{"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/pop-music-highlighter-marking-the-emotion","title":"Pop Music Highlighter: Marking the Emotion Keypoints","arxiv_id":"1802.10495","date":"2018-02-28","proceeding":null,"authors":["Yu-Siang Huang","Szu-Yu Chou","Yi-Hsuan Yang"],"abstract":"The goal of music highlight extraction is to get a short consecutive segment\nof a piece of music that provides an effective representation of the whole\npiece. In a previous work, we introduced an attention-based convolutional\nrecurrent neural network that uses music emotion classification as a surrogate\ntask for music highlight extraction, for Pop songs. The rationale behind that\napproach is that the highlight of a song is usually the most emotional part.\nThis paper extends our previous work in the following two aspects. First,\nmethodology-wise we experiment with a new architecture that does not need any\nrecurrent layers, making the training process faster. Moreover, we compare a\nlate-fusion variant and an early-fusion variant to study which one better\nexploits the attention mechanism. Second, we conduct and report an extensive\nset of experiments comparing the proposed attention-based methods against a\nheuristic energy-based method, a structural repetition-based method, and a few\nother simple feature-based methods for this task. Due to the lack of\npublic-domain labeled data for highlight extraction, following our previous\nwork we use the RWC POP 100-song data set to evaluate how the detected\nhighlights overlap with any chorus sections of the songs. The experiments\ndemonstrate the effectiveness of our methods over competing methods. For\nreproducibility, we open source the code and pre-trained model at\nhttps://github.com/remyhuang/pop-music-highlighter/.","url_abs":"http://arxiv.org/abs/1802.10495v2","url_pdf":"http://arxiv.org/pdf/1802.10495v2.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":"pop-music-highlighter-marking-the-emotion","repo_url":"https://github.com/remyhuang/pop-music-highlighter","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"GPL-3.0"}}],"tasks":[{"task_slug":"emotion-classification","task_name":"Emotion Classification"}],"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}