentication of a distribution network dispatcher based on an improved Synth2 Aug;
fig. 2 is a view showing the construction of Tacotron2 model.
Detailed Description
The technical scheme of the invention is further described below with reference to the accompanying drawings and examples. The following examples are only for more clearly illustrating the technical solutions of the present invention and are not intended to limit the scope of protection of the present application.
The existing Synth2Aug uses TTS (Text-To-Speech) for model training, and Text content Text and Speech content Speech are completely identical. In actual dispatching work, specific contents of dispatching languages are greatly different due to the fact that different dispatching personnel understand and express habits of the same dispatching operation.
Therefore, the invention firstly gives out the key words of the dispatching instructions as the main intention of dispatching personnel, then generates dispatching texts by using MTTG (fusing-to-Text generation), respectively generates two different dispatching Text contents (Text 1 and Text 2) by adopting LSTM and GRU models, and then generates corresponding virtual dispatching voice samples (VAudio 1 and VAudio 2) by using the TTS technology. Further, the virtual dispatch language samples are mixed with the original dispatch voices from different dispatcher to form training samples of the dispatcher voiceprint recognition model. And finally, performing effect test on the model according to the test sample of the dispatcher.
As shown in fig. 1, the distribution network dispatcher identity verification method based on the improved Synth2Aug is as follows:
(1) Giving a group of scheduling instruction keywords (Key words);
(2) Forming different dispatching Text instructions (Text 1 and Text 2) by using different MTTG models (MTTG_1 and MTTG_2) according to given dispatching instruction keywords;
(3) Converting the dispatch Text instruction obtained in the step (2) into virtual dispatch instruction voices (VAudio 1 and VAudio 2) by using a Text-To-Speech conversion module (Text-To-Speech);
(4) Dividing an original recording sampling sample of a dispatcher into dispatcher training voice (Audio Train) and dispatcher testing voice (Audio Test) according to a certain proportion;
(5) Mixing the virtual dispatching instruction voice and the dispatcher training voice to form an integral training sample of the voiceprint recognition model, and training the dispatcher voiceprint recognition model;
(6) After the model is successfully trained, testing the model by using an original test voice sample of a dispatcher; and outputting a dispatcher identity authentication result, and determining whether the dispatcher identity authentication is successful.
The following illustrates a distribution network dispatcher identity verification method based on an improved Synth2Aug, which comprises the following specific steps:
(1) Giving a group of scheduling instruction keywords (Key words);
and according to the running condition of the power distribution network, a keyword of a scheduling instruction is given. For example, the sum, the line, the switch.
(2) Forming different dispatching Text instructions (Text 1 and Text 2) by using different MTTG models (MTTG_1 and MTTG_2) according to given dispatching instruction keywords;
in order to generate different texts by using a group of keywords, different MTTG models (MTTG_1 and MTTG_2) are used in the MTTG_1 model, a bidirectional long and short time memory network (bidirectional LSTM) is used as a core module, and modification components such as 'fixed language, scholartree language, complement language' and the like used in text generation are fewer. In the mttg_2 model, a gate control loop unit (GRU) is used as a core module, and the generated scheduling instruction text uses corresponding modifier components such as 'fixed language, scholarly language, and complement'.
According to the given scheduling command keywords "on", "off", "on" and "off", the mttg_1 model is used to generate the scheduling Text command Text1 as "on", "on" and "off", and the mttg_2 model is used to generate the Text command Text2 as "on" and "on" the line where the request must be immediately on ".
(3) Converting the dispatch Text instruction obtained in the step (2) into virtual dispatch instruction voices (VAudio 1 and VAudio 2) by using a Text-To-Speech conversion module (Text-To-Speech);
the Text-to-speech conversion module is implemented by adopting the existing Tacotron2 model, as shown in fig. 2, and uses the text_1 and the text_2 generated in the step (2) as the input of the model respectively, and outputs the dispatcher virtual dispatching instruction speech VAudio1 and VAudio2.
(4) Dividing an original recording sample of a dispatcher into dispatcher training voice (Audio Train) and dispatcher testing voice (Audio Test) according to a ratio of 7:3;
(5) Mixing the virtual dispatching instruction voice and the dispatcher training voice to form an integral training sample of the voiceprint recognition model, and training the dispatcher voiceprint recognition model;
the virtual scheduling instruction voice and the scheduling personnel training voice are respectively divided into N subsets with different sizes (N=10 is preferable according to experience values), namely a virtual scheduling instruction voice subset (VAudio_k, k=1, 2,..10) and an original scheduling personnel training voice subset (Audio_train_k, k=1, 2,..10), and subsets with equal k are respectively taken out from the virtual scheduling instruction voice subset and the original scheduling personnel training voice subset to be mixed, so that 10 mixed subsets are formed. Finally, 10 mixed subsets are input into a voiceprint recognition training model to train the model.
(6) After the model is successfully trained, testing the model by using an original test voice sample of a dispatcher; and outputting a dispatcher identity authentication result, and determining whether the dispatcher identity authentication is successful.
The invention also provides a distribution network dispatcher identity verification system based on the improved Synth2Aug, which comprises: the system comprises a dispatching text instruction generation module, a text voice conversion module, a dispatcher original recording sampling module, a dispatcher voiceprint recognition model training module and a distribution network dispatcher identity verification module.
And the dispatch text instruction generation module forms different dispatch text instructions by using different MTTG models according to given dispatch instruction keywords.
And the text-to-speech conversion module is used for converting the obtained dispatching text instruction into virtual dispatching instruction speech.
And the dispatcher original recording sampling module is used for carrying out dispatcher original recording sampling and dividing the sampling sample into dispatcher training voice and dispatcher testing voice according to a certain proportion.
And the dispatcher voiceprint recognition model training module mixes the virtual dispatching instruction voice with dispatcher training voice to form an integral training sample of the voiceprint recognition model, and trains the dispatcher voiceprint recognition model.
And the distribution network dispatcher identity verification module is used for testing the model by using a dispatcher test voice sample, outputting a dispatcher identity verification result and determining whether the dispatcher identity verification is successful.
Compared with the prior art, the voice recognition method has the advantages that virtual instruction voice of the dispatcher can be generated under the condition that the voice samples of the dispatcher are fewer, and further training of the voiceprint model is achieved together with the real original instruction voice. The virtual instruction voice is automatically generated according to the dispatching instruction keywords, and the principle that the main meanings are consistent and the expression modes are different is met among a plurality of virtual instruction voice samples, and the virtual instruction voice is identical to the scene of the dispatching field working condition.
While the applicant has described and illustrated the embodiments of the present invention in detail with reference to the drawings, it should be understood by those skilled in the art that the above embodiments are only preferred embodiments of the present invention, and the detailed description is only for the purpose of helping the reader to better understand the spirit of the present invention, and not to limit the scope of the present invention, but any improvements or modifications based on the spirit of the present invention should fall within the scope of the present invention.
Claims (5)
1. An improved Synth2 Aug-based distribution network dispatcher identity verification method, comprising the steps of:
(1) Giving a group of scheduling instruction keywords;
(2) According to a given scheduling instruction keyword, different MTTG models are utilized to form different scheduling text instructions; the method comprises an MTTG_1 model and an MTTG_2 model, wherein the MTTG_1 model uses a bidirectional LSTM as a core module to generate a dispatching text instruction, and the use of modification components is less; the MTTG_2 model uses GRU as a core module, a scheduling text instruction is generated, and corresponding modification components are used, wherein the modification components of the MTTG_2 model are more than those of the MTTG_1 model;
(3) Converting the dispatching text instruction obtained in the step (2) into virtual dispatching instruction voice by utilizing a text voice conversion module;
(4) Dividing an original recording sampling sample of a dispatcher into dispatcher training voice and dispatcher testing voice according to a certain proportion;
(5) Mixing the virtual dispatching instruction voice and the dispatcher training voice to form an integral training sample of the voiceprint recognition model, and training the dispatcher voiceprint recognition model;
(6) After the model is successfully trained, testing the model by using a dispatcher to test a voice sample; and outputting a dispatcher identity authentication result, and determining whether the dispatcher identity authentication is successful.
2. The improved Synth2Aug based distribution network dispatcher identity verification method of claim 1, wherein,
in the step (3), the text-to-speech conversion module is realized by adopting a Tacotron2 model, takes a dispatching text instruction as the input of the model, and outputs the dispatching personnel virtual dispatching instruction speech.
3. The improved Synth2Aug based distribution network dispatcher identity verification method of claim 1, wherein,
in the step (4), the original recording sample of the dispatcher is divided into dispatcher training voice and dispatcher testing voice according to the proportion of 7:3.
4. The improved Synth2Aug based distribution network dispatcher identity verification method of claim 1, wherein,
in the step (5), the virtual scheduling instruction speech and the scheduler training speech are respectively divided into N subsets with different sizes, namely a virtual scheduling instruction speech subset (vaudio_k, k=1, 2,..n) and an original scheduler training speech subset (audio_train_k, k=1, 2,..n), and the subsets with equal k are respectively taken out from the virtual scheduling instruction speech subset and the original scheduler training speech subset to be mixed, so as to form N mixed subsets, and finally the N mixed subsets are input into a voiceprint recognition training model to perform model training.
5. The distribution network dispatcher identity verification system based on the improved Synth2Aug is characterized by comprising a dispatching text instruction generation module, a text voice conversion module, a dispatcher original recording sampling module, a dispatcher voiceprint recognition model training module and a distribution network dispatcher identity verification module;
the dispatch text instruction generation module forms different dispatch text instructions by using different MTTG models according to given dispatch instruction keywords; the method comprises an MTTG_1 model and an MTTG_2 model, wherein the MTTG_1 model uses a bidirectional LSTM as a core module to generate a dispatching text instruction, and the use of modification components is less; the MTTG_2 model uses GRU as a core module, a scheduling text instruction is generated, and corresponding modification components are used, wherein the modification components of the MTTG_2 model are more than those of the MTTG_1 model;
the text-to-speech conversion module is used for converting the obtained dispatching text instruction into virtual dispatching instruction speech;
the dispatcher original recording sampling module is used for carrying out dispatcher original recording sampling and dividing the sampling sample into dispatcher training voice and dispatcher testing voice according to a certain proportion;
the dispatcher voiceprint recognition model training module mixes the virtual dispatching instruction voice with dispatcher training voice to form an integral training sample of the voiceprint recognition model, and trains the dispatcher voiceprint recognition model;
and the distribution network dispatcher identity verification module is used for testing the model by using a dispatcher test voice sample, outputting a dispatcher identity verification result and determining whether the dispatcher identity verification is successful.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110896813.3A CN113488061B (en) | 2021-08-05 | 2021-08-05 | Distribution network dispatcher identity verification method and system based on improved Synth2Aug |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110896813.3A CN113488061B (en) | 2021-08-05 | 2021-08-05 | Distribution network dispatcher identity verification method and system based on improved Synth2Aug |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113488061A CN113488061A (en) | 2021-10-08 |
| CN113488061B true CN113488061B (en) | 2024-02-23 |
Family
ID=77945676
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110896813.3A Active CN113488061B (en) | 2021-08-05 | 2021-08-05 | Distribution network dispatcher identity verification method and system based on improved Synth2Aug |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113488061B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114120975A (en) * | 2021-11-11 | 2022-03-01 | 北京有竹居网络技术有限公司 | Method, apparatus and storage medium for speech recognition punctuation recovery |
| CN114171011A (en) * | 2021-12-25 | 2022-03-11 | 中国人民解放军战略支援部队信息工程大学 | Monophonic voice recognition method and system based on railway scheduling |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1362430A2 (en) * | 2001-02-20 | 2003-11-19 | Michael D. Ellis | Modular personal network systems and methods |
| CN102708867A (en) * | 2012-05-30 | 2012-10-03 | 北京正鹰科技有限责任公司 | Method and system for identifying faked identity by preventing faked recordings based on voiceprint and voice |
| CN104732974A (en) * | 2013-12-19 | 2015-06-24 | 国家电网公司 | Intelligent recording and recognizing system |
| CN110224483A (en) * | 2019-04-25 | 2019-09-10 | 宁波天灵信息科技有限公司 | 3D dispatching of power netwoks information display system |
| CN111489746A (en) * | 2020-03-05 | 2020-08-04 | 国网浙江省电力有限公司 | Power grid dispatching voice recognition language model construction method based on BERT |
| CN111755001A (en) * | 2020-05-07 | 2020-10-09 | 国网山东省电力公司信息通信公司 | Artificial intelligence-based power grid rapid dispatching and commanding system and method |
| CN112164404A (en) * | 2020-10-28 | 2021-01-01 | 广西电网有限责任公司贺州供电局 | Remote identity authentication method and system based on voiceprint recognition technology |
| CN112420025A (en) * | 2020-11-09 | 2021-02-26 | 国网山东省电力公司临沂供电公司 | Intelligent voice recognition conversion method and system for power dispatching |
| CN113094681A (en) * | 2021-04-09 | 2021-07-09 | 广东电网有限责任公司 | Identity recognition method, system, equipment and storage medium |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050065779A1 (en) * | 2001-03-29 | 2005-03-24 | Gilad Odinak | Comprehensive multiple feature telematics system |
| CN110619867B (en) * | 2019-09-27 | 2020-11-03 | 百度在线网络技术（北京）有限公司 | Training method and device of speech synthesis model, electronic equipment and storage medium |
-
2021
- 2021-08-05 CN CN202110896813.3A patent/CN113488061B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1362430A2 (en) * | 2001-02-20 | 2003-11-19 | Michael D. Ellis | Modular personal network systems and methods |
| CN102708867A (en) * | 2012-05-30 | 2012-10-03 | 北京正鹰科技有限责任公司 | Method and system for identifying faked identity by preventing faked recordings based on voiceprint and voice |
| CN104732974A (en) * | 2013-12-19 | 2015-06-24 | 国家电网公司 | Intelligent recording and recognizing system |
| CN110224483A (en) * | 2019-04-25 | 2019-09-10 | 宁波天灵信息科技有限公司 | 3D dispatching of power netwoks information display system |
| CN111489746A (en) * | 2020-03-05 | 2020-08-04 | 国网浙江省电力有限公司 | Power grid dispatching voice recognition language model construction method based on BERT |
| CN111755001A (en) * | 2020-05-07 | 2020-10-09 | 国网山东省电力公司信息通信公司 | Artificial intelligence-based power grid rapid dispatching and commanding system and method |
| CN112164404A (en) * | 2020-10-28 | 2021-01-01 | 广西电网有限责任公司贺州供电局 | Remote identity authentication method and system based on voiceprint recognition technology |
| CN112420025A (en) * | 2020-11-09 | 2021-02-26 | 国网山东省电力公司临沂供电公司 | Intelligent voice recognition conversion method and system for power dispatching |
| CN113094681A (en) * | 2021-04-09 | 2021-07-09 | 广东电网有限责任公司 | Identity recognition method, system, equipment and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| SYNTH2AUG: CROSS-DOMAIN SPEAKER RECOGNITION WITH TTS SYNTHESIZED SPEECH;Yiling Huang 等;《IEEE Spoken Language Technology Workshop》;第316-319页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113488061A (en) | 2021-10-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113488061B (en) | Distribution network dispatcher identity verification method and system based on improved Synth2Aug | |
| CN111667812B (en) | Speech synthesis method, device, equipment and storage medium | |
| CN109241330A (en) | The method, apparatus, equipment and medium of key phrase in audio for identification | |
| CN112086086A (en) | Speech synthesis method, device, equipment and computer readable storage medium | |
| CN107657017A (en) | Method and apparatus for providing voice service | |
| Komatani et al. | User modeling in spoken dialogue systems to generate flexible guidance | |
| DeVault et al. | Can I finish? Learning when to respond to incremental interpretation results in interactive dialogue | |
| CN108184032B (en) | Service method and device of customer service system | |
| CN103366729A (en) | Speech dialogue system, terminal apparatus, and data center apparatus | |
| CA2540557A1 (en) | Apparatus and method for model adaptation for spoken language understanding | |
| CN112632288A (en) | Power dispatching system and method based on knowledge graph | |
| CN112489618A (en) | Neural text-to-speech synthesis using multi-level contextual features | |
| CN111145719A (en) | Data labeling method and device for Chinese-English mixing and tone labeling | |
| CN110517668A (en) | A kind of Chinese and English mixing voice identifying system and method | |
| CN111627420A (en) | Specific-speaker emotion voice synthesis method and device under extremely low resources | |
| CN113140219A (en) | Regulation and control instruction generation method and device, electronic equipment and storage medium | |
| CN102196100A (en) | Instant call translation system and method | |
| CN113299272B (en) | Speech synthesis model training and speech synthesis method, equipment and storage medium | |
| CN113689851A (en) | Scheduling professional language understanding system and method | |
| CN113159483A (en) | Task scheduling method and device based on RPA and AI, robot and medium | |
| CN116366768A (en) | Intelligent speech engine capability sharing model building and evaluating method and system | |
| Lin | Research on vehicle service simulation dispatching telephone system based on natural language processing | |
| CN111369975A (en) | University music scoring method, device, equipment and storage medium based on artificial intelligence | |
| CN113515950B (en) | Natural language processing semantic analysis method suitable for intelligent power dispatching | |
| CN114987581B (en) | Train dispatching method, train dispatching device and medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |