iex0 · Stable Audio Space

Text-to-audio (and audio-to-audio where supported) for the Stable Audio family. Designed to be called via gradio_client — see the README for the API contract.

Model

Stable Audio Open 1.0 (1B · ≤47 s) — Stereo 44.1 kHz · best general SFX + ambience baseline · gated repo

Prompt

Negative prompt

Duration (s)

1 47

Steps

4 300

CFG scale

0 15

Sampler

sigma_min

0 5

sigma_max

10 1000

Seed (-1 = random)

Init audio (optional, audio-to-audio)

Init noise level

0 1

Output

Seed used

API

From Python:

from gradio_client import Client, handle_file
client = Client('00cyre/iex0-audio-space', hf_token='hf_…')
wav, info, seed = client.predict(
    model_key='stable-audio-open-1.0',
    prompt='deep zombie growl, wet, distant',
    negative_prompt='', duration_s=8.0, steps=100, cfg_scale=7.0,
    sigma_min=0.3, sigma_max=500.0, sampler_type='dpmpp-3m-sde',
    seed=-1, init_audio=None, init_strength=0.5,
    api_name='/generate',
)