Adobe opened Firefly's music, speech and sound effects generators to everyone

Adobe pushed Firefly’s three audio generators out of limited release and made them available to everyone, the company announced on August 20, 2026. Generate Music, Generate Speech and Generate Sound Effects now sit in the same browser workspace as Firefly’s image and video tools.

The pitch is aimed at people cutting video, not at people making records. Adobe’s framing is about the moment a video is nearly done and the sound is missing.

What Adobe Firefly’s three audio generators do

Generate Music runs on the Firefly Music Model. You give it a length and a mood, and it returns an original track cut to fit. Adobe’s own Firefly product page leans on the phrase “universally licensed” and on the promise your music travels with your content without a takedown landing on it.

Generate Speech turns a script into a voiceover with control over voice, pacing and emotion. It defaults to the Firefly Speech Model, and Adobe lets you switch the engine to ElevenLabs inside the same step. That’s a notable choice: Adobe is shipping a competitor’s model as an option rather than walling its own off. ElevenLabs sits in Firefly next to Google, Kling AI, Luma AI, OpenAI and Runway.

Generate Sound Effects builds custom effects matched to the action and timing on screen. It runs on the Firefly Audio Model and is the tool aimed at the gap where a library search returns nothing usable.

Adobe also added Gemini Omni Flash to the model roster. It takes video, audio and image inputs alongside text, so a reference clip can become a storyboarded first cut without leaving the app.

The licence covers your output, not the training set

The commercial-safety claim is the reason a video editor would pick this over ElevenLabs Music or a stock library. Adobe is saying: use the output, ship it, we stand behind it.

Read the claim precisely. “Universally licensed” describes what Adobe grants you over the file it hands back. The August announcement says nothing about what the Firefly Music Model learned from, or whether the musicians in that data were paid. Those are two separate agreements, and only one of them was published. Adobe has been building audio models in the open since Project Music GenAI Control, and the training disclosure has been the quiet part throughout.

Firefly’s audio tools are live now at firefly.adobe.com, and Adobe pairs the release with a free AI Assistant tier that includes a daily generation allowance.