Video editing

How to remove filler words from a video (ums, ahs and dead air)

By Tushar Agarwal, Founder, Explaino · Updated October 17, 2026 · 6 min read

Quick answer

To remove filler words from a video, transcribe it, find every 'um', 'uh', 'like' and long pause in the transcript, and cut the matching moments from the timeline. Doing this by hand takes roughly 3 to 5 minutes of editing per minute of footage; AI tools like Explaino detect and cut fillers and dead air automatically when the recording is processed.

Watch: Removing filler words from a screen recording automatically

Video walkthrough coming soon

This video is being made with Explaino from a single screen recording. You can make your own version in a few minutes.

Make this video free

Made with Explaino: recorded once, then narrated, zoomed and captioned automatically.

Step by step

1
Transcribe the video
Work from text, not the waveform. A transcript shows every filler at a glance, while scrubbing audio makes you re-listen to the whole take.
2
Mark fillers and dead air
Flag 'um', 'uh', 'you know', 'like', restarts of a sentence, and any silence longer than about 1.5 seconds. Keep natural half-second pauses — removing every gap makes narration sound robotic.
3
Cut on the transcript, not the timeline
Text-based editors delete the video frames that match deleted words. If you edit a timeline manually, cut a few frames before and after the filler so mouths and cursors don't jump.
4
Smooth the seams
Listen to each cut. Where the audio jumps, add a tiny crossfade; where the screen jumps, prefer cutting during a still moment rather than mid-scroll.
5
Or let AI do the whole pass
In Explaino, filler words, false starts and dead air are detected and removed automatically after you stop recording, and the cleaned narration can be re-voiced with an AI voiceover so no seams are audible at all.

Why fillers hurt more in tutorial and demo videos

In a casual team update, a few ums are fine. In a product demo or help-center tutorial they compound: viewers are trying to follow steps, and every hesitation makes the step less clear and the video longer. Trimming fillers and dead air routinely shortens a raw recording by 20 to 40 percent without losing a single step.

The hidden cost is maintenance. If you polish a take by re-recording until it's perfect, every product change means re-performing the whole take. If you record naturally and clean it in software, updates stay cheap.

Manual vs. AI cleanup

Manual editing gives you frame-level control and is the right call for cinematic content. For screen recordings, the economics favour automation: the edits are repetitive (find silence, find filler, cut, smooth) and the footage is forgiving, because the screen rarely moves during a filler.

Manual: full control, roughly 15–30 minutes of work per 5-minute video.
Text-based editing: faster, still one pass of human review per video.
Automatic (Explaino): fillers, restarts and dead air removed on upload; the script can be re-voiced so cuts are inaudible.

Record it once. Explaino does the rest.

Turn one screen recording into a narrated video with smart zoom and captions, plus a step-by-step article, in 70+ languages.

Start free in Explaino Studio Get the Chrome extension

Free plan · no credit card required

Frequently asked questions

Should I remove every pause?

No. Pauses under about a second are natural pacing, and tutorials need a beat after each click so viewers can follow. Remove hesitations and dead air, keep deliberate pauses.

Will cuts make the video look jumpy?

On a talking-head video, hard cuts are visible as jump cuts. On a screen recording they are usually invisible because the screen is still during fillers. Re-voicing the cleaned script with an AI voice removes audible seams entirely.

Can I keep my own voice?

Yes. You can keep your cleaned original audio, or clone your voice from a short sample and have the AI read the cleaned script in a voice that sounds like you.

How to remove filler words from a video (ums, ahs and dead air)

Watch: Removing filler words from a screen recording automatically

Step by step

Transcribe the video

Mark fillers and dead air

Cut on the transcript, not the timeline

Smooth the seams

Or let AI do the whole pass