Adobe brings AI-generated soundscapes to Premiere Pro

Adobe Research and the Premiere product team have built Generate Soundscape, a new artificial-intelligence feature inside Premiere Pro Desktop that studies a video clip and automatically lays down four synchronized audio tracks on the timeline — two ambient layers and two foreground effects — instead of leaving editors to build a scene’s sound from nothing.

Justin Salamon, who leads Adobe Research’s Sound Design AI team, calls the gap the feature targets the “blank page problem”: even when individual effects can already be generated on demand, an editor still has to figure out which sounds a scene needs, write a prompt for each one, and drag every clip into place by hand. “Generate Soundscape helps creators reach a good starting point from which they can iterate and really make it their own,” Salamon says.

The feature draws on video-guided sound-generation research Adobe presented at CVPR 2025, which trained a model to tell which of several simultaneous on-screen actions should trigger a given sound — distinguishing, say, a door closing from a person crossing the frame at the same moment. That same underlying work also drives Sync to Video, a separate Premiere Desktop tool that times one prompted sound effect to the pace and intensity of the action on screen. The idea of stitching several generated sounds into one full scene first surfaced in Project Sound Stager, a prototype Adobe Research engineer Oriol Nieto demonstrated as an Adobe MAX Sneak in 2025; the response pushed Adobe Research and the Premiere team into the collaboration that produced Generate Soundscape. Adobe Research engineer Johannes Imort, who helped bring the capability into Premiere Desktop, says feedback from professional sound designers during testing flagged gaps such as footstep sounds, which the team then used to refine the underlying model.

Unlike a single rendered audio track, each of the four generated layers stays separate and editable: creators can raise or lower one track, trim its length, or regenerate just one element — replacing a footstep sound, for instance — without touching the rest of the mix. Generate Soundscape is live now in Premiere Pro. At Adobe MAX 2026, the research team also showed Project Sound Bender, a further prototype aimed at letting creators reshape and manipulate audio interactively.

Why it matters

Generate Soundscape turns sound design — one of video editing’s more manual, time-consuming steps — into an AI-assisted first draft that creators edit rather than build from scratch, and Adobe is already pushing further: Project Sound Bender, shown at Adobe MAX 2026, points toward sound that editors can reshape interactively rather than only regenerate.