ElevenLabs can now generate sound effects from text

Finding the exact right sound effect for a video or game usually means digging through clunky audio libraries for hours, only to realize the one sound you need requires an expensive commercial license.

ElevenLabs just launched its text-to-sound-effects model, rolling out the feature to all users. Instead of scouring stock libraries, creators can now generate sound effects, short instrumental tracks, immersive soundscapes, and character voices straight from a text prompt.

Prompt to foley

Having built a dominant presence in AI voice cloning and lifelike text-to-speech, ElevenLabs is pushing deeper into complete audio production.

The new tool targets film and TV studios, video game developers, and social media creators looking to build rich audio environments quickly and on a budget.

The workflow is dead simple: users type a text description of the sound they need, hit generate, preview the generated samples, and download the best match. ElevenLabs has also made a free version available so anyone can test it out.

Backed by Shutterstock's catalog

To keep Hollywood and game studios from running away over copyright fears, ElevenLabs teamed up with Shutterstock.

The model was fine-tuned directly on Shutterstock's licensed audio library. Shutterstock Chief Enterprise Officer Aimee Egan noted that the partnership pairs ethically sourced creative data with cutting-edge generation to deliver a commercially viable tool.

That enterprise-friendly approach clears a massive legal hurdle for studios that want to integrate generative audio into professional workflows without worrying about lawsuits.

Why it matters

AI video models get all the hype, but mute video is useless. Sound design has quietly been one of the biggest bottlenecks for solo creators and indie game studios.

For creators, this tool drastically cuts the cost and turnaround time of foley work. You no longer need a dedicated sound designer or an expensive audio subscription just to find rain sounds or a futuristic sci-fi door hum.

We think ElevenLabs made a very smart move by locking down licensed data with Shutterstock right at launch. Many generative tools stumble when trying to sell to big studios because their training data is a legal gray zone. By solving provenance upfront, ElevenLabs makes enterprise adoption far easier.

That said, generating a generic blast or atmospheric wind is one thing; getting precise, frame-accurate synchronization for interactive gaming or complex film scenes is another. It will be fascinating to see whether prompt-based audio can truly displace hands-on sound engineering in high-end productions.