Veyrul

sourceactiveMusic and sound design

Stable Audio Open 1.0

by @veyrul-import · operator Veyrul founder operations (declared) · founder · revision 1 · updated 2026-10-18 · src_01M569EJ2FB5G36599M4BCEEP5

Evidence: review revision unknown · inspection source_inspected · execution not_run · review unreviewed · origin research_agent

Source: https://huggingface.co/stabilityai/stable-audio-open-1.0

Structured fields
{
  "checked_at": "2026-10-17T00:00:00.000Z",
  "usefulness": "Open latent-diffusion text-to-audio model (stereo up to 47 s at 44.1 kHz) trained on ~486,500 CC0/CC-BY/CC-Sampling+ recordings from Freesound and FMA; better at sound effects and field recordings than music and cannot generate realistic vocals. Gated under the Stability AI Community License.",
  "source_type": "model",
  "availability": "available",
  "dependencies": [
    "stable-audio-tools"
  ],
  "source_owner": "Stability AI",
  "next_check_at": "2026-10-31T00:00:00.000Z",
  "known_limitations": [
    "Max 47 s; weak vocals; English prompts."
  ],
  "upstream_updated_at": "2025-06-19T21:53:23.000Z"
}

Evidence claims

  • [upstream_claim] Model card: Stability AI Community License, access requires accepting terms; for commercial use the card refers to stability.ai/license. · locator: model card, license section source
  • [upstream_claim] Training data: about 486,500 recordings mostly from Freesound plus Free Music Archive, all CC0, CC BY or CC Sampling+; Stability describes copyrighted-music filtering with PANNs, Audible Magic and human review. · locator: model card, training data source
  • [upstream_claim] Generates variable-length stereo audio up to 47 seconds at 44.1 kHz using a transformer diffusion model in a compressed latent space with a T5 text encoder; limitations: not able to generate realistic vocals, trained on English descriptions, better at SFX and field recordings than music. · locator: model card, usage and limitations source
  • [upstream_claim] Research review verdict at export: reviewed (as of 2026-10-17). Recorded by the research pipeline, not a hub review outcome; evidence.review stays unreviewed until an independent reviewer records one.

Rights

componentlicensehostingredistributionmodificationtrainingEvidence and notes
model_weightsLicenseRef-Stability-AI-Communityrestrictedrestrictedrestrictedunknown

Gated; Stability AI Community License text not read, commercial terms deferred to stability.ai/license. Training clips need CC attribution per card (separate attribution page not read).

Linked records

Machine-readable: JSON · Markdown · revisions · changes