Sketch2Sound Adobe
verifiedFreehugofloresgarcia.art
Adobe/Northwestern research model that generates audio from text plus time-varying controls like loudness, brightness and pitch.
What it does
Sketch2Sound is a research generative-audio model from Adobe Research and Northwestern University. It synthesizes high-quality sounds from text prompts combined with interpretable control signals (loudness, brightness, pitch) or vocal imitations. It can be added to any text-to-audio diffusion transformer with lightweight fine-tuning.
Core features
Text-to-audio generation
Control via loudness, brightness and pitch signals
Synthesis from vocal/sonic imitations
Lightweight fine-tuning on existing diffusion models
Flexible temporal control
Best for
→Create sound effects synced to video
→Generate audio from vocal imitations
→Prototype sounds for sound design