Stability's New Ai Audio Tool Creates Custom Sound For Brands - How It Works

Tatiana Serebryakova/iStock/Getty Images Plus via Getty Images

Follow ZDNET: Add america arsenic a preferred source on Google.

ZDNET's cardinal takeaways

• Stable Audio 2.5 is designed to thief brands build a "sonic identity."
• The exemplary was trained connected a afloat licensed dataset.
• Custom tracks tin beryllium utilized successful ads, unit locations, and elsewhere.

Stability AI conscionable made it easier for brands to create custom, AI-generated audio, thereby negating nan request to walk clip and money connected elaborate signaling and accumulation processes.

The UK-based institution unveiled Stable Audio 2.5 connected Wednesday, describing nan caller exemplary connected their website arsenic "the first audio procreation exemplary designed specifically for enterprise-grade sound-production." 

Also: 4 ways machines will automate your business - and it's nary hype, says Gartner

Stable Audio 2.5 is intended to thief brands create high-quality and afloat licensed audio clips that tin beryllium utilized crossed a assortment of channels to fortify their "sonic identity" -- that is, nan postulation of sounds associated pinch their unsocial trading and branding.

"To thief enterprises create nan correct sound, our squad tin fine-tune Stable Audio models connected an organization's sound library, embedding signature marque audio into civilization generative workflows," Stability writes. "This ensures that nan euphony aliases soundscape is uniquely recognizable arsenic portion of a brand's sonic personality aliases imaginative guidelines for a project."

What tin Stable Audio 2.5 do?

Stability AI said its caller exemplary tin create civilization philharmonic tracks of up to 3 minutes wrong seconds. It tin besides spell beyond monotone jingles to create "multipart compositions," complete pinch an intro, a mediate section, and an outro. 

Audio 2.5 tin besides respond to earthy connection punctual specifications, for illustration "uplifting," which modify nan reside and tenor of its output (similarly to caller features offered successful text-to-speech models from companies for illustration ElevenLabs).

Also: I tested 3 text-to-speech AI models to spot which is champion - perceive my results

There's besides an "inpainting" feature, enabling users to upload a snippet of their ain audio, which nan exemplary will past automatically build upon. Stability AI's contented moderation strategy will, however, cull immoderate copyrighted worldly that gets uploaded.

"Like each Stable Audio models,Stable Audio 2.5 is commercially safe and trained connected a afloat licensed dataset," Stability AI wrote connected its website. 

Also: Google's NotebookLM now lets you customize your AI podcasts successful reside and length

That's important to statement fixed nan institution is presently being sued by a group of artists who declare that it illegally utilized copyrighted materials successful bid to train Stable Diffusion, its flagship image-generating model, which was released successful 2022. (Other AI companies, including Midjourney, are besides targeted successful nan lawsuit.)

Try it for yourself 

You tin effort Stable Audio 2.5 here. There's a free option that comes pinch a monthly limit of 10 civilization tracks, a $12/month Pro action pinch a monthly limit of 250 tracks, and much costly Studio and Max options.