Cut the silence out of a video

The audio is genuinely decoded and measured in short windows — there's no shortcut based on file size or a fixed pattern. Anything quieter than the threshold, for at least the minimum length, becomes a candidate cut. A stretch that's quiet but too short (a breath, a comma) is left alone on purpose, because slicing every micro-pause out of natural speech makes a recording sound chopped rather than tightened.

Threshold sets how quiet counts as silent — raise it (toward 0) to catch more background hiss and room tone, lower it (toward -60) to only catch true dead air. Minimum length stops short pauses from being cut at all. Padding leaves a sliver of the quiet attached to the speech on each side of a cut, so a word's onset or ending never gets clipped — the timeline updates live, so you can see the effect of each change before committing to a full render.

The video is genuinely re-encoded, because stitching kept ranges together means changing which frames exist — there is no wrapper trick for that. Audio doesn't need to be re-encoded, so it's copied across as compressed packets exactly like the site's other video tools do, provided it's AAC. Any other codec can still be measured for silence — removing the quiet parts from the picture is still useful — but can't be carried into the output this way, so it's dropped rather than silently guessed at, and the plan says so before you convert anything.

Listening to the audio, deciding what to cut, and re-encoding the video all happen in this page. There is no server here to send the video to, and the page's security policy blocks any connection to anywhere but this site. Here is how that is enforced.