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.

Stable Audio Open

Stability AI · 2024

Synthesizes variable-length stereo audio, sound effects, and musical stems from text prompts.

Application types: Text-to-musicText-to-sample

Architecture:Diffusion

Essential categories

Category	Status	Notes
Source code	✔︎ Open	Code for data processing, training pipeline, and inference is available in the stable-audio-tools repository. Architecture of the model is specified in the form of a config file.
Training data	✔︎ Open	Training data can be mapped from Attribution files. Dataset attribution files can also be accessed in HuggingFace after accepting Stability AI conditions (https://huggingface.co/stabilityai/stable-audio-open-1.0/blob/main/fma_dataset_attribution2.csv and https://huggingface.co/stabilityai/stable-audio-open-1.0/blob/main/freesound_dataset_attribution2.csv).
Model weights	✔︎ Open	Weights are available in the HuggingFace repository. However, accessibility is subject to accepting Stability AI conditions.
Code documentation	✔︎ Open	Documentation of the code is limited for training replicating the model. Although there are not specific config files for Stable Audio Open, those from Stable Audio could be used.
Training procedure	✔︎ Open	Described in pre-print.
Evaluation procedure	✔︎ Open	They use stable-audio-metrics for evaluation, which has examples of evaluation pipelines.
Research paper	✔︎ Open	Accepted at ICASSP 2026.
Licensing	~ Partial	Stability AI Community License. Not an Open Source Initiate (OSI) or RAIL approved licences.

Desirable categories

Category	Status	Notes
Model card	⭐ Included	Model card in HuggingFace.
Datasheet	∅ Not included
Package	⭐ Included	The model can be used with 1) the https://github.com/Stability-AI/stable-audio-tools library and 2) the https://huggingface.co/docs/diffusers/main/en/index library.
User-oriented application	⭐ Included	Gradio interface in source code.
Supplementary material page	⭐ Included	A complementary site with sound examples and demonstration of the model's capabilities is available.

Raw YAML file with complete evaluation.