Remove silences in Premiere Pro in one reviewed pass

Um Out reads the whole dialogue track on your timeline and proposes every gap it could remove, with the minimum length under your control. You see the count and the total time it would save before a single frame changes.

Each proposed cut is a tick box. Untick the beat before a punchline, the breath between two ideas, the pause you took on purpose. What is left is applied as ordinary Premiere edits you can trim by hand afterwards.

Picture and dialogue are cut together. Because Premiere’s plugin API has no razor tool, Um Out rebuilds the visible segments rather than slicing them, which is why the video track keeps up with the audio instead of being left behind. This is the failure that makes most automatic silence removers unusable on a talking head, and it is the first thing we fixed.

Two settings do most of the work. Room before speech and room after speech hold back a little quiet on each side of the speech that is kept, so words do not start or end abruptly. The presets set both for you, and Relaxed, Balanced and Fast are there because a tutorial and a vlog want different pacing.

J and L cuts let audio lead or trail the picture across a join. An L-cut keeps the current audio running as the next shot begins; a J-cut brings the next audio in early. Both are applied as real keyframed edits you own, and both are what stop a tightened video sounding like a list of sentences.

Detection runs on your computer from the waveform alone. No transcript, no internet connection, and no footage leaving your drive, which also means no per-minute charge for watching your own recording.

A four-hour file is limited by your machine rather than by an upload, so it works the same on a plane as it does at your desk. Web editors charge by the minute because they pay for the compute. You already own the compute.