CN117133294A - Improved intelligent work order system for extracting voice feature unit based on LSTM model - Google Patents
Improved intelligent work order system for extracting voice feature unit based on LSTM model Download PDFInfo
- Publication number
- CN117133294A CN117133294A CN202311395210.0A CN202311395210A CN117133294A CN 117133294 A CN117133294 A CN 117133294A CN 202311395210 A CN202311395210 A CN 202311395210A CN 117133294 A CN117133294 A CN 117133294A
- Authority
- CN
- China
- Prior art keywords
- data
- module
- model
- unit
- work order
- 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.)
- Granted
Links
- 238000012545 processing Methods 0.000 claims abstract description 99
- 238000000605 extraction Methods 0.000 claims abstract description 71
- 238000003860 storage Methods 0.000 claims description 33
- 238000012986 modification Methods 0.000 claims description 19
- 230000004048 modification Effects 0.000 claims description 19
- 238000000034 method Methods 0.000 claims description 14
- 238000012937 correction Methods 0.000 claims description 13
- 239000000284 extract Substances 0.000 claims description 11
- 230000005540 biological transmission Effects 0.000 claims description 7
- 230000010354 integration Effects 0.000 abstract description 7
- 238000012423 maintenance Methods 0.000 description 9
- 238000010586 diagram Methods 0.000 description 6
- 238000004891 communication Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000004519 manufacturing process 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
- G10L15/26—Speech to text systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/232—Orthographic correction, e.g. spell checking or vowelisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
-
- 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
- G10L15/02—Feature extraction for speech recognition; Selection of recognition unit
-
- 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)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Human Computer Interaction (AREA)
- Acoustics & Sound (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Machine Translation (AREA)
Abstract
The invention relates to the technical field of worksheet combination, in particular to an improved intelligent worksheet system based on voice feature unit extraction of an LSTM model. The system comprises an audio processing unit, a feature extraction unit, a model operation unit and a work order output unit. According to the invention, the audio data are converted into text data through the audio processing unit, the audio processing unit is acquired and processed through the model operation unit, the characteristics in the text data are acquired, and the work order is manufactured through the work order output unit according to the extracted characteristic data, so that the audio data in a plurality of subsystems are integrated together, the accurate extraction of the characteristics is realized, the integration of the subsequent staff on the data of each subsystem is facilitated, and the speed of integrating each system by the staff is increased.
Description
Technical Field
The invention relates to the technical field of worksheet combination, in particular to an improved intelligent worksheet system based on voice feature unit extraction of an LSTM model.
Background
In the field of operation and maintenance, there are a series of difficulties and challenges including history carry-over problems, complex system architecture, continuous business needs, insufficient personnel and difficulties in terms of safety, operation and maintenance personnel need to continuously learn new technologies, update knowledge, enhance communication and cooperation, ensure high availability and safety of the system, and in order to overcome these problems, the operation and maintenance personnel need to have high level of technical capability and professional literacy, and meanwhile need to have flexibility and innovation capability.
The operation and maintenance work order system has the following problems in the using process: the operation and maintenance work order system is provided with a plurality of subsystems, the data patterns in the subsystems are different, such as text data, audio data and the like, the text data is convenient to process, but the audio data cannot be integrated with the text data when being processed, so that information and data transmission is difficult, and communication and processing problems are caused.
Disclosure of Invention
The present invention is directed to an improved intelligent work order system for extracting speech feature units based on the LSTM model, so as to solve the above-mentioned problems in the prior art.
In order to achieve the above-mentioned purpose, the invention provides an improved intelligent work order system based on the voice characteristic unit extraction of LSTM model, comprising an audio processing unit, a characteristic extraction unit, a model operation unit and a work order output unit;
the audio processing unit receives the audio data and performs text content identification according to the content of the audio data;
the model operation unit receives text data recognized by the audio processing unit, processes the text data according to the established model, and transmits the data to the feature extraction unit after the data is processed;
the feature extraction unit is used for respectively receiving the text data transmitted by the audio processing unit and the data transmitted by the model operation unit, respectively extracting features of the data transmitted by the audio processing unit and the model operation unit, and transmitting the extracted data to the model operation unit for storage;
the work order output unit receives the data stored in the model operation unit, compares the data extracted by the audio processing unit and the model operation unit by the feature extraction unit, and after the comparison is completed, carries out model modification feedback to the model operation unit, and after the comparison of the work order output units is consistent, carries out work order output.
As a further improvement of the technical scheme, the audio processing unit comprises an audio acquisition module, an audio voice recognition module and a text contrast error correction module;
the audio acquisition module acquires audio data of different ends to acquire audio data content;
the audio voice recognition module recognizes the content in the audio data acquired by the audio acquisition module, and acquires text information data corresponding to the audio data;
the text comparison error correction module receives the text information data acquired in the audio voice recognition module and the audio data acquired in the audio acquisition module, compares the audio data with the text information data, and modifies the text information data when errors occur in comparison.
As a further improvement of the technical scheme, the model running unit comprises a model building module and a model processing module;
the model establishing module establishes a feature extraction model through an LSTM algorithm and transmits the established feature extraction model to the model processing module;
the model processing module receives the text information data checked by the text comparison error correction module, and extracts the characteristics in the text data through the characteristic extraction model to obtain the characteristic data in the text information data.
As a further improvement of the technical scheme, the feature extraction model extracts features in text data by the steps of:
(1) firstly, carrying out feature extraction on text information data for the first time, and marking the extracted information as first-time data;
(2) selectively forgetting the first data, extracting the characteristics of the text information data for the second time after forgetting the first data, and eliminating the data which has no meaning in the information extraction at the time of extraction to obtain the second data;
(3) superposing the information of the first data and the second data, and integrating the characteristic data in the superposed data, namely combining the same characteristic data in the first data and the second data, integrating different characteristic data, wherein the integrated data is defined as second complete data;
(4) selectively forgetting the data after the superposition in the step (3), extracting the characteristics of the text information data for the third time after forgetting, and superposing the data extracted for the third time and the second complete data to obtain third complete data;
(5) repeating the steps (3) and (4), overlapping the data acquired before and the newly formed data until the characteristic data in the newly formed data are all in the previously overlapped data, stopping extracting the characteristics at the moment, and outputting the overlapped data.
As a further improvement of the technical scheme, the model running unit further comprises a data set storage module;
the data set storage module stores work order data in the past period, receives data processed by the model processing module, records the step of processing the data by the model processing module, receives the data transmitted by the feature extraction unit, and stores data information transmitted by the feature extraction unit.
As a further improvement of the technical scheme, the feature extraction unit comprises a feature recognition module and a feature comparison module;
the feature recognition module receives the text data transmitted by the text comparison error correction module, recognizes key words in the text data and acquires feature data in a maximum range;
the feature comparison module receives the data in the model processing module and the feature recognition module, compares the data in the model processing module and the feature recognition module with the data stored in the data set storage module, determines whether the same or similar data are stored in the data set storage module, marks the same or similar data in the feature recognition module and the model processing module as the data set storage module when the same or similar data exist, and transmits the data transmitted by the model processing module and the feature recognition module to the data set storage module for storage after the data are marked.
As a further improvement of the technical scheme, the work order output unit comprises a data merging processing module, a data comparison module and a work order output module;
the data merging processing module receives the data transmitted by the data set storage module, merges the same characteristics in the data transmitted by the data set storage module and the data transmitted by the model processing module, transmits the merged data to the work order output module after merging, and simultaneously transmits the uncombined data to the data comparison module;
the data comparison module receives the data which are not combined and transmitted by the data combination processing module, compares the data which are not combined with each other, and determines the difference between the data which are not combined in the comparison process;
the work order output module receives the data combined in the data combining processing module and the data with smaller difference in the data comparison module, and makes the work order according to the received data, and transmits the made work order data.
As a further improvement of the technical scheme, the model operation unit comprises a model modification module, and the work order output unit comprises a model data feedback module;
the model data feedback module receives the characteristic data with larger difference in the data comparison module, identifies the data with larger difference, determines the content data in the text data corresponding to the characteristic data with larger difference, and transmits the partial content data to the model modification module;
the model modification module receives the data transmitted by the model data feedback module, extracts the section of content from the text information data according to the transmission content in the model data feedback module, and transmits the section of content to the model establishment module, so that the model establishment module carries out model modification according to the section of content and the characteristic data with larger difference.
Compared with the prior art, the invention has the beneficial effects that:
1. in the improved intelligent work order system based on the LSTM model voice feature unit extraction, the audio data are converted into text data through the audio processing unit, the audio processing unit is acquired and processed through the model operation unit to acquire the features in the text data, and the work order is manufactured through the work order output unit according to the extracted feature data, so that the audio data in a plurality of subsystems are integrated together, the features are accurately extracted, the integration of subsequent staff to each subsystem data is facilitated, and the speed of integrating each system by the staff is accelerated.
2. In the improved intelligent work order system based on the LSTM model and for extracting the voice feature unit, text data in an audio processing unit is received through a feature extraction unit, feature extraction is carried out according to the text data, data processed by a model processing module are received, the data processed by the model processing module and the feature data extracted by the feature extraction unit are stored in a data set storage module, judgment is carried out through a work order output unit, data with larger difference are fed back through a model data feedback module, a model modification module controls a model building module to adjust a model according to fed back content, so that the accuracy of the model on extracting the feature data in the text data is improved, continuous optimization of the model is carried out, the processing of data in a subsystem by workers is facilitated, and the integration of data in each subsystem by subsequent workers is facilitated.
Drawings
FIG. 1 is an overall block diagram of embodiment 1 of the present invention;
fig. 2 is a block diagram of an audio processing unit according to embodiment 1 of the present invention;
FIG. 3 is a block diagram of a feature extraction unit according to embodiment 1 of the present invention;
FIG. 4 is a block diagram of a model execution unit according to embodiment 1 of the present invention;
FIG. 5 is a block diagram of a work order output unit according to embodiment 1 of the present invention;
fig. 6 is an overall combined block diagram of embodiment 1 of the present invention.
The meaning of each reference sign in the figure is:
1. an audio processing unit; 11. an audio acquisition module; 12. an audio speech recognition module; 13. a text contrast error correction module;
2. a feature extraction unit; 21. a feature recognition module; 22. the characteristic comparison module;
3. a model operation unit; 31. a model building module; 32. a model processing module; 33. a data set storage module; 34. a model modification module;
4. a work order output unit; 41. a data merging processing module; 42. a data comparison module; 43. a work order output module; 44. and the model data feedback module.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Example 1
The operation and maintenance work order system has the following problems in the using process: the operation and maintenance work order system is provided with a plurality of subsystems, the data patterns in the subsystems are different, such as text data, audio data and the like, the text data is convenient to process, but the audio data cannot be integrated with the text data when being processed, so that information and data transmission is difficult, and communication and processing problems are caused.
The invention provides an improved intelligent work order system for extracting voice characteristic units based on an LSTM model, which is shown in figures 1-6, and comprises an audio processing unit 1, a characteristic extracting unit 2, a model running unit 3 and a work order output unit 4;
the audio processing unit 1 receives the audio data, performs text content identification according to the content of the audio data, and facilitates the integration processing of worksheet output unit 4 by identifying the content in the audio data and converting the content into a text form so as to facilitate the extraction of characteristic data according to the text data, and simultaneously converts the audio data into a text form, so that the data among a plurality of subsystems are convenient for mutual transmission and identification, and the data transmission among each subsystem is accelerated;
the audio processing unit 1 comprises an audio acquisition module 11, an audio voice recognition module 12 and a text contrast error correction module 13;
the audio acquisition module 11 acquires audio data of different ends to acquire audio data content, and meanwhile, in the process of acquiring the audio data by the audio acquisition module 11, denoising and denoising the acquired audio data to remove content affecting text recognition in the audio data so as to improve the accuracy of subsequent processing;
the audio voice recognition module 12 recognizes the content in the audio data acquired by the audio acquisition module 11, and acquires text information data corresponding to the audio data;
the text comparison error correction module 13 receives the text information data acquired in the audio voice recognition module 12 and the audio data acquired in the audio acquisition module 11, compares the audio data with the text information data, improves the accuracy of the text information data corresponding to the audio data acquired by the audio voice recognition module 12, and modifies the text information data when errors occur in comparison, so that the accuracy of converting the text information data by the audio data is improved, and the text information data acquired by the audio voice recognition module 12 is processed conveniently.
The model running unit 3 receives the text data identified by the audio processing unit 1, processes the text data according to the established model, and transmits the data to the feature extraction unit 2 after the data is processed;
the model running unit 3 includes a model establishing module 31 and a model processing module 32;
the model establishing module 31 establishes a feature extraction model through an LSTM algorithm and transmits the established feature extraction model to the model processing module 32, the model established by the model establishing module 31 is an unfixed model, and in the data processing process, when the data processed by the model has problems, the model is modified conveniently;
the model processing module 32 receives the text information data checked by the text comparison error correction module 13, and extracts the features in the text data through the feature extraction model to obtain feature data in the text information data;
the feature extraction model extracts features in text data by the steps of:
(1) firstly, carrying out feature extraction on text information data for the first time, and marking the extracted information as first-time data;
(2) selectively forgetting the first data, extracting the characteristics of the text information data for the second time after forgetting the first data, and eliminating the data which has no meaning in the information extraction at the time of extraction to obtain the second data;
(3) superposing the information of the first data and the second data, and integrating the characteristic data in the superposed data, namely combining the same characteristic data in the first data and the second data, integrating different characteristic data, wherein the integrated data is defined as second complete data;
(4) selectively forgetting the data after the superposition in the step (3), extracting the characteristics of the text information data for the third time after forgetting, and superposing the data extracted for the third time and the second complete data to obtain third complete data;
(5) repeating the steps (3) and (4), overlapping the data acquired before and the newly formed data until the characteristic data in the newly formed data are all in the previously overlapped data, stopping extracting the characteristics at the moment, and outputting the overlapped data.
The feature extraction model established by adopting the LSTM algorithm can process the feature extraction after multiple inputs of a user, combine the extracted features, extract the features of a group of data for multiple times, gradually correct the result in the output process and improve the accuracy of extracting the features of the data.
The model running unit 3 further comprises a data set storage module 33;
the data set storage module 33 stores work order data of the past period, receives data processed by the model processing module 32, records the step of processing the data by the model processing module 32, receives the data transmitted by the feature extraction unit 2, stores data information transmitted by the feature extraction unit 2, judges the condition that the model established by the model establishment module 31 performs feature extraction on text data by recording the step of processing the data by the model processing module 32 each time, and when the number of the steps of processing the data by the model processing module 32 is small, the effect of extracting features in the text data by the model established by the model establishment module 31 is good, otherwise, the effect of extracting features by the established model is poor.
The feature extraction unit 2 receives the text data transmitted by the audio processing unit 1 and the data transmitted by the model operation unit 3 respectively, extracts features of the data transmitted by the audio processing unit 1 and the model operation unit 3 respectively, and transmits the extracted data to the model operation unit 3 for storage;
the feature extraction unit 2 includes a feature recognition module 21 and a feature comparison module 22;
the feature recognition module 21 receives the text data transmitted by the text comparison error correction module 13, recognizes key words in the text data and acquires the feature data in the maximum range;
the feature comparison module 22 receives the data in the model processing module 32 and the feature recognition module 21, compares the data in the model processing module 32, the feature recognition module 21 and the data stored in the data set storage module 33, determines whether the same or similar data is stored in the data set storage module 33, marks the same or similar data in the model processing module 32 and the feature recognition module 21 as the data set storage module 33 when the same or similar data exists, and after the data is marked, the feature comparison module 22 transmits the data transmitted by the model processing module 32 and the feature recognition module 21 to the data set storage module 33 for storage.
The work order output unit 4 receives the data stored in the model operation unit 3, compares the data extracted by the audio processing unit 1 and the model operation unit 3 by the feature extraction unit 2, and after the comparison is completed, carries out model modification feedback to the model operation unit 3, and after the comparison of the work order output unit 4 is consistent, carries out work order output;
the work order output unit 4 includes a data merging processing module 41, a data comparison module 42, and a work order output module 43;
the data merging processing module 41 receives the data transmitted by the data set storage module 33, merges the same features in the data transmitted by the data set storage module 33 and the data transmitted by the model processing module 32, transmits the merged data to the work order output module 43 after merging, and simultaneously transmits the uncombined data to the data comparison module 42;
the data comparison module 42 receives the data which are not combined and transmitted by the data combining processing module 41, compares the data which are not combined with each other, determines the difference between the data which are not combined in the comparison process, and integrates and transmits the data to the work order output module 43 when the difference between the data is smaller and the data only belong to the same nouns of different vocabulary descriptions;
the work order output module 43 receives the data combined in the data combining processing module 41 and the data with smaller difference in the data comparison module 42, and makes work orders according to the received data, and transmits the made work order data;
after the data is compared by the data comparison module 42, the data processed by the feature extraction unit 2 and the same content of the data processed by the model operation unit 3 are subjected to work order manufacture, so that the accuracy of the manufactured work order content is ensured, the work order content of a plurality of subsystems is conveniently processed by workers, and the integration of the data by the workers is facilitated.
The audio data are converted into text data through the audio processing unit 1, the audio processing unit 1 is acquired and processed through the model running unit 3, the characteristics in the text data are acquired, and the work order is manufactured through the work order output unit 4 according to the extracted characteristic data, so that the audio data in a plurality of subsystems are integrated together, the accurate extraction of the characteristics is achieved, the integration of the subsequent staff on the data of each subsystem is facilitated, and the speed of integrating each system by the staff is increased.
The model running unit 3 comprises a model modification module 34, and the work order output unit 4 comprises a model data feedback module 44;
the model data feedback module 44 receives the feature data with larger difference in the data comparison module 42, identifies the data with larger difference, determines content data in text data corresponding to the feature data with larger difference, and transmits the part of the content data to the model modification module 34;
the model modification module 34 receives the data transmitted from the model data feedback module 44, extracts the piece of content from the text information data according to the transmission content in the model data feedback module 44, and transmits the piece of content to the model establishment module 31, so that the model establishment module 31 carries out model modification according to the piece of content and the characteristic data with larger difference.
The text data in the audio processing unit 1 is received through the feature extraction unit 2, feature extraction is carried out according to the text data, meanwhile, data processed by the model processing module 32 is received, the data processed by the model processing module 32 and the feature data extracted by the feature extraction unit 2 are stored in the data set storage module 33, judgment is carried out through the work order output unit 4, data with larger difference are fed back through the model data feedback module 44, the model modification module 34 controls the model establishment module 31 to adjust the model according to the fed back content, the accuracy of the model for extracting the feature data in the text data is improved, continuous optimization of the model is carried out, the processing of the data in the subsystem by workers is facilitated, and the integration of the data in each subsystem by the workers is facilitated.
The data in the audio processing unit 1 and the model running unit 3 are respectively extracted by the feature extraction unit 2, the respectively extracted data are compared by the work order output unit 4, so that whether the model established by the model running unit 3 is correct or not is judged, the model is repaired when the model is wrong, the model is updated so as to facilitate the subsequent use of the model, and meanwhile, after the comparison of the work order output unit 4, the data in different work orders are combined together, the data in different positions acquired by the audio processing unit 1 are integrated, and unified processing of the data is achieved, so that the processing of operation and maintenance personnel on the data is facilitated, and the work efficiency and the work quality of the operation and maintenance personnel are improved.
The foregoing has shown and described the basic principles, principal features and advantages of the invention. It will be understood by those skilled in the art that the present invention is not limited to the above-described embodiments, and that the above-described embodiments and descriptions are only preferred embodiments of the present invention, and are not intended to limit the invention, and that various changes and modifications may be made therein without departing from the spirit and scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (8)
1. An improved intelligent work order system based on LSTM model voice feature unit extraction, its characterized in that: the system comprises an audio processing unit (1), a feature extraction unit (2), a model operation unit (3) and a work order output unit (4);
the audio processing unit (1) receives audio data and performs text content identification according to the content of the audio data;
the model running unit (3) receives the text data recognized by the audio processing unit (1), processes the text data according to the established model, and transmits the data to the feature extraction unit (2) after the data is processed;
the feature extraction unit (2) is used for respectively receiving the text data transmitted by the audio processing unit (1) and the data transmitted by the model operation unit (3), respectively extracting features of the data transmitted by the audio processing unit (1) and the model operation unit (3), and transmitting the extracted data to the model operation unit (3) for storage;
the work order output unit (4) receives the data stored in the model operation unit (3), compares the data extracted by the audio processing unit (1) and the model operation unit (3) by the feature extraction unit (2), and after the comparison is completed, carries out model modification feedback to the model operation unit (3), and outputs the work order after the comparison of the work order output unit (4) is consistent.
2. The improved intelligent work order system for LSTM model based speech feature unit extraction of claim 1, wherein: the audio processing unit (1) comprises an audio acquisition module (11), an audio voice recognition module (12) and a text contrast error correction module (13);
the audio acquisition module (11) acquires audio data of different ends to acquire audio data content;
the audio voice recognition module (12) recognizes the content in the audio data acquired by the audio acquisition module (11) and acquires text information data corresponding to the audio data;
the text comparison error correction module (13) receives the text information data acquired in the audio voice recognition module (12) and the audio data acquired in the audio acquisition module (11), compares the audio data with the text information data, and modifies the text information data when errors occur in comparison.
3. The improved intelligent work order system for LSTM model based speech feature unit extraction of claim 2, wherein: the model running unit (3) comprises a model establishing module (31) and a model processing module (32);
the model establishing module (31) establishes a feature extraction model through an LSTM algorithm and transmits the established feature extraction model to the model processing module (32);
the model processing module (32) receives the text information data checked by the text comparison error correction module (13), and extracts the characteristics in the text data through the characteristic extraction model to obtain the characteristic data in the text information data.
4. The improved intelligent work order system for LSTM model based speech feature unit extraction of claim 3, wherein: the feature extraction model extracts features in text data, and the feature extraction model comprises the following steps:
(1) firstly, carrying out feature extraction on text information data for the first time, and marking the extracted information as first-time data;
(2) selectively forgetting the first data, extracting the characteristics of the text information data for the second time after forgetting the first data, and eliminating the data which has no meaning in the information extraction at the time of extraction to obtain the second data;
(3) superposing the information of the first data and the second data, and integrating the characteristic data in the superposed data, namely combining the same characteristic data in the first data and the second data, integrating different characteristic data, wherein the integrated data is defined as second complete data;
(4) selectively forgetting the data after the superposition in the step (3), extracting the characteristics of the text information data for the third time after forgetting, and superposing the data extracted for the third time and the second complete data to obtain third complete data;
(5) repeating the steps (3) and (4), overlapping the data acquired before and the newly formed data until the characteristic data in the newly formed data are all in the previously overlapped data, stopping extracting the characteristics at the moment, and outputting the overlapped data.
5. The improved intelligent work order system for LSTM model based speech feature unit extraction of claim 3, wherein: the model running unit (3) further comprises a data set storage module (33);
the data set storage module (33) stores work order data in the past period, receives data processed by the model processing module (32) at the same time, records the step of processing the data by the model processing module (32), receives the data transmitted by the feature extraction unit (2), and stores data information transmitted by the feature extraction unit (2).
6. The improved intelligent work order system for LSTM model based speech feature unit extraction of claim 5, wherein: the feature extraction unit (2) comprises a feature identification module (21) and a feature comparison module (22);
the feature recognition module (21) receives the text data transmitted by the text comparison error correction module (13), recognizes key words in the text data and acquires the feature data in the maximum range;
the feature comparison module (22) receives data in the model processing module (32) and the feature recognition module (21), compares the data in the model processing module (32), the feature recognition module (21) and the data stored in the data set storage module (33), determines whether the same or similar data are stored in the data set storage module (33), marks the same or similar data in the model processing module (21) and the model processing module (32) as the data in the data set storage module (33) when the same or similar data exist, and after the data are marked, the feature comparison module (22) transmits the data transmitted by the model processing module (32) and the feature recognition module (21) to the data set storage module (33) for storage.
7. The improved intelligent work order system for LSTM model based speech feature unit extraction of claim 6, wherein: the work order output unit (4) comprises a data merging processing module (41), a data comparison module (42) and a work order output module (43);
the data merging processing module (41) receives the data transmitted by the data set storage module (33), merges the same features in the data transmitted by the data set storage module (33) and the data transmitted by the model processing module (32), transmits the merged data to the work order output module (43) after merging, and simultaneously transmits the uncombined data to the data comparison module (42);
the data comparison module (42) receives the data which are not combined and transmitted by the data combination processing module (41), compares the data which are not combined with each other, and determines the difference between the data which are not combined in the comparison process;
the work order output module (43) receives the data combined in the data combination processing module (41) and the data with smaller difference in the data comparison module (42), and makes work orders according to the received data, and transmits the made work order data.
8. The improved intelligent work order system for LSTM model based speech feature unit extraction of claim 7, wherein: the model running unit (3) comprises a model modification module (34), and the work order output unit (4) comprises a model data feedback module (44);
the model data feedback module (44) receives the characteristic data with larger difference in the data comparison module (42), identifies the data with larger difference, determines content data in text data corresponding to the characteristic data with larger difference, and transmits the partial content data to the model modification module (34);
the model modification module (34) receives the data transmitted by the model data feedback module (44), extracts the section of content from the text information data according to the transmission content in the model data feedback module (44), and transmits the section of content to the model establishment module (31), so that the model establishment module (31) carries out model modification according to the section of content and the characteristic data with larger difference.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311395210.0A CN117133294B (en) | 2023-10-26 | 2023-10-26 | Improved intelligent work order system for extracting voice feature unit based on LSTM model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311395210.0A CN117133294B (en) | 2023-10-26 | 2023-10-26 | Improved intelligent work order system for extracting voice feature unit based on LSTM model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117133294A true CN117133294A (en) | 2023-11-28 |
| CN117133294B CN117133294B (en) | 2024-01-05 |
Family
ID=88854895
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311395210.0A Active CN117133294B (en) | 2023-10-26 | 2023-10-26 | Improved intelligent work order system for extracting voice feature unit based on LSTM model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117133294B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20050101695A (en) * | 2004-04-19 | 2005-10-25 | 대한민국(전남대학교총장) | A system for statistical speech recognition using recognition results, and method thereof |
| US20080270137A1 (en) * | 2007-04-27 | 2008-10-30 | Dickson Craig B | Text to speech interactive voice response system |
| CN112567459A (en) * | 2018-08-24 | 2021-03-26 | 三菱电机株式会社 | Sound separation device, sound separation method, sound separation program, and sound separation system |
| CN113010680A (en) * | 2021-03-19 | 2021-06-22 | 国网河北省电力有限公司营销服务中心 | Electric power work order text classification method and device and terminal equipment |
| US11355122B1 (en) * | 2021-02-24 | 2022-06-07 | Conversenowai | Using machine learning to correct the output of an automatic speech recognition system |
| CN116226372A (en) * | 2022-12-28 | 2023-06-06 | 天津科技大学 | Bi-LSTM-CNN-based multi-modal voice emotion recognition method |
| CN116562270A (en) * | 2023-07-07 | 2023-08-08 | 天津亿科科技有限公司 | Natural language processing system supporting multi-mode input and method thereof |
-
2023
- 2023-10-26 CN CN202311395210.0A patent/CN117133294B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20050101695A (en) * | 2004-04-19 | 2005-10-25 | 대한민국(전남대학교총장) | A system for statistical speech recognition using recognition results, and method thereof |
| US20080270137A1 (en) * | 2007-04-27 | 2008-10-30 | Dickson Craig B | Text to speech interactive voice response system |
| CN112567459A (en) * | 2018-08-24 | 2021-03-26 | 三菱电机株式会社 | Sound separation device, sound separation method, sound separation program, and sound separation system |
| US11355122B1 (en) * | 2021-02-24 | 2022-06-07 | Conversenowai | Using machine learning to correct the output of an automatic speech recognition system |
| CN113010680A (en) * | 2021-03-19 | 2021-06-22 | 国网河北省电力有限公司营销服务中心 | Electric power work order text classification method and device and terminal equipment |
| CN116226372A (en) * | 2022-12-28 | 2023-06-06 | 天津科技大学 | Bi-LSTM-CNN-based multi-modal voice emotion recognition method |
| CN116562270A (en) * | 2023-07-07 | 2023-08-08 | 天津亿科科技有限公司 | Natural language processing system supporting multi-mode input and method thereof |
Non-Patent Citations (2)
| Title |
|---|
| 范华 等: "基于Bi-LSTM和TFIDF的工单事件提取", 电脑知识与技术, vol. 16, no. 04, pages 291 - 293 * |
| 黄秀彬 等: "基于ResNet-BiLS...模型的电力客服工单分类研究", 电子设计工程, vol. 30, no. 22, pages 179 - 183 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117133294B (en) | 2024-01-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107562918A (en) | A kind of mathematical problem knowledge point discovery and batch label acquisition method | |
| CN110890088B (en) | Voice information feedback method and device, computer equipment and storage medium | |
| CN107862005A (en) | User view recognition methods and device | |
| CN105261356A (en) | Voice recognition system and method | |
| CN111324708A (en) | Natural language processing system based on human-computer interaction | |
| CN105654025A (en) | Fingerprint identification method, apparatus and electronic equipment thereof | |
| CN116416480B (en) | Visual classification method and device based on multi-template prompt learning | |
| CN111695763B (en) | Scheduling system and method based on voice question and answer | |
| CN113129898A (en) | Machine-assisted conference recording system and method | |
| CN113841156B (en) | Control method and device based on image recognition | |
| CN110196897B (en) | Case identification method based on question and answer template | |
| CN116562270A (en) | Natural language processing system supporting multi-mode input and method thereof | |
| CN117133294B (en) | Improved intelligent work order system for extracting voice feature unit based on LSTM model | |
| CN107766256A (en) | A kind of method for transformation and system of software unit testing use-case | |
| CN110705217A (en) | Wrongly-written character detection method and device, computer storage medium and electronic equipment | |
| CN111081218A (en) | Voice recognition method and voice control system | |
| CN111341324B (en) | Fasttext model-based recognition error correction and training method | |
| CN113326367A (en) | Task type dialogue method and system based on end-to-end text generation | |
| CN115798485A (en) | Scheduling shift-to-shift method, system, equipment and storage medium | |
| CN112820292B (en) | Method, device, electronic device and storage medium for generating meeting summary | |
| US8051026B2 (en) | Rules collector system and method with user interaction | |
| CN111078674A (en) | Data identification and error correction method for distribution network equipment | |
| CN111950875A (en) | Intelligent contract reviewing method | |
| CN111739519B (en) | Speech recognition-based dialogue management processing method, device, equipment and medium | |
| CN111914777B (en) | Method and system for identifying robot instruction in cross-mode manner |
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 |