[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-661196-105":59,"doc-detail-661196-en":130},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":123,"head_meta":125,"extra_data":127,"updated_unix":129},105,"en","yue-scaling-open-foundation-models-for-long-form-music-generation","YUE - Scaling Open Foundation Models for Long-Form Music Generation","","YuE (乐) is an open-source family of foundation models for long-form music generation, focused on the difficult lyrics-to-song setting. YuE scales to trillions of tokens and generates up to five minutes of music while preserving lyrical alignment, coherent musical structure, and engaging vocal melodies with appropriate accompaniment. Track-decoupled next-token prediction and structural progressive conditioning enable robust long-context control. A redesigned in-context learning method supports bidirectional creation and style cloning. Extensive evaluation shows YuE matches or exceeds certain proprietary systems in musicality and vocal agility.",{"@graph":69,"@context":122},[70,84,105],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/yue-scaling-open-foundation-models-for-long-form-music-generation/661196/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/yue-scaling-open-foundation-models-for-long-form-music-generation/661196.png","ImageObject",300,407,{"name":92,"@type":93},"Himbo","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-30","2026-10-28",true,{"@type":102,"interactionType":103,"userInteractionCount":24},"InteractionCounter",{"@type":104},"ViewAction",{"@type":106,"mainEntity":107},"FAQPage",[108,114,118],{"name":109,"@type":110,"acceptedAnswer":111},"What task does YuE focus on?","Question",{"text":112,"@type":113},"YuE targets long-form music generation, especially the lyrics-to-song problem, generating full songs with vocals and accompaniment from lyrics.","Answer",{"name":115,"@type":110,"acceptedAnswer":116},"How does YuE maintain long-context lyrical alignment and structure?",{"text":117,"@type":113},"It uses structural progressive conditioning designed for long-form generation, enabling song-level lyrics following and structure control.",{"name":119,"@type":110,"acceptedAnswer":120},"What technique does YuE use to handle dense mixture signals in audio tokens?",{"text":121,"@type":113},"YuE introduces track-decoupled next-token prediction that separately models different audio tracks at the frame level.","https://schema.org",{"og:url":83,"og:type":124,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":126,"canonical":83},"index,follow",{"doc_id":128,"site_id":62},661196,1793159566,{"code":4,"msg":5,"data":131},{"doc_id":128,"user_id":132,"nickname":92,"user_avatar":133,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":24,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":139,"language":140,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":67,"update_tm":129,"read_time":144},687197100911,"https://ap-avatar.wpscdn.com/avatar/a000239b6f1da00475?x-image-process=image/resize,m_fixed,w_180,h_180&k=1785132997149421697","YUE: SCALING OPEN FOUNDATION MODELS FOR LONG-FORM MUSIC GENERATION  \n Multimodal Art Projection∗, Hong Kong University of Science and Technology∗  \nABSTRACT  \nWe tackle the task of long-form music generation, particularly the challenging lyrics-to-song problem, by introducing YuE (乐), a family of open-source music generation foundation models. Specifically, YuE scales to trillions of tokens and generates up to five minutes of music while maintaining lyrical alignment, coherent musical structure, and engaging vocal melodies with appropriate accompaniment.  \nIt achieves this through track-decoupled next-token prediction to overcome dense mixture signals, and structural progressive conditioning for long-context lyrical alignment. In addition, we redesign the in-context learning technique for music generation, enabling bidirectional content creation, style cloning, and improving musicality. Through extensive evaluation, we demonstrate that YuE matches or even surpasses some of the proprietary systems in musicality and vocal agility (as of 2025-01) . We strongly encourage readers to listen to our demo1.  \n1 INTRODUCTION  \nNeural music generation is transforming artistic creativity and holds significant commercial and cultural potential (Ma et al., 2024) . Among music generation tasks, lyrics-to-song, creating full songs with vocals and accompaniment from lyrics, remains highly challenging. By the release time of our system, no open-source system can reliably achieve this at scale, restricting innovation and accessibility despite proprietary solutions like Suno and Udio2 demonstrating promising results.  \nKey challenges of lyrics-to-song generation include: 1) Long-range dependencies: Music exhibits complex temporal structures spanning several minutes, making it difficult for models to maintain coherence over extended durations. 2) Signal complexity: Unlike speech or environmental sounds, music is inherently polyphonic, requiring precise coordination between multiple instrumental and vocal components. 3) Linguistic distortion: Singing alters phonemes, durations, and prosody in ways that differ significantly from spoken language, complicating the alignment between lyrics and melody. 4) Data scarcity: The lack of large-scale, high-quality paired datasets of lyrics, vocals, and accompaniment limits model training and generalization capabilities.  \nRelated Work. Existing models for music generation and singing voice synthesis (SVS) are often limited in several key aspects. For instance, models such as Jukebox (Dhariwal et al., 2020) and MusicLM (Agostinelli et al., 2023) typically generate instrumental music with short durations (around 30 seconds) and lack coherent lyrical semantics when attempting to incorporate vocals. While some SVS systems (Chen et al., 2020; Liu et al., 2022) have been successful in generating vocals, they often rely on explicit melodic guidance and do not seamlessly integrate vocals with instrumental accompaniments, especially over long-form durations. Moreover, academic models such as MelodyLM (Li et al., 2024) and SongCreator (Lei et al., 2024) struggle to maintain high-quality, long-duration synthesis beyond 30 seconds, often suffering from artifacts or limited controllability. Industry-developed systems, such as Tiangong Music and Suno, offer better performance but remain proprietary, with limited technical transparency.  \n∗ Co-first affiliation. These institutions contributed equally. 1 [https://map-yue.github.io/](https://map-yue.github.io/)  \n2 [https://suno.com/](https://suno.com/), [https://www.udio.com/](https://www.udio.com/)  \nIn this paper, we introduce YuE, the first3 open-source foundation model family for high-quality, full-song lyrics-to-song generation. Built upon LLaMA2 architechture (Touvron et al., 2023b; Zhang et al., 2024) and trained on trillions of tokens, YuE generates coherent, lyrically-aligned, and musically interesting songs up to five minutes in length, matching or surpassing propri","cbCaipekEwS93r9A","https://ap.wps.com/l/cbCaipekEwS93r9A","pdf",2272449,42,"English","# Introduction\n# YUE\n## Overview\n## Track-Decoupled Next-Token Prediction","[{\"question\":\"What task does YuE focus on?\",\"answer\":\"YuE targets long-form music generation, especially the lyrics-to-song problem, generating full songs with vocals and accompaniment from lyrics.\"},{\"question\":\"How does YuE maintain long-context lyrical alignment and structure?\",\"answer\":\"It uses structural progressive conditioning designed for long-form generation, enabling song-level lyrics following and structure control.\"},{\"question\":\"What technique does YuE use to handle dense mixture signals in audio tokens?\",\"answer\":\"YuE introduces track-decoupled next-token prediction that separately models different audio tracks at the frame level.\"}]","YUE - Scaling Open Foundation Models for Long-Form Music Generation | PDF",106]