Automatic Analysis and Segmentation Annotation of Teaching Repertoire Structure by Beat Feature Fusion

Automatic Analysis and Segmentation Annotation of Teaching Repertoire Structure by Beat Feature Fusion | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Automatic Analysis and Segmentation Annotation of Teaching Repertoire Structure by Beat Feature Fusion Xue Li, Shengqiang Zhang This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9432354/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 5 You are reading this latest preprint version Abstract In digital music education, manually annotating the structure of teaching repertoires often suffers from imprecise boundaries, perceptual delays (typically 0.3–0.6 s), and low efficiency for long pieces. To address these challenges, this study proposes a deep learning model for automatic structural analysis based on beat feature fusion, named MERT–Conformer Hierarchical Beat-Feature Fusion Network (MCH-BFNet). The model treats beat as fundamental units and integrates MERT pre-trained semantic representations with multidimensional acoustic features. This dual “feature–structure” optimization improves parsing accuracy. The Conformer backbone extracts local time-frequency details and captures global long-range dependencies. A hierarchical attention mechanism (HAM) enables structured mapping from motives to larger sections. Experiments on the SALAMI public dataset ( https://github.com/DDMAL/salami-data-public ) show that MCH-BFNet achieves an F1 score of 0.85 for boundary detection with a 3-second window and approximately 0.65 with a 0.5-second window. This represents a 7% improvement over the baseline model (ResNet + Self-Attention). Compared with manual annotation, the model provides millisecond-level boundary regression through beat-synchronous alignment, eliminating the 0.3–0.6 s perceptual delay in human labeling. Segment annotation accuracy, measured by the Pairwise F-measure (PWF), consistently exceeds 0.8. Overall, this study offers a reliable technical reference for automated structural navigation and high-precision segmentation in intelligent teaching systems, addressing the inefficiency and errors of manual annotation. Biological sciences/Computational biology and bioinformatics Physical sciences/Mathematics and computing teaching repertoire automatic structural analysis MCH-BFNet beat feature fusion Conformer Full Text Additional Declarations No competing interests reported. Supplementary Files Data.rar Cite Share Download PDF Status: Under Review Version 1 posted Reviewers invited by journal 24 Apr, 2026 Editor invited by journal 23 Apr, 2026 Editor assigned by journal 17 Apr, 2026 Submission checks completed at journal 17 Apr, 2026 First submitted to journal 15 Apr, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9432354","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":634177356,"identity":"ca92dc40-ea22-44ff-85c0-85411427e037","order_by":0,"name":"Xue Li","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAw0lEQVRIiWNgGAWjYBACPmYQWQHl8RCjhQ2s5QxJWkAEYxtJWth5zCQ+zrOT3XD+AOODt20M8uaEHcaWJjlzW7LxhgMHmA3ntjEY7mwgqIX5mDTvtgOJGw42sEnztjEkGBwgqIWxTfrvHKCWwwzsv4nUArSFsQGo5RiQTaQWtmTLnmPJxjPPMDZLzjknYbiBkBZ+/jOGN37U2Mn2nT988MObMht5grbAAOOCA4wNQFqCSPUgLfMbiFc8CkbBKBgFIwwAAFoIOjQrXRcDAAAAAElFTkSuQmCC","orcid":"","institution":"Chongqing Preschool Education College","correspondingAuthor":true,"prefix":"","firstName":"Xue","middleName":"","lastName":"Li","suffix":""},{"id":634177357,"identity":"e8a76be3-1d4c-4b6e-a419-817c78761e65","order_by":1,"name":"Shengqiang Zhang","email":"","orcid":"","institution":"Chongqing Preschool Education College","correspondingAuthor":false,"prefix":"","firstName":"Shengqiang","middleName":"","lastName":"Zhang","suffix":""}],"badges":[],"createdAt":"2026-04-16 02:55:38","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9432354/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9432354/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108468129,"identity":"bb18ea47-0ba0-41b2-953b-fa62587d3977","added_by":"auto","created_at":"2026-05-05 04:25:27","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":779168,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9432354/v1_covered_bb8de1f8-bcdf-4e8b-8c35-e7177e9c0137.pdf"},{"id":108468117,"identity":"4338bb20-235a-4a09-94e6-7bf0f29d5786","added_by":"auto","created_at":"2026-05-05 04:25:19","extension":"rar","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":2354239,"visible":true,"origin":"","legend":"","description":"","filename":"Data.rar","url":"https://assets-eu.researchsquare.com/files/rs-9432354/v1/b04593ea203661c353c5e129.rar"}],"financialInterests":"No competing interests reported.","formattedTitle":"Automatic Analysis and Segmentation Annotation of Teaching Repertoire Structure by Beat Feature Fusion","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"teaching repertoire, automatic structural analysis, MCH-BFNet, beat feature fusion, Conformer","lastPublishedDoi":"10.21203/rs.3.rs-9432354/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9432354/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"In digital music education, manually annotating the structure of teaching repertoires often suffers from imprecise boundaries, perceptual delays (typically 0.3–0.6 s), and low efficiency for long pieces. To address these challenges, this study proposes a deep learning model for automatic structural analysis based on beat feature fusion, named MERT–Conformer Hierarchical Beat-Feature Fusion Network (MCH-BFNet). The model treats beat as fundamental units and integrates MERT pre-trained semantic representations with multidimensional acoustic features. This dual “feature–structure” optimization improves parsing accuracy. The Conformer backbone extracts local time-frequency details and captures global long-range dependencies. A hierarchical attention mechanism (HAM) enables structured mapping from motives to larger sections. Experiments on the SALAMI public dataset (https://github.com/DDMAL/salami-data-public) show that MCH-BFNet achieves an F1 score of 0.85 for boundary detection with a 3-second window and approximately 0.65 with a 0.5-second window. This represents a 7% improvement over the baseline model (ResNet + Self-Attention). Compared with manual annotation, the model provides millisecond-level boundary regression through beat-synchronous alignment, eliminating the 0.3–0.6 s perceptual delay in human labeling. Segment annotation accuracy, measured by the Pairwise F-measure (PWF), consistently exceeds 0.8. Overall, this study offers a reliable technical reference for automated structural navigation and high-precision segmentation in intelligent teaching systems, addressing the inefficiency and errors of manual annotation.","manuscriptTitle":"Automatic Analysis and Segmentation Annotation of Teaching Repertoire Structure by Beat Feature Fusion","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-05 04:24:14","doi":"10.21203/rs.3.rs-9432354/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewersInvited","content":"","date":"2026-04-24T09:10:03+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-04-23T06:06:53+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-17T04:29:58+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-04-17T04:29:38+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2026-04-16T02:45:19+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"9daf270e-1e90-4de1-ac6f-5b9c7e46c046","owner":[],"postedDate":"May 5th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":67489382,"name":"Biological sciences/Computational biology and bioinformatics"},{"id":67489383,"name":"Physical sciences/Mathematics and computing"}],"tags":[],"updatedAt":"2026-05-05T04:24:14+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-05 04:24:14","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9432354","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9432354","identity":"rs-9432354","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}