ElevenLabs teases upcoming AI sound effects

When OpenAI dropped Sora last week, the internet lost its mind over how cinematic AI video had become. But once the dust settled, everyone noticed the obvious catch: every single clip was a completely silent movie.

ElevenLabs saw the opening and moved fast. The audio startup just showcased its upcoming AI sound effects model by doing the most logical flex possible, slapping realistic audio onto Sora's best silent clips.

Foley artists in a prompt box

ElevenLabs essentially built a digital foley stage. You feed the model natural language descriptions of the scene, and it generates matching ambient sounds and audio effects.

In their public demo, the team used straightforward prompts like waves crashing, metal clanging, birds chirping, and racing car engines.

They layered these generated sounds over Sora footage with striking accuracy. A futuristic bipedal robot now hums with mechanical whirs and digital beeps, busy city street scenes feature footsteps under ambient traffic drone, and they even threw in cinematic Hollywood-style narration over dramatic shots.

Founded in 2022 by ex-Google machine learning engineer Piotr Dabkowski and ex-Palantir strategist Mati Staniszewski, ElevenLabs previously built its reputation on voice cloning and multilingual dubbing. Sound effects mark their first major push beyond spoken human dialogue.

The feature isn't widely available yet. There is no official release date, but the company has opened up a waitlist for early access.

Why it matters

For creators and small video teams, audio sourcing is notoriously tedious. Creators often burn hours digging through generic stock audio libraries just to find a decent door slam or natural room tone. Being able to conjure custom ambient soundscapes with a few words could remove friction from post-production immediately.

Strategically, ElevenLabs played this brilliantly. Instead of racing to build a me-too video generator, they positioned themselves as the missing audio puzzle piece to the biggest tech story of the month. It establishes their brand as the default audio layer for generative video.

That said, we think the long-term risk is whether stand-alone sound generation remains necessary. Right now, this requires humans to prompt audio separately and sync it manually to video. If next-generation video models learn to natively output synchronized, rich multi-channel audio directly alongside the pixels, third-party foley tools could end up being a short-lived bridge. Still, if you produce video content, getting on this waitlist is well worth your time.