Description

Stable Audio 3.0 is a generative audio model family from Stability AI, trained on fully licensed data and designed to support the audio community’s next wave of creation.

The page presents Stable Audio as an open invitation to experiment, with open weights available for download and building on.

How we innovate

What stands out is the combination of fully licensed training data and open weights, making Stable Audio 3.0 both a foundation model and an open invitation to experiment.

Use Case / Scenario

1. Experiment with generative audio

Use Stable Audio 3.0 as a foundation for trying out audio generation ideas and exploring what can be built with open weights.

2. Build on licensed training data

Work with a model family trained on fully licensed data, making it positioned for responsible audio development.

3. Download open weights for development

Download the open weights and build directly on the model family for further experimentation and implementation.

Visit Website