Adobe makes Firefly a full audio studio, and the licence is the actual product
Generate Music, Generate Speech and Generate Sound
Effects came out of limited release on 20 August. Adobe's pitch rests on
provenance, at a moment when more than half of everything uploaded to Deezer
each day is AI-generated with no traceable origin.
Adobe has flipped its three Firefly audio tools to
general availability. Generate Music, Generate Speech and Generate Sound
Effects are now open to everyone on Firefly's web and mobile apps, announced on 20 August 2026 by Adobe VP of product
marketing and general manager for creators, Mike Polner.
The audio quality is fine. It's the licence
attached to the output that's worth your attention, and that's a deliberate
choice by Adobe.
Generate Music runs on the Firefly Music Model.
Point it at a video and it reads the footage, drafts an editable prompt, then
returns 4 options with adjustable tempo, energy and duration, cut to the length
of your clip. Adobe calls the output "universally licensed", meaning
it's meant to travel wherever your content goes without triggering a takedown.
Generate Speech runs on the Firefly Speech Model,
with ElevenLabs Multilingual v2 sitting
in the model dropdown as an alternative. You get control over voice, pacing and
emotion, plus expressivity and pronunciation tags, across more than 20
languages, exporting as .wav.
Generate Sound Effects runs on the Firefly Audio
Model. Describe the sound, or perform it yourself as a voice reference, and it
matches the action and timing of the scene.
"Sound is usually the last thing you add and
the first thing people notice," Polner wrote in the announcement. That's a
good line, and it's also the commercial argument: audio is where creative flow
breaks and people go shopping for a second subscription.
Leave A Comment