Adobe Firefly Audio Tools Are Now Generally Available

Adobe Firefly audio tools are now generally available, effective August 20, 2026. Generate Music, Generate Speech, and Generate Sound Effects all exit beta today, usable inside any paid Firefly plan starting at $9.99 a month. Firefly AI Assistant also gains a free daily-generation tier, and Gemini Omni Flash joins Firefly’s roster of partner models.

None of this costs extra beyond a Firefly subscription, and that is where Adobe’s “commercially safe” pitch gets more complicated. Adobe’s own Firefly-trained audio models carry the company’s legal indemnity. The ElevenLabs voice option built into Generate Speech does not, a distinction Adobe’s announcement never spells out.

Generate Music and Generate Sound Effects Leave Beta

Adobe Firefly’s Generate Music tool uses the Firefly Music Model to produce what Adobe calls universally licensed original tracks, tuned automatically to a video’s length and mood rather than requiring manual trimming. It moved out of beta today alongside its two siblings, more than nine months after Adobe first showed it at Adobe MAX in October 2025 under the working name Generate Soundtrack.

Generate Sound Effects runs on the Firefly Audio Model and works the same way in spirit: describe the sound you need, and the Adobe Firefly sound effects generator matches it to the action, timing, and energy of your footage. It’s the most mature of the three. It entered public beta back in July 2025 with impact sounds and atmospheric noise, and today it exits alongside music and speech.

Adobe built the launch post around creator testimonials rather than spec sheets. Madeline Salazar, one of the creators quoted, put the appeal in blunt terms: “I know the music is safe for commercial use.” Another, Esther Rehema, said the hunt for usable background music used to eat hours of her time. Now it takes seconds.

For a broader look at how Firefly’s music generator stacks up against dedicated tools, see our comparison of the best AI music generators in 2026.

What Is Generate Speech’s ElevenLabs Option?

Generate Speech turns a script into a voiceover, with controls for voice, pacing, and emotion, the same category of settings that matter in dedicated AI dubbing tools. Adobe Firefly’s Generate Speech runs on the Firefly Speech Model by default, but the interface also lets you swap in ElevenLabs as the voice engine for a given generation.

That choice is presented as a style option, not a legal one. At its Adobe MAX beta debut in October 2025, Generate Speech shipped with more than 50 voices across more than 20 languages, according to The Verge’s coverage at the time. Adobe hasn’t published updated voice or language counts for today’s general-availability release, so treat those beta-era figures as historical rather than current.

The two engines look interchangeable inside the Firefly interface. They are not interchangeable in Adobe’s contract terms, which is the part worth reading before you commit a client project to one or the other.

Does Adobe’s Indemnity Cover the ElevenLabs Option?

Adobe’s IP indemnity covers Generate Music, Generate Speech, and the sound-effects features when they run on Adobe’s own Firefly models. Generate Music and Generate Speech were added to the covered list in a July 13, 2026 legal update. Speech generated with the ElevenLabs option is not covered, because Adobe’s terms exclude output from non-Adobe trained models.

Adobe’s Firefly indemnity applies only to what its legal Product Description calls “Eligible Firefly Features.” The sound-effects features joined that list back on September 30, 2025, listed as Text to Sound Effects and Voice to Sound Effects. Generate Speech and Generate Music were added far more recently, in a July 13, 2026 update, five weeks before today’s launch.

The same legal document carries an exclusion clause that matters more than the addition does. It excludes capabilities denoted in the interface as powered by non-Adobe trained models, along with beta and trial features. Adobe’s own documentation treats the ElevenLabs engine as a partner model rather than an Adobe-trained one, which puts speech generated through it outside the indemnity that covers the Adobe-model default.

Adobe does maintain a separate “Creative Partner Model Supplemental Coverage” for select outside models, but it only applies to Google and OpenAI’s image and video models, and only on specific enterprise licenses. ElevenLabs, Kling AI, Luma AI, and Runway aren’t on that supplemental list, and audio partner models aren’t mentioned in it at all.

None of this is a legal requirement in the way that content-labeling rules under Article 50 of the EU AI Act are. It’s a private contractual carve-out Adobe chose to write. But the practical takeaway for anyone using Firefly commercially is similar either way: don’t assume “made in a commercially-safe AI tool” and “covered by Adobe’s indemnity” mean the same thing for every option inside that tool.

Should You Bundle with Firefly or Pay ElevenLabs Direct?

A Firefly subscription buys access to more than Adobe’s own models. The platform integrates outside models from Google, ElevenLabs, Kling AI, Luma AI, OpenAI, and Runway, which became a preferred Firefly video partner under a multi-year deal announced in December. Today’s release adds Gemini Omni Flash to that lineup, letting creators prompt with video, audio, and image inputs together for a storyboarded first cut. It’s a similar bundling logic to Adobe’s node-based Project Graph workflow editor: pay for one Firefly plan, get a growing shelf of models to pull from.

The question is whether that bundle beats paying for the pieces separately, and for voice specifically, it often doesn’t.

Adobe is also running a limited-time promotion on its bigger plans, though the discounted prices aren’t in the announcement, so the table above reflects list pricing only.

Read that table by use case rather than by which number is smaller. A video-first creator already paying for Firefly to generate B-roll, images, and Premiere workflows gets Generate Music and Generate Sound Effects for free inside that same subscription, and picking Adobe’s own Speech Model keeps the indemnity intact. A voice-heavy workflow (dubbing, audiobooks, character voices) is a different case: ElevenLabs Starter at $6/mo with a commercial license undercuts every paid Firefly tier on price for that one job, and it comes with instant voice cloning that Firefly’s bundled option doesn’t replicate. Neither path is the wrong one. They’re just answering different questions.

FAQ

---

Sources

1. Adobe Blog: Adobe Firefly Expands Its Creative AI Studio - August 20, 2026
2. The Verge: Adobe’s AI audio generators exit beta - August 20, 2026
3. The Verge: Adobe Firefly adds AI audio generation - October 28, 2025
4. Adobe: Firefly Product Description (IP Indemnification)
5. Adobe: Creative Partner Models Product Description
6. Adobe: Firefly Plans
7. Adobe: Generative Credits FAQ
8. ElevenLabs: Pricing (elevenlabs.io/pricing)
9. The Verge: Adobe’s new AI tool turns silly noises into realistic audio effects - July 17, 2025