Adobe has taken three AI audio tools out of testing and made them broadly available inside Firefly, its creative AI platform. Generate Music builds royalty free tracks for videos. Generate Speech turns a written script into a voiceover. Generate Sound Effects creates audio for a single scene. Adobe says all three are cleared for commercial use.
The commercial use claim is the part worth paying attention to. Adobe has spent years building its pitch around being the "safe" AI company for business, offering to cover legal costs if a customer gets sued over content made with Firefly. That pitch has real teeth right now, because two of the best known AI music tools, Suno and Udio, are being sued by major record labels over how they were trained.
For a company producing marketing videos, training content, or client work, that difference is not small. Using an AI tool with an uncertain legal history to make music for a paid ad campaign is a real risk. Adobe is betting that businesses will pay for certainty rather than chase the flashiest output.
The tools themselves lower a real cost. Hiring a composer for background music, booking a voice actor for a script, or licensing stock sound effects all take time and money. A small business or in house marketing team can now generate all three inside one platform, in minutes, without contracts or licensing fees.
That said, this is not a reason for anyone in professional voiceover, jingle composition, or stock audio licensing to panic overnight. These tools are still generic. A voiceover for a training video is one thing. A voice that needs to carry a brand's personality, or music written to match specific emotional beats in a film, is a different job. The freelancers most exposed are the ones doing low complexity, high volume work: short ad reads, simple background tracks, basic sound design.
Adobe also folded in Google's newest video model, Gemini Omni Flash, into Firefly. This model can take video, audio, and images as input, not just text, and edit based on instructions across multiple turns. It now sits alongside other outside models from Google, Kling AI, Luma AI, and Runway that Adobe has added to Firefly over the past year.
This pattern is the real story here. Adobe is no longer trying to build the single best AI model for every job. It is turning Firefly into a marketplace, where a user picks whichever model, its own or someone else's, works best for a given task, all inside tools like Photoshop and Premiere that businesses already use.
For any company already paying for Adobe's Creative Cloud subscription, this is a low effort way to test whether AI generated audio and video can replace parts of a production workflow that used to require outside vendors. The tools are already inside software many teams use daily, which lowers the barrier to trying them far more than a separate AI app would.