CN114332903A - Lute music score identification method and system based on end-to-end neural network - Google Patents
Lute music score identification method and system based on end-to-end neural network Download PDFInfo
- Publication number
- CN114332903A CN114332903A CN202111458277.5A CN202111458277A CN114332903A CN 114332903 A CN114332903 A CN 114332903A CN 202111458277 A CN202111458277 A CN 202111458277A CN 114332903 A CN114332903 A CN 114332903A
- Authority
- CN
- China
- Prior art keywords
- lute
- music score
- music
- file
- sub
- 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
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 36
- 238000000034 method Methods 0.000 title claims abstract description 31
- 238000012549 training Methods 0.000 claims abstract description 26
- 238000007781 pre-processing Methods 0.000 claims abstract description 23
- 230000003595 spectral effect Effects 0.000 claims description 24
- 230000033764 rhythmic process Effects 0.000 claims description 12
- 238000012937 correction Methods 0.000 claims description 10
- 238000006243 chemical reaction Methods 0.000 claims description 6
- 230000000306 recurrent effect Effects 0.000 claims description 6
- 230000006870 function Effects 0.000 description 16
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000002123 temporal effect Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
Images
Landscapes
- Auxiliary Devices For Music (AREA)
- Character Discrimination (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides a lute music score recognition method and a lute music score recognition system based on an end-to-end neural network, which belong to the technical field of music score recognition, and the method comprises the following steps: s10, acquiring a large number of lute music score pictures and music XML files corresponding to the lute music score pictures, and preprocessing the lute music score pictures; step S20, cutting each music score picture according to lines to obtain lute music score sub-pictures, and cutting corresponding music XML files based on the lute music score sub-pictures to obtain music XML sub-files; step S30, converting each MusicXML subfile into an MEI file, and converting each MEI file into a semantic file based on a preset semantic dictionary; step S40, establishing a lute music score recognition model based on an end-to-end neural network, and training the lute music score recognition model by utilizing sub-pictures of the lute music scores and semantic files; and step S50, carrying out lute music score recognition by using the trained lute music score recognition model. The invention has the advantages that: the lute music score can be automatically identified, and the digitization efficiency of the lute music score is greatly improved.
Description
Technical Field
The invention relates to the technical field of music score identification, in particular to a lute music score identification method and system based on an end-to-end neural network.
Background
Music score is a method for recording music by symbols, conventionally, music is mostly spread by hand-written music score, and in order to protect and spread the music heritage, it is important to digitize and store the music score. However, the manual recording of music scores one by one is time-consuming, labor-consuming and error-prone, and therefore, a need for automatic recognition of music scores arises.
However, in the prior art, there is no method for automatically identifying the lute music score. Therefore, how to provide a lute music score recognition method and system based on an end-to-end neural network to realize automatic recognition of the lute music score becomes a technical problem to be solved urgently.
Disclosure of Invention
The invention aims to solve the technical problem of providing a lute music score recognition method and system based on an end-to-end neural network, and realizing automatic recognition of the lute music score.
In a first aspect, the invention provides a lute music score recognition method based on an end-to-end neural network, which comprises the following steps:
s10, acquiring a large number of lute music score pictures and music XML files corresponding to the lute music score pictures, and preprocessing the lute music score pictures;
step S20, cutting each lute music score picture according to lines to obtain a plurality of lute music score sub-pictures, and cutting a corresponding music XML file based on the lute music score sub-pictures to obtain a plurality of music XML sub-files;
step S30, converting each MusicXML subfile into an MEI file, and converting each MEI file into a semantic file based on a preset semantic dictionary;
step S40, establishing a lute music score recognition model based on an end-to-end neural network, and training the lute music score recognition model by utilizing sub-pictures and semantic files of the lute music score;
and step S50, automatically recognizing the lute music score by using the trained lute music score recognition model.
Further, in the step S10, the preprocessing the lute music score pictures specifically includes:
and sequentially carrying out the preprocessing of graying, binaryzation, noise reduction and inclination correction on each lute music score picture.
Further, the step S20 is specifically:
identifying the upper spectral line of each line of music score in each lute music score picture based on a template matching method, and further cutting each lute music score picture according to lines based on the upper spectral line to obtain a plurality of lute music score sub-pictures;
identifying vertical spectral lines in each lute music score sub-picture, judging the number of subsections contained in the lute music score sub-picture by using the vertical spectral lines, and cutting the musicXML file based on the number of subsections and a label carried by the musicXML file to obtain a plurality of musicXML sub-files.
Further, the step S30 is specifically:
converting each MusicXML subfile into an MEI file through an OMR tool website, and converting each MEI file into a semantic file based on a preset semantic dictionary;
the semantic file is a note information sequence comprising note information, fingering information and rhythm information.
Further, the step S40 is specifically:
establishing a lute music score recognition model based on an end-to-end convolution recurrent neural network, taking each lute music score sub-picture as the input of the lute music score recognition model, taking a semantic file corresponding to each lute music score sub-picture as the output of the lute music score recognition model, and training the lute music score recognition model;
and a CTC function is adopted as a loss function of the lute music score identification model.
In a second aspect, the invention provides a lute music score recognition system based on an end-to-end neural network, which comprises the following modules:
the lute music score image and music XML file acquisition module is used for acquiring a large number of lute music score images and music XML files corresponding to the lute music score images and preprocessing the lute music score images;
the lute music score picture and music XML file cutting module is used for cutting each lute music score picture according to lines to obtain a plurality of lute music score sub-pictures, and cutting a corresponding music XML file based on the lute music score sub-pictures to obtain a plurality of music XML sub-files;
the semantic conversion module is used for converting each MusicXML subfile into an MEI file and converting each MEI file into a semantic file based on a preset semantic dictionary;
the lute music score recognition model training module is used for establishing a lute music score recognition model based on an end-to-end neural network and training the lute music score recognition model by utilizing sub-pictures and semantic files of the lute music score;
and the lute music score automatic identification model is used for carrying out lute music score automatic identification by utilizing the trained lute music score identification model.
Further, in the lute music score picture and music xml file obtaining module, the preprocessing of each lute music score picture specifically comprises:
and sequentially carrying out the preprocessing of graying, binaryzation, noise reduction and inclination correction on each lute music score picture.
Further, the lute music score picture and music XML file cutting module specifically comprises:
identifying the upper spectral line of each line of music score in each lute music score picture based on a template matching method, and further cutting each lute music score picture according to lines based on the upper spectral line to obtain a plurality of lute music score sub-pictures;
identifying vertical spectral lines in each lute music score sub-picture, judging the number of subsections contained in the lute music score sub-picture by using the vertical spectral lines, and cutting the musicXML file based on the number of subsections and a label carried by the musicXML file to obtain a plurality of musicXML sub-files.
Further, the semantic conversion module is specifically:
converting each MusicXML subfile into an MEI file through an OMR tool website, and converting each MEI file into a semantic file based on a preset semantic dictionary;
the semantic file is a note information sequence comprising note information, fingering information and rhythm information.
Further, the lute music score recognition model training module specifically comprises:
establishing a lute music score recognition model based on an end-to-end convolution recurrent neural network, taking each lute music score sub-picture as the input of the lute music score recognition model, taking a semantic file corresponding to each lute music score sub-picture as the output of the lute music score recognition model, and training the lute music score recognition model;
and a CTC function is adopted as a loss function of the lute music score identification model.
The invention has the advantages that:
1. the lute music score recognition method includes the steps that a plurality of lute music score sub-pictures are obtained by cutting each lute music score picture according to lines, a corresponding music XML file is cut based on the lute music score sub-pictures and converted into an MEI file, the MEI file is converted into a semantic file comprising note information, fingering information and rhythm information in a semantic mode, a created lute music score recognition model is trained based on the lute music score sub-pictures and the semantic file, and finally automatic recognition of the lute music score can be achieved by the aid of the trained lute music score recognition model, so that automatic recognition of the lute music score is achieved, and compared with traditional manual recording, the digitization efficiency of the lute music score is greatly improved.
2. By cutting the lute music score picture according to lines and carrying out the preprocessing of graying, binaryzation, noise reduction and inclination correction, the method greatly improves the accuracy of identification compared with the method for directly identifying the whole lute music score picture.
Drawings
The invention will be further described with reference to the following examples with reference to the accompanying drawings.
Fig. 1 is a flow chart of a lute score identification method based on an end-to-end neural network.
Fig. 2 is a schematic structural diagram of a lute score recognition system based on an end-to-end neural network.
Fig. 3 is a schematic structural diagram of the lute music score recognition model of the invention.
Detailed Description
The technical scheme in the embodiment of the application has the following general idea: the method comprises the steps of cutting lute music score pictures according to lines to obtain a plurality of lute music score sub-pictures, cutting and semantically converting music XML files corresponding to the lute music score sub-pictures to obtain semantic files, training a lute music score recognition model based on the lute music score sub-pictures and the semantic files, and finally automatically recognizing lute music scores by using the trained lute music score recognition model.
Referring to fig. 1 to fig. 3, a lute score recognition method according to a preferred embodiment of the present invention includes the following steps:
s10, acquiring a large number of lute music score pictures and music XML files corresponding to the lute music score pictures, and preprocessing the lute music score pictures;
music Extensible Markup Language (MusicExtensible Markup Language) is an open Music symbol file format based on XML, and is used for Music exchange and Music distribution, and the MusicXML aims at creating a universal Music notation format;
step S20, cutting each lute music score picture according to lines to obtain a plurality of lute music score sub-pictures, and cutting a corresponding music XML file based on the lute music score sub-pictures to obtain a plurality of music XML sub-files;
step S30, converting each MusicXML subfile into an MEI file, and converting each MEI file into a semantic file based on a preset semantic dictionary;
step S40, establishing a lute music score recognition model based on an end-to-end neural network, and training the lute music score recognition model by utilizing sub-pictures and semantic files of the lute music score;
and step S50, automatically identifying the lute music score by using the trained lute music score identification model, namely cutting and preprocessing the lute music score picture to be identified according to lines and inputting the lute music score identification model for identification.
The lute music score recognition model is different from other models in training modes, namely the lute music score recognition model does not expect a data set to provide position information of music symbols in the lute music score, and from the music perspective, the position information of each music symbol in a lute music score picture does not need to be retrieved, and only corresponding note information, fingering information and rhythm information in the lute music score (staff) need to be known; therefore, only the context of the corresponding note information sequence needs to be searched, and finally, only one note information sequence needs to be output, the corresponding information of the note information sequence has corresponding explanation in the semantic dictionary, and according to the semantic dictionary, a user can convert the corresponding note information sequence into a control command of the robot.
In the step S10, the preprocessing of the lute music score pictures specifically includes:
and sequentially carrying out the preprocessing of graying, binaryzation, noise reduction and inclination correction on each lute music score picture. Graying is to convert a colorful lute music score picture into a grayscale image with only one channel, binarization is to convert the lute music score picture into black and white, noise reduction is to remove noise points of the lute music score picture, and inclination correction is to better identify and cut the music score of each line.
The step S20 specifically includes:
identifying the upper spectral line of each line of music score in each lute music score picture based on a template matching method, and further cutting each lute music score picture according to lines based on the upper spectral line to obtain a plurality of lute music score sub-pictures;
identifying vertical spectral lines in each lute music score sub-picture, judging the number of subsections contained in the lute music score sub-picture by using the vertical spectral lines, and cutting the musicXML file based on the number of subsections and a label carried by the musicXML file to obtain a plurality of musicXML sub-files. Namely, the musicXML file is cut by utilizing the corresponding relation between the number attribute and the number of the measure labels carried by the musicXML file.
The step S30 specifically includes:
converting each MusicXML subfile into an MEI file through an OMR tool website, and converting each MEI file into a semantic file based on a preset semantic dictionary; the semantic dictionary stores the one-to-one correspondence of each measure of the lute music score with note information, fingering information and rhythm information;
the semantic file is a note information sequence comprising note information, fingering information and rhythm information, and the semantic file is a complete data unit in a data set and is used for training the lute music score recognition model.
The OMR tool is a highly integrated open source code C and C + + component, can be used for constructing a large number of languages, and supports a plurality of different hardware and operating system platforms during running; these components include, but are not limited to: memory management, thread processing, platform port (abstraction) libraries, diagnostic support, monitoring support, garbage collection, and local real-time compilation.
The step S40 specifically includes:
establishing a lute music score recognition model based on an end-to-end Convolution Recurrent Neural Network (CRNN), taking each lute music score sub-picture as the input of the lute music score recognition model, taking a semantic file corresponding to each lute music score sub-picture as the output of the lute music score recognition model, and performing supervised training on the lute music score recognition model;
and a CTC function is adopted as a loss function of the lute music score identification model.
The convolution recursive neural network is formed by connecting the last layer of the CNN with the input of the first layer of the RNN, all output channels of the convolution part are connected into one image, and the column of the image is regarded as a single module of the convolution block; in the aspect of the identification accuracy of the lute music score identification model, in the end-to-end semantic information, for each music score, a training set only provides a corresponding expected note information sequence (semantic file) without position information of the note information sequence in an image, so that a CTC function is adopted as a loss function; the ctc (connectionist Temporal classification) function is a function for avoiding manual alignment of input and output, and is suitable for speech recognition or OCR.
The invention discloses a preferred embodiment of a lute music score recognition system based on an end-to-end neural network, which comprises the following modules:
the lute music score image and music XML file acquisition module is used for acquiring a large number of lute music score images and music XML files corresponding to the lute music score images and preprocessing the lute music score images;
music Extensible Markup Language (MusicExtensible Markup Language) is an open Music symbol file format based on XML, and is used for Music exchange and Music distribution, and the MusicXML aims at creating a universal Music notation format;
the lute music score picture and music XML file cutting module is used for cutting each lute music score picture according to lines to obtain a plurality of lute music score sub-pictures, and cutting a corresponding music XML file based on the lute music score sub-pictures to obtain a plurality of music XML sub-files;
the semantic conversion module is used for converting each MusicXML subfile into an MEI file and converting each MEI file into a semantic file based on a preset semantic dictionary;
the lute music score recognition model training module is used for establishing a lute music score recognition model based on an end-to-end neural network and training the lute music score recognition model by utilizing sub-pictures and semantic files of the lute music score;
the lute music score automatic identification model is used for carrying out lute music score automatic identification by utilizing the trained lute music score identification model, namely cutting a to-be-identified lute music score picture according to lines and inputting the to-be-identified lute music score identification model for identification after preprocessing.
The lute music score recognition model is different from other models in training modes, namely the lute music score recognition model does not expect a data set to provide position information of music symbols in the lute music score, and from the music perspective, the position information of each music symbol in a lute music score picture does not need to be retrieved, and only corresponding note information, fingering information and rhythm information in the lute music score (staff) need to be known; therefore, only the context of the corresponding note information sequence needs to be searched, and finally, only one note information sequence needs to be output, the corresponding information of the note information sequence has corresponding explanation in the semantic dictionary, and according to the semantic dictionary, a user can convert the corresponding note information sequence into a control command of the robot.
In the lute music score picture and music XML file obtaining module, the preprocessing of the lute music score pictures specifically comprises the following steps:
and sequentially carrying out the preprocessing of graying, binaryzation, noise reduction and inclination correction on each lute music score picture. Graying is to convert a colorful lute music score picture into a grayscale image with only one channel, binarization is to convert the lute music score picture into black and white, noise reduction is to remove noise points of the lute music score picture, and inclination correction is to better identify and cut the music score of each line.
The lute music score picture and music XML file cutting module specifically comprises:
identifying the upper spectral line of each line of music score in each lute music score picture based on a template matching method, and further cutting each lute music score picture according to lines based on the upper spectral line to obtain a plurality of lute music score sub-pictures;
identifying vertical spectral lines in each lute music score sub-picture, judging the number of subsections contained in the lute music score sub-picture by using the vertical spectral lines, and cutting the musicXML file based on the number of subsections and a label carried by the musicXML file to obtain a plurality of musicXML sub-files. Namely, the musicXML file is cut by utilizing the corresponding relation between the number attribute and the number of the measure labels carried by the musicXML file.
The semantic conversion module is specifically as follows:
converting each MusicXML subfile into an MEI file through an OMR tool website, and converting each MEI file into a semantic file based on a preset semantic dictionary; the semantic dictionary stores the one-to-one correspondence of each measure of the lute music score with note information, fingering information and rhythm information;
the semantic file is a note information sequence comprising note information, fingering information and rhythm information, and the semantic file is a complete data unit in a data set and is used for training the lute music score recognition model.
The OMR tool is a highly integrated open source code C and C + + component, can be used for constructing a large number of languages, and supports a plurality of different hardware and operating system platforms during running; these components include, but are not limited to: memory management, thread processing, platform port (abstraction) libraries, diagnostic support, monitoring support, garbage collection, and local real-time compilation.
The lute music score recognition model training module specifically comprises:
establishing a lute music score recognition model based on an end-to-end Convolution Recurrent Neural Network (CRNN), taking each lute music score sub-picture as the input of the lute music score recognition model, taking a semantic file corresponding to each lute music score sub-picture as the output of the lute music score recognition model, and performing supervised training on the lute music score recognition model;
and a CTC function is adopted as a loss function of the lute music score identification model.
The convolution recursive neural network is formed by connecting the last layer of the CNN with the input of the first layer of the RNN, all output channels of the convolution part are connected into one image, and the column of the image is regarded as a single module of the convolution block; in the aspect of the identification accuracy of the lute music score identification model, in the end-to-end semantic information, for each music score, a training set only provides a corresponding expected note information sequence (semantic file) without position information of the note information sequence in an image, so that a CTC function is adopted as a loss function; the ctc (connectionist Temporal classification) function is a function for avoiding manual alignment of input and output, and is suitable for speech recognition or OCR.
In summary, the invention has the advantages that:
1. the lute music score recognition method includes the steps that a plurality of lute music score sub-pictures are obtained by cutting each lute music score picture according to lines, a corresponding music XML file is cut based on the lute music score sub-pictures and converted into an MEI file, the MEI file is converted into a semantic file comprising note information, fingering information and rhythm information in a semantic mode, a created lute music score recognition model is trained based on the lute music score sub-pictures and the semantic file, and finally automatic recognition of the lute music score can be achieved by the aid of the trained lute music score recognition model, so that automatic recognition of the lute music score is achieved, and compared with traditional manual recording, the digitization efficiency of the lute music score is greatly improved.
2. By cutting the lute music score picture according to lines and carrying out the preprocessing of graying, binaryzation, noise reduction and inclination correction, the method greatly improves the accuracy of identification compared with the method for directly identifying the whole lute music score picture.
Although specific embodiments of the invention have been described above, it will be understood by those skilled in the art that the specific embodiments described are illustrative only and are not limiting upon the scope of the invention, and that equivalent modifications and variations can be made by those skilled in the art without departing from the spirit of the invention, which is to be limited only by the appended claims.
Claims (10)
1. A lute music score recognition method based on an end-to-end neural network is characterized by comprising the following steps: the method comprises the following steps:
s10, acquiring a large number of lute music score pictures and music XML files corresponding to the lute music score pictures, and preprocessing the lute music score pictures;
step S20, cutting each lute music score picture according to lines to obtain a plurality of lute music score sub-pictures, and cutting a corresponding music XML file based on the lute music score sub-pictures to obtain a plurality of music XML sub-files;
step S30, converting each MusicXML subfile into an MEI file, and converting each MEI file into a semantic file based on a preset semantic dictionary;
step S40, establishing a lute music score recognition model based on an end-to-end neural network, and training the lute music score recognition model by utilizing sub-pictures and semantic files of the lute music score;
and step S50, automatically recognizing the lute music score by using the trained lute music score recognition model.
2. The lute music score recognition method based on the end-to-end neural network as claimed in claim 1, wherein: in the step S10, the preprocessing of the lute music score pictures specifically includes:
and sequentially carrying out the preprocessing of graying, binaryzation, noise reduction and inclination correction on each lute music score picture.
3. The lute music score recognition method based on the end-to-end neural network as claimed in claim 1, wherein: the step S20 specifically includes:
identifying the upper spectral line of each line of music score in each lute music score picture based on a template matching method, and further cutting each lute music score picture according to lines based on the upper spectral line to obtain a plurality of lute music score sub-pictures;
identifying vertical spectral lines in each lute music score sub-picture, judging the number of subsections contained in the lute music score sub-picture by using the vertical spectral lines, and cutting the musicXML file based on the number of subsections and a label carried by the musicXML file to obtain a plurality of musicXML sub-files.
4. The lute music score recognition method based on the end-to-end neural network as claimed in claim 1, wherein: the step S30 specifically includes:
converting each MusicXML subfile into an MEI file through an OMR tool website, and converting each MEI file into a semantic file based on a preset semantic dictionary;
the semantic file is a note information sequence comprising note information, fingering information and rhythm information.
5. The lute music score recognition method based on the end-to-end neural network as claimed in claim 1, wherein: the step S40 specifically includes:
establishing a lute music score recognition model based on an end-to-end convolution recurrent neural network, taking each lute music score sub-picture as the input of the lute music score recognition model, taking a semantic file corresponding to each lute music score sub-picture as the output of the lute music score recognition model, and training the lute music score recognition model;
and a CTC function is adopted as a loss function of the lute music score identification model.
6. A lute music score recognition system based on an end-to-end neural network is characterized in that: the system comprises the following modules:
the lute music score image and music XML file acquisition module is used for acquiring a large number of lute music score images and music XML files corresponding to the lute music score images and preprocessing the lute music score images;
the lute music score picture and music XML file cutting module is used for cutting each lute music score picture according to lines to obtain a plurality of lute music score sub-pictures, and cutting a corresponding music XML file based on the lute music score sub-pictures to obtain a plurality of music XML sub-files;
the semantic conversion module is used for converting each MusicXML subfile into an MEI file and converting each MEI file into a semantic file based on a preset semantic dictionary;
the lute music score recognition model training module is used for establishing a lute music score recognition model based on an end-to-end neural network and training the lute music score recognition model by utilizing sub-pictures and semantic files of the lute music score;
and the lute music score automatic identification model is used for carrying out lute music score automatic identification by utilizing the trained lute music score identification model.
7. The lute music score recognition system based on end-to-end neural network as claimed in claim 6, wherein: in the lute music score picture and music XML file obtaining module, the preprocessing of the lute music score pictures specifically comprises the following steps:
and sequentially carrying out the preprocessing of graying, binaryzation, noise reduction and inclination correction on each lute music score picture.
8. The lute music score recognition system based on end-to-end neural network as claimed in claim 6, wherein: the lute music score picture and music XML file cutting module specifically comprises:
identifying the upper spectral line of each line of music score in each lute music score picture based on a template matching method, and further cutting each lute music score picture according to lines based on the upper spectral line to obtain a plurality of lute music score sub-pictures;
identifying vertical spectral lines in each lute music score sub-picture, judging the number of subsections contained in the lute music score sub-picture by using the vertical spectral lines, and cutting the musicXML file based on the number of subsections and a label carried by the musicXML file to obtain a plurality of musicXML sub-files.
9. The lute music score recognition system based on end-to-end neural network as claimed in claim 6, wherein: the semantic conversion module is specifically as follows:
converting each MusicXML subfile into an MEI file through an OMR tool website, and converting each MEI file into a semantic file based on a preset semantic dictionary;
the semantic file is a note information sequence comprising note information, fingering information and rhythm information.
10. The lute music score recognition system based on end-to-end neural network as claimed in claim 6, wherein: the lute music score recognition model training module specifically comprises:
establishing a lute music score recognition model based on an end-to-end convolution recurrent neural network, taking each lute music score sub-picture as the input of the lute music score recognition model, taking a semantic file corresponding to each lute music score sub-picture as the output of the lute music score recognition model, and training the lute music score recognition model;
and a CTC function is adopted as a loss function of the lute music score identification model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111458277.5A CN114332903A (en) | 2021-12-02 | 2021-12-02 | Lute music score identification method and system based on end-to-end neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111458277.5A CN114332903A (en) | 2021-12-02 | 2021-12-02 | Lute music score identification method and system based on end-to-end neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114332903A true CN114332903A (en) | 2022-04-12 |
Family
ID=81047820
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111458277.5A Pending CN114332903A (en) | 2021-12-02 | 2021-12-02 | Lute music score identification method and system based on end-to-end neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114332903A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115146649A (en) * | 2022-06-24 | 2022-10-04 | 厦门大学 | Method and device for identifying music book on drum |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110443127A (en) * | 2019-06-28 | 2019-11-12 | 天津大学 | In conjunction with the musical score image recognition methods of residual error convolutional coding structure and Recognition with Recurrent Neural Network |
| CN110852375A (en) * | 2019-11-09 | 2020-02-28 | 北京工业大学 | End-to-end music score note identification method based on deep learning |
| WO2020215683A1 (en) * | 2019-04-26 | 2020-10-29 | 平安科技（深圳）有限公司 | Semantic recognition method and apparatus based on convolutional neural network, and non-volatile readable storage medium and computer device |
| KR102306765B1 (en) * | 2021-02-17 | 2021-09-30 | 쿤크리에이터스 주식회사 | Sheet Music recognition method using artificial intelligence processing |
-
2021
- 2021-12-02 CN CN202111458277.5A patent/CN114332903A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020215683A1 (en) * | 2019-04-26 | 2020-10-29 | 平安科技（深圳）有限公司 | Semantic recognition method and apparatus based on convolutional neural network, and non-volatile readable storage medium and computer device |
| CN110443127A (en) * | 2019-06-28 | 2019-11-12 | 天津大学 | In conjunction with the musical score image recognition methods of residual error convolutional coding structure and Recognition with Recurrent Neural Network |
| CN110852375A (en) * | 2019-11-09 | 2020-02-28 | 北京工业大学 | End-to-end music score note identification method based on deep learning |
| KR102306765B1 (en) * | 2021-02-17 | 2021-09-30 | 쿤크리에이터스 주식회사 | Sheet Music recognition method using artificial intelligence processing |
Non-Patent Citations (1)
| Title |
|---|
| 黄志清;贾翔;郭一帆;张菁;: "基于深度学习的端到端乐谱音符识别", 天津大学学报(自然科学与工程技术版), no. 06, 27 April 2020 (2020-04-27), pages 107 - 114 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115146649A (en) * | 2022-06-24 | 2022-10-04 | 厦门大学 | Method and device for identifying music book on drum |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1218274C (en) | On-line handwrited script mode identifying editing device and method | |
| US10592704B2 (en) | System and method for electronic automated printed circuit design | |
| CN1834955A (en) | Multilingual translation memory, translation method, and translation program | |
| CN114005123A (en) | System and method for digitally reconstructing layout of print form text | |
| CN111209728B (en) | Automatic labeling and inputting method for test questions | |
| CN110781648A (en) | Test paper automatic transcription system and method based on deep learning | |
| CN109830230A (en) | A kind of data mask method and device based on self-teaching algorithm | |
| CN113129898A (en) | Machine-assisted conference recording system and method | |
| CN114332903A (en) | Lute music score identification method and system based on end-to-end neural network | |
| CN116052193A (en) | RPA interface dynamic form picking and matching method and system | |
| CN110674802B (en) | Improved text detection method for parallelogram candidate box | |
| CN111860487A (en) | Inscription marking detection and recognition system based on deep neural network | |
| CN113283231A (en) | Method for acquiring signature bit, setting system, signature system and storage medium | |
| CN116341525A (en) | Text examination and correction system based on natural language processing | |
| JP5227120B2 (en) | Character string recognition apparatus and method, and program | |
| CN115988149A (en) | Method for generating video by AI intelligent graphics context | |
| CN116166845A (en) | Universal analysis system and method for radar admission data | |
| CN112052652B (en) | Automatic generation method and device for electronic courseware script | |
| CN112102670B (en) | Teaching auxiliary system based on voice control and working method thereof | |
| CN115393875A (en) | Method and system for staff identification and numbered musical notation conversion based on MobileNet V3 | |
| CN111209924B (en) | System for automatically extracting medical advice and application | |
| CN110866400B (en) | Automatic change lexical analysis system of update | |
| CN113449504A (en) | Intelligent marking method and system | |
| CN113205527A (en) | Intelligent test paper cutting method and system and storage medium | |
| CN112201225A (en) | Corpus obtaining method and device, readable storage medium and electronic equipment |
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 |