Stable Audio 3 Small-Music — GGUF (for sa3.cpp)
GGUF conversions of
stabilityai/stable-audio-3-small-music
for
sa3.cpp — a portable C++/GGML port of
Stable Audio 3, no PyTorch in the loop. Runs on CPU, CUDA, Vulkan, or Metal (Apple Silicon). The small-music model is a
lighter, faster music generator (SAME-S autoencoder, 0.5B DiT). Validated against the PyTorch
reference at cosine similarity ~1.0.
Files
Multi-file model. Grab the
DiT +
SAME at your chosen precision and the
conditioner, plus
the shared
encoder + tokenizer from
t5gemma-b-b-ul2-GGUF.
| component | file | notes |
|---|
| DiT (diffusion transformer) | stable-audio-3-small-music-dit-0.5B-v1.0-{F32,F16,Q8_0,Q5_K_M,Q4_K_M}.gguf | pick one encoding |
| autoencoder (SAME-S) | stable-audio-3-small-music-same-s-v1.0-{F32,F16,Q8_0,Q5_K_M,Q4_K_M}.gguf | must match the DiT |
| conditioner | stable-audio-3-small-music-conditioner-v1.0-F32.gguf | tiny sidecar (prompt padding + seconds_total) |
| encoder + tokenizer | → t5gemma-b-b-ul2-GGUF | shared across all SA3 variants |
note: SAME-S needs an even --frames count (the packed sequence must divide the chunk size).
Encodings
sa3-generate --encoding resolves the DiT and the SAME with the same suffix, so download the pair.
| encoding | DiT | SAME-S | total |
|---|
| F32 | 1751 MB | 413 MB | 2164 MB |
| F16 | 877 MB | 207 MB | 1084 MB |
| Q8_0 | 481 MB | 110 MB | 591 MB |
| Q5_K_M | 343 MB | 79 MB | 422 MB |
| Q4_K_M | 312 MB | 72 MB | 384 MB |
q4_k_m and q5_k_m promote the attention V, feed-forward down and embedding tensors to Q6_K;
q8_0 is uniform. Every tier passes sa3-quant-check with below-threshold=0 at cosine 0.990
against the F16 reference, for the DiT and the SAME alike.
Quantization buys footprint everywhere and speed only on some backends. CUDA and Vulkan gain
roughly 33% end to end. Metal is flat — Q8_0 is 1.7% faster and Q4_K_M 2.1% slower than F16,
because the load-time saving and the added per-step dequant cancel out. On a Mac, pick a quant for
the memory, not for the speed.
Usage
1python tools/download_models.py --variant small-music --encoding f16
2
3# --model resolves the gguf set in ./models by name
4sa3-generate --model small-music --prompt "lo-fi hip hop beat, warm vinyl, mellow keys" --out song.wav
For a quantized set, pass the encoding to both — the downloader and the generator use the same names:
1python tools/download_models.py --variant small-music --encoding q4_k_m
2sa3-generate --model small-music --encoding q4_k_m --prompt "lo-fi hip hop beat" --out song.wav
Performance
Roughly
1.7s for a 12s clip at f16 on an 8GB laptop GPU (RTX 5070) — about 2× faster than the medium
model. The sliding-window decoder keeps long generations linear. Full numbers + levers:
docs/BENCHMARKS.md.
License
These are format conversions of
stabilityai/stable-audio-3-small-music,
whose weights Stability AI releases under the
Stability AI Community License:
free for organizations under $1M annual revenue, with commercial use, fine-tuning, and derivative works
permitted within that threshold (above it, contact Stability AI for an Enterprise License). Outputs are yours.
That license carries over to these converted weights.
The upstream
stable-audio-3 source code is released
separately under MIT. Pair these with the shared T5Gemma text encoder, which is Google's under the
Gemma Terms of Use.
Relationship to the original
Format conversions (weights → GGUF) for inference in sa3.cpp — no retraining. See
sa3.cpp/docs/DISTRIBUTION.md.
