Editing
How to Remove Silences and Dead Air from a Video (Without Making It Sound Robotic)
By the ManyMotion team · · 9 min read
You film yourself, watch the take back, and it feels slow. Your content isn't the problem: it's the dead air. The silences between sentences, the "ums", the sentences restarted three times. In conversation, nobody notices them. In a video, each one is a chance to scroll away. Here's why you should cut them, how to do it cleanly by hand or with AI, and when it's better to keep a pause.
Why silences drive viewers away
When you talk to someone face to face, pauses are useful: they give you time to find your words and give the other person time to process. Video is different. Your viewer isn't captive. The next video is a thumb swipe away, and their brain is constantly looking for a reason to stay or leave.
Half a second of silence doesn't seem like much. But in an improvised two-minute take, those little gaps add up fast. Add hesitations and restarted sentences, and a big chunk of your video says nothing at all. That's attention spent for nothing.
Creators who hold their audience figured this out long ago: they cut almost every breath. That very tight style, with visible cuts, has become the norm on TikTok, Reels and Shorts, and has largely taken over YouTube too.
What to cut (and what to keep)
Not all gaps are equal. Before you cut, know what you're looking for:
| Type | Example | Cut it? |
|---|---|---|
| Silence between sentences | A second of dead air while you think | Yes, almost always |
| Filler word | "um", "uh", "like", a repeated "so" | Yes |
| False start | "So today... no. Today I'm going to..." | Yes: keep the complete version |
| Repeat | The same sentence said three times to get it right | Yes: keep the best take |
| Dramatic pause | A silence right before a reveal | No, if it's deliberate |
| Natural breath | An inhale before a long sentence | Depends on the tone you want |
The rule: cut what's unintentional, keep what's intentional. A pause before announcing a price, a silence after a moving line, a beat to let people laugh: those are storytelling tools. Removing them makes the video feel mechanical.
Tone matters too. On a high-energy TikTok, you remove everything. On a ten-minute training video or a personal story, you keep a few breaths: a pace that's too tight tires the ear over time.
The manual method: cutting silences in an editor
If you edit on a timeline yourself, here's the most efficient method:
- Show the audio waveform. Silences appear as flat areas between peaks. It's much faster than listening.
- Do a rough first pass. Cut the big gaps and the botched takes without worrying about precision. The goal is to reduce the material.
- Do a fine second pass. Listen again and tighten each cut as close as possible to the last word and the next, without clipping a syllable.
- Remove hesitations mid-sentence. This is the slowest part: an "um" stuck to a word requires a very precise cut.
- Listen to the whole thing without watching. Your ear catches cuts that are too abrupt or clipped words better when your eyes aren't busy.
Expect to spend three to five times the length of your take on a careful cut. A five-minute take is often half an hour of work before you even add captions.
The AI method: cutting based on words, not volume
The most reliable way to cut silences is to start from a word-by-word transcript. If you know exactly when each word starts and ends, you know precisely where the gaps are, and you can cut in the space between two words, where the ear doesn't notice anything.
That's how ManyMotion works. You upload your video (MP4, MOV or WebM, up to 30 minutes, even shot on a phone), and the AI:
- transcribes every word with its exact timestamp, in English, French, Spanish, Portuguese, German and many other languages;
- detects silences, but also hesitations, false starts and repeated sentences;
- keeps the best take when you restarted a sentence;
- always cuts between two words, never in the middle;
- can hide the cuts with a zoom or B-roll, if you want.
You set the pace you want in one sentence: "very tight, TikTok-style" or "natural pace, keep a few breaths". If a passage you wanted to keep has disappeared, just ask: "put back the sentence about opening hours". Everything is explained on our remove silences from video page.
Example request: "Cut all the silences and ums from my video, keep only the best version of my intro, tight pace, and add yellow pop captions."
The jump cut: flaw or style?
When you cut a chunk out of the middle of a static talking-head shot, your head makes a small jump between the frame before and the frame after. That's the jump cut. Long considered an editing mistake, it has become an accepted convention of talking-head videos.
On social media, a jump cut signals "we're getting to the point". So you don't need to hide it every time. But on a calmer video, or when the jumps are very frequent, you may want to soften it:
- The punch-in zoom: the frame after the cut is slightly tighter on your face. The jump becomes a deliberate change of shot, and it adds movement.
- Alternating framing: a wide shot, then a close shot, then a wide shot. Each cut changes the shot size.
- B-roll: while your voice keeps going, viewers see what you're talking about. The cut disappears under the image.
- The transition: a flash, a blur, a whip. Best kept for section changes, otherwise it gets tiring.
ManyMotion combines these techniques automatically: snappy or smooth zooms on your face thanks to face detection, B-roll from another video you've uploaded (up to 5 videos per message), transitions at topic changes.
Filming to make cutting easier
Editing starts when you film. A few habits make the cuts cleaner, whether you edit by hand or with AI:
- Mount your phone on a stand. A moving frame makes every cut visible and jarring.
- When you slip up, pause, then restart the whole sentence. A clean retake is easy to cut; a word corrected on the fly, much less so.
- No music while filming. It would get chopped at every cut. Music is added during the edit, continuously.
- Clean audio: a quiet room, a lavalier mic or a phone close to you. The quality of the transcript, and therefore of the cuts, depends on it.
- Don't hold back: speak naturally and retake as often as you need. Silences aren't a problem once they get cut.
Mistakes that make a video sound robotic
- Cutting mid-word: a clipped syllable is instantly audible and sounds amateurish.
- Removing every breath on a long video: after three minutes, the speech becomes suffocating.
- Cutting dramatic pauses: the punchline of a joke or the reveal of a number loses all its impact.
- Leaving audio jumps: if your background noise changes abruptly at every cut, music in the edit helps smooth the transitions.
- Ignoring consistency: a sentence stitched together from two different takes can change intonation halfway through. Keep a full take instead.
Once the silences are gone, your video is ready for the rest of the edit: auto captions, zooms, music. And to go further on pacing, read our editing techniques that keep viewers watching to the end.
Frequently asked questions
How do I remove silences from a video automatically?
+
Use a tool that cuts based on a word-by-word transcript rather than volume. With ManyMotion, you upload your video and the AI removes silences, filler words, false starts and repeats, always cutting between two words.
Should I cut every silence?
+
No. Cut unintentional silences (hesitating, searching for words) and keep intentional pauses, like the one before a reveal. On a long video or an emotional testimonial, also keep a few breaths.
What is a jump cut?
+
It's a visible cut within a static shot that makes the person on screen "jump" slightly. It's an accepted convention of talking-head videos. You can soften it with a zoom, B-roll or alternating framing.
How much does removing silences with ManyMotion cost?
+
Uploading costs 2 credits per started minute, then 1 credit per 2 seconds of final video. A 3-minute take edited down to 1 minute uses 36 credits. You get 60 free credits at sign-up. See pricing.