{"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/deep-convolutional-neural-networks-for-6","title":"Deep convolutional neural networks for predominant instrument recognition in polyphonic music","arxiv_id":"1605.09507","date":"2016-05-31","proceeding":null,"authors":["Yoonchang Han","Jaehun Kim","Kyogu Lee"],"abstract":"Identifying musical instruments in polyphonic music recordings is a\nchallenging but important problem in the field of music information retrieval.\nIt enables music search by instrument, helps recognize musical genres, or can\nmake music transcription easier and more accurate. In this paper, we present a\nconvolutional neural network framework for predominant instrument recognition\nin real-world polyphonic music. We train our network from fixed-length music\nexcerpts with a single-labeled predominant instrument and estimate an arbitrary\nnumber of predominant instruments from an audio signal with a variable length.\nTo obtain the audio-excerpt-wise result, we aggregate multiple outputs from\nsliding windows over the test audio. In doing so, we investigated two different\naggregation methods: one takes the average for each instrument and the other\ntakes the instrument-wise sum followed by normalization. In addition, we\nconducted extensive experiments on several important factors that affect the\nperformance, including analysis window size, identification threshold, and\nactivation functions for neural networks to find the optimal set of parameters.\nUsing a dataset of 10k audio excerpts from 11 instruments for evaluation, we\nfound that convolutional neural networks are more robust than conventional\nmethods that exploit spectral features and source separation with support\nvector machines. Experimental results showed that the proposed convolutional\nnetwork architecture obtained an F1 measure of 0.602 for micro and 0.503 for\nmacro, respectively, achieving 19.6% and 16.4% in performance improvement\ncompared with other state-of-the-art algorithms.","url_abs":"http://arxiv.org/abs/1605.09507v3","url_pdf":"http://arxiv.org/pdf/1605.09507v3.pdf","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":[{"paper_slug":"deep-convolutional-neural-networks-for-6","repo_url":"https://github.com/iooops/CS221-Audio-Tagging","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"information-retrieval","task_name":"Information Retrieval"},{"task_slug":"instrument-recognition","task_name":"Instrument Recognition"},{"task_slug":"music-information-retrieval","task_name":"Music Information Retrieval"},{"task_slug":"music-transcription","task_name":"Music Transcription"},{"task_slug":"retrieval","task_name":"Retrieval"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}