Adobe Firefly adds music, voice and sound to the edit

Adobe Firefly audio tools are now generally available: Generate Music, Generate Speech and Generate Sound Effects let teams create soundtracks, voiceovers and effects from a prompt, inside Firefly ^[1]. In September Adobe followed with a Generative Media Tool in Premiere that creates video and sound effects directly in the editing timeline ^[2]. For brands that produce a lot of short video, this closes the last big gap between AI images and finished, publishable content.

We read both announcements with one question in mind: what changes for the teams that make social, product and campaign videos every week?

What Adobe Firefly audio includes

On August 20, 2026, Adobe announced that its audio features in the Firefly app are generally available ^[1]. Each tool runs on its own Adobe model:

• Generate Music, powered by the Firefly Music Model, creates original, universally licensed tracks matched to the length and mood of a video ^[1].
• Generate Speech, powered by the Firefly Speech Model with ElevenLabs as an option, turns a script into a voiceover with control over voice, pacing and emotion ^[1].
• Generate Sound Effects, powered by the Firefly Audio Model, creates sounds timed to the action in a clip ^[1].

Adobe also added Google’s Gemini Omni Flash to the partner models available in Firefly, alongside models from Kling AI, Luma AI, OpenAI and Runway ^[1]. The Firefly AI Assistant, which first appeared in public beta in April as an agent that works across Photoshop, Premiere, Illustrator and other Creative Cloud apps, now has a free daily generation tier and new skills such as Create Storyboard and Create Brand Kit ^{[1, 3]}.

Generating inside the Premiere timeline

On September 8, Adobe brought generation into its professional video tools ^[2]. The new Generative Media Tool in Premiere lets editors generate context-aware, fully editable video and sound effect clips without leaving the project, and choose between Adobe Firefly and partner models including Google Veo, Kling, Runway and Luma ^[2].

• Generate Video uses reference frames from the project so new footage matches what surrounds it ^[2].
• Generate Sound Effects uses Adobe’s commercially safe audio model, with voice-guided timing and intensity ^[2].
• Generate Soundscape, in beta, analyses up to 15 seconds of video and layers ambience and effects automatically ^[2].
• Generate Music, in beta, produces original music with tempo and looping controls and sync rights for any media type ^[2].
• After Effects gets an AI Assistant in public beta that handles multi-step tasks, including expressions, without code ^[2].

Premiere also gains audio clean-up in public beta, including Enhance Audio with separate control over dialogue, music, ambience and effects, and Separate Crosstalk, which splits overlapping speakers into their own tracks ^[2].

Why licensing is the real headline

AI music and voice tools have existed for a while. What most brand teams lacked was confidence about rights. A track pulled from a random generator, or a song used without a licence, can get a video muted or taken down after a campaign goes live. Adobe positions the Firefly Music Model output as universally licensed and says it removes copyright takedown concerns ^[1], and describes Premiere’s Generate Music output as including sync rights for any media type ^[2].

That claim still needs to sit inside your own approval process. Partner models are a different matter from Adobe’s own models, and the licensing terms for each are set by the model provider and your plan, so check them before using partner output in paid media.

What changes for small content teams

For a two- or three-person social team, the time sink was rarely the edit itself. It was finding a track that fit, recording a clean voiceover, and filling the gaps where a shot was missing. Those are the exact tasks Adobe targeted ^{[1, 2]}.

The risk is sameness. If every brand prompts for an upbeat corporate track, every brand will sound alike. The teams that get value from these tools will be the ones that give them a clear, specific brief, and that keep a human editor deciding what stays in the cut.

We also expect review work to shift. When a voiceover or track takes seconds to generate, the slow part becomes deciding which version is right and checking it against the brand. That is a skill worth building on purpose: a short checklist for tone, pronunciation of product names, and music that does not clash with the brand’s existing sound.

Key takeaways

• Firefly’s Generate Music, Generate Speech and Generate Sound Effects became generally available on August 20, 2026 ^[1].
• Premiere’s Generative Media Tool generates video and sound effects in the timeline, with a choice of Adobe and partner models ^[2].
• Adobe positions its own music and audio output as commercially safe and licensed ^{[1, 2]}.
• Brand guidelines should now cover sound and voice, not only visuals.
• Check licensing per model before using partner model output in paid campaigns.

Quick answers

Do we need a separate subscription for the audio tools? Adobe says no separate subscriptions are required for them in Firefly ^[1].

Can we use our own brand voice? Generate Speech gives control over voice, pacing and emotion, with ElevenLabs as an alternative model ^[1]. Test it against your guidelines first.

Is generated footage ready for final delivery? Treat it as fill and b-roll, review every clip, and keep real product footage for anything that makes a claim about the product.