@inproceedings{jiang-etal-2026-muse,
    title = "Muse: Towards Reproducible Long-Form Song Generation with Fine-Grained Style Control",
    author = "Jiang, Changhao  and
      Chen, Jiahao  and
      Xiang, Zhenghao  and
      Yang, Zhixiong  and
      Wang, Hanchen  and
      Zhuang, Jiabao  and
      Che, Xinmeng  and
      Sun, Jiajun  and
      Li, Hui  and
      Cao, Yifei  and
      Dou, Shihan  and
      Zhang, Ming  and
      Ye, Junjie  and
      Ji, Tao  and
      Gui, Tao  and
      Zhang, Qi  and
      Huang, Xuanjing",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl/2026.findings-acl.1129/",
    pages = "22492--22512",
    ISBN = "979-8-89176-395-1",
    abstract = "Recent commercial systems such as Suno demonstrate strong capabilities in long-form song generation, while academic research remains largely non-reproducible due to the lack of publicly available training data, hindering fair comparison and progress. To this end, we release a fully open-source system for long-form song generation with fine-grained style conditioning, including a licensed synthetic dataset, training and evaluation pipelines, and Muse, an easy-to-deploy song generation model. The dataset consists of 116k fully licensed synthetic songs with automatically generated lyrics and style descriptions paired with audio synthesized by SunoV5. We train Muse via single-stage supervised finetuning of a Qwen-based language model extended with discrete audio tokens using MuCodec, without task-specific losses, auxiliary objectives, or additional architectural components. Our evaluations find that although Muse is trained with a modest data scale and model size, it achieves competitive performance on phoneme error rate, text{--}music style similarity, and audio aesthetic quality, while enabling controllable segment-level generation across different musical structures. All data, model weights, and training and evaluation pipelines will be publicly released, paving the way for continued progress in controllable long-form song generation research."
}