MusGO Framework: Assessing Openness in Music-Generative AI

This website builds on the paper MusGO: A Community-Driven Framework for Assessing Openness in Music-Generative AI, authored by Roser Batlle-Roca, Laura Ibáñez-Martínez, Xavier Serra, Emilia Gómez, and Martín Rocamora, and published in the Proceedings of the 26th International Society for Music Information Retrieval Conference (ISMIR 2025). It serves not only as a companion to the publication, but also as a living resource, which is continuously updated and shaped by contributions from the community.

YuE (乐)

The Hong Kong University of Science and Technology and Mutimodal Art Projection (M-A-P) · 2025

YuE is an autoregressive LM-based framework tailored for lyrics-to-song generation.

Application types: Lyrics-to-songStyle transferText-to-song

Architecture:Autoregressive transformer

Essential categories

Category	Status	Notes
Source code	~ Partial	Source code is available in the GitHub repo, including inference code and model architecture. They also provide instructions on how to fine-tune YuE for specific data (see https://github.com/multimodal-art-projection/YuE/tree/main/finetune), including data processing and training procedure. Code to train the model from scratch is missing.
Training data	✘ Closed	Speech-related datasets are introduced (i.e., WeNetSpeech (zh), LibriHeavy (en), and GigaSpeech (en)). But for music, they only indicate that they rely on 650K hours of music recordings. No information on specific data sources, acquisition methods, and data licensing is given. See "Data Setup" in Section 5.1. of the technical report.
Model weights	✔︎ Open	Checkpoints are available in Hugging Face.
Code documentation	✔︎ Open	Codebase is documented in the GitHub repository, including installation requirements and usage instructions.
Training procedure	✔︎ Open	Training procedure is properly described, including hardware requirements and configurations (see Section 4 and Section 5.1).
Evaluation procedure	✔︎ Open	Evaluation procedure is described in Section 5.2 of the technical report, including both objective and subjective evaluation methods. Evaluation metrics implementations are referenced in the text.
Research paper	~ Partial	Technical report available, but no peer-reviewed publication yet.
Licensing	✔︎ Open	Apache License 2.0

Desirable categories

Category	Status	Notes
Model card	∅ Not included
Datasheet	∅ Not included
Package	∅ Not included
User-oriented application	∅ Not included	No official user-oriented application is available. But there are user-oriented applications available for YuE, for both Windows and Linux. For Windows, check out: https://github.com/multimodal-art-projection/YuE?tab=readme-ov-file#-windows-users-quickstart, and for Linux: https://github.com/multimodal-art-projection/YuE?tab=readme-ov-file#-linuxwsl-users-quickstart.
Supplementary material page	⭐ Included	A demo page is available showcasing the use of YuE.

Raw YAML file with complete evaluation.