{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/bytecover2-towards-dimensionality-reduction","title":"BYTECOVER2: TOWARDS DIMENSIONALITY REDUCTION OF LATENT EMBEDDING FOR EFFICIENT COVER SONG IDENTIFICATION","arxiv_id":null,"date":"2022-04-27","proceeding":"ICASSP 2022 4","authors":["Xingjian Du","Ke Chen","Zijie Wang","Bilei Zhu","Zejun Ma"],"abstract":"Convolutional neural network (CNN)-based methods have\r\ndominated the recent research of cover song identification\r\n(CSI). A typical example is the ByteCover system we proposed, which has achieved state-of-the-art results on all the\r\nmainstream datasets of CSI. In this paper, we propose an upgraded version of ByteCover, termed ByteCover2, which further improves ByteCover in both identification performance\r\nand efficiency. Compared with ByteCover, ByteCover2 is\r\ndesigned with an additional PCA-FC module, which integrates the capability of principal component analysis (PCA)\r\nand fully-connected (FC) neural network for dimensionality reduction of the audio embedding, allowing ByteCover2\r\nto perform CSI in a more precise and efficient way. We\r\nevaluated ByteCover2 on multiple datasets in different dimension sizes and training settings, where ByteCover2 beat\r\nall the compared methods including ByteCover, even with a\r\ndimension size of 128, which is 15 times smaller than that of\r\nByteCover.","url_abs":"https://ieeexplore.ieee.org/abstract/document/9747630","url_pdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9747630","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[],"tasks":[{"task_slug":"cover-song-identification","task_name":"Cover song identification"},{"task_slug":"dimensionality-reduction","task_name":"Dimensionality Reduction"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/cover-song-identification-on-covers80","task":"Cover song identification","dataset":"Covers80","model":"ByteCover2","rank_in_archive_order":1,"of":5,"metrics":{"MAP":"0.928"},"uses_additional_data":false},{"leaderboard":"/sota/cover-song-identification-on-da-tacos","task":"Cover song identification","dataset":"Da-TACOS","model":"ByteCover2","rank_in_archive_order":1,"of":4,"metrics":{"mAP":"0.791"},"uses_additional_data":false},{"leaderboard":"/sota/cover-song-identification-on-shs100k-test","task":"Cover song identification","dataset":"SHS100K-TEST","model":"Bytecover","rank_in_archive_order":2,"of":8,"metrics":{"mAP":"0.864"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}