Stable Audio Open 1.0

Stable Audio Open 1.0 is Stability AI's open-source text-to-audio model. It generates variable-length stereo audio at 44.1kHz from text descriptions - from drum loops to sound effects. Based on a latent diffusion transformer (DiT) architecture.

Capabilities

• Variable length - up to 47 seconds
• 44.1kHz stereo - CD-quality output
• Text conditioning - T5-based text embeddings
• Sound effects - excels at SFX and field recordings
• Music loops - drum patterns, instrumentals

Usage

from diffusers import StableAudioPipeline
import torch
import soundfile as sf
 
pipe = StableAudioPipeline.from_pretrained(
    "stabilityai/stable-audio-open-1.0",
    torch_dtype=torch.float16
).to("cuda")
 
audio = pipe(
    "128 BPM tech house drum loop",
    num_inference_steps=200,
    audio_end_in_s=10.0,
).audios
 
sf.write("output.wav", audio[0].T.float().cpu().numpy(), pipe.vae.sampling_rate)

When to use

• Music production - generate drum loops and instrumentals
• Sound design - SFX for games and video
• Content creation - background music for social media
• Audio prototyping - quick audio sketches

Limitations

• No realistic vocals
• English prompts only
• Better at SFX than music
• Requires prompt engineering