Stable Audio 3 · Small Music / SFX
Choose a music or sound-effects specialist for local generation with a smaller runtime than Medium.
GenerationOpen-source code433M
- Maru integration
Maru workflow available
Generate loops with Small Music; one-shots and effects use Small SFX. Sign-in and service availability apply; completed sounds enter the public collection.
- Start here
- Compare the Music and SFX variants on the same short prompt before paying the memory cost of a larger model.
- Training data
- The cards report 806,284 AudioSparx recordings and 472,618 Freesound recordings; T5Gemma supplies text conditioning.
- Training rights
- Licensed AudioSparx audio and CC0, CC BY and CC Sampling+ Freesound sources are reported. Gemma component terms apply separately.
- License & output terms
- Stability AI Community License plus component terms.
- Limitations
- Music and SFX are different checkpoints. English descriptions, no designed speech generation; published speed measurements depend on hardware and duration.
- Technique
- Distilled latent diffusion / SAME-S
- Runtime
- 433M diffusion transformer with the smaller SAME-S autoencoder. The upstream project includes CPU, Apple Silicon and GPU paths; test your actual configuration.
- Parameters
- 433M
Small Music card (opens in a new tab)Small SFX card (opens in a new tab)Variant & LoRA guide (opens in a new tab)
Sources reviewed . Training and rights notes summarize provider disclosures. A model license does not clear every source recording or generated output. Parameter counts refer to the named variant or component, not total runtime memory.