Skip to content
Transcut Studio
← All posts

October 23, 2026

How to cut silences from a talking video without clipping words

Long pauses are the main reason a talking video feels slow. Shortening them is easy to explain (“cut where it’s quiet”) and surprisingly easy to get wrong. Here’s what goes wrong, and the settings that fix it.

What “silence” means to a computer

A recording is never truly silent. There’s always room noise: a fan, traffic, the hum of the microphone. So a silence remover needs a silence level: anything quieter than that level counts as silence, anything louder counts as sound.

  • Set it too high and quiet word endings (“…left”, “…this”) count as silence and get cut. That’s the clipped-word sound.
  • Set it too low and the room noise counts as sound, and nothing gets cut.

The right level sits between your room’s noise and your voice. A good tool measures both and puts the level in the gap. If there’s music or constant noise under your voice, there may be no clear gap, and you’ll need to set it yourself.

The two settings that matter

Almost every silence tool has some version of these:

  1. Minimum length: only shorten silences longer than this. Short gaps between words are part of how speech sounds; take them out and it sounds like a robot. Around half a second is a good starting point.
  2. Padding (what to keep): how much silence to leave next to each word. This is the one that prevents clipped words: it gives every word’s ending a little room. Around 0.2 seconds on each side keeps the speech breathing.

If your cuts sound choppy, raise the padding. If the video still drags, lower the minimum length.

Why cutting by the transcript can clip words

Some editors shorten pauses using the transcript’s word timestamps: “the gap between this word’s end and the next word’s start”. The trouble is that those timestamps are estimates. They can be a fraction of a second off, and sometimes a timestamp lands in the middle of a silence, splitting one long pause into two short ones that never get shortened. Cutting by the timestamps alone can both clip words and miss pauses.

Measuring the silence in the sound avoids both: a cut only ever lands where the recording is actually quiet.

Doing it by hand

In any editor with a waveform: zoom in, find the flat stretches between words, split at both ends, leave a little room next to each word, delete the middle, and close the gap. It’s precise and very slow for a long video. It’s worth doing for the few pauses that matter most (a dramatic beat you want to keep, for instance) after an automatic pass.

Doing it in Transcut Studio

Transcut measures the silences in the sound, not from the transcript. Its Remove silences window has exactly the settings above: shorten silences longer than (0.6 s by default), silence to keep beside each word (0.2 s), and the silence level (set automatically from your room noise and your voice, or by hand). While you move the sliders it tells you what will happen, for example “14 silences: saves 0:20, new length 0:48 (was 1:08)”. One undo brings everything back.

When we compared it with a careful hand edit of a 12-minute video, it removed within 6.4% of the same amount of silence, and none of its cuts contained speech. More on the silence removal page.

After the silences: jump cuts

Tighter pauses mean more visible cuts in a talking-head shot. Three common ways to hide them: cut to B-roll over the join, zoom in slightly on every other section so each cut looks like a camera change, or leave them in: on YouTube, jump cuts are normal and keep the pace up.

Try Transcut free

No card, no account: download it with just your email. For Macs with Apple silicon.

Want Transcut on Windows?

Transcut runs on Mac for now. Windows comes later: leave your email and we'll send you one message when it's ready.

One email when Transcut is on Windows, and news emails only for the boxes you tick. We email you a link first, to make sure the address is yours. Privacy

Have a Mac? Download for Mac