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.
Generative model that creates new audio by blending two sources: one audio stream to set the style or timbre, and another input (either audio or MIDI) to shape the structure over time.
Application types:MIDI-to-audioAudio synthesisStyle transfer
Datasets used for training the original model (SLAKH, MaestroV2, GuitarSet and URMP) are publicly available and described in the paper, but the datasets used for training the current model are unknown.
Documentation is available in the GitHub repository, including installation requirements, examples of scripts usage and configuration files. The codebase could still benefit from more detailed documentation.
Training procedure of the original model is fully described in the paper, but relevant changes in the model architecture and training procedure are not properly documented.
Evaluation procedure of the original model is fully described in the paper, including used metrics and implementation details. However, no evaluation is provided for the current model.
