How to Remove Filler Words (Um, Uh, Like) From a Video

Short answer To remove filler words from a video, cut each "um" out by hand (split the clip just before and just after it, then delete the piece), or use a tool that finds them for you. On iPhone, VCUTS removes the ums, uhs, false starts and long pauses automatically, and you can tap any cut to bring it back. On a computer, transcript editors such as Descript and Adobe Premiere Pro do the same from the text.

You hit record, you talk, and the idea comes out fine. Then you watch it back and hear every "um", every "so, uh", every sentence you started twice. Re-recording takes another ten tries, and the tenth one sounds stiff. The better fix is to keep the take where you sounded like yourself and cut the filler out of it.

This guide covers what to cut, what to keep, and three ways to do it, from free to fast.

What counts as a filler word?

Fillers come in a few kinds, and a clean edit deals with all of them:

• Hesitation sounds: um, uh, er, hmm. They carry no meaning; they hold your place while you think.
• Filler words: like, you know, so, basically, actually, I mean, kind of. These are real words, so they're only filler in some sentences: "I like this lens" needs its "like", "it was, like, so good" doesn't.
• False starts and stutters: "We're go— we are going." "The, the best part."
• Restarts and earlier takes: you say a line, don't like it, and say it again. Only the last version belongs in the video.
• Off-script remarks: "Wait, let me start again." "Okay, one more time."
• Dead air: the long pause while you look for the next thought.

That's why a simple search for the word "like" doesn't work: whether a word is filler depends on the sentence it's in. Restarts and repeated takes don't contain any filler word at all, yet they are often the biggest thing to cut.

Should you remove every filler word?

No. Cut the hesitation sounds, the restarts and the long pauses; be gentle with the rest. A few natural "so"s and a breath between sentences are what make you sound like a person rather than a voice-over. Linguists have found that speakers use "uh" and "um" to signal that a pause is coming ( Clark and Fox Tree, 2002), so in conversation they aren't simply mistakes. They're also normal: in one large study of conversation, people said about 2.6 ums and uhs per 100 words, on top of repeated words and restarted sentences (Bortfeld et al., 2001). In a 60-second video, though, every second counts, and the ums are the first thing to go.

A good rule of thumb:

• Always cut: um, uh, false starts, earlier takes, "let me start again".
• Usually cut: pauses longer than about half a second, "you know" and "basically" used as padding.
• Keep if it's you: a "so" that starts a thought, a "like" that's part of how you talk, short breaths between sentences.

Method 1: Cut them by hand on iPhone (free, slow)

Any editor that can split a clip will do, such as iMovie or CapCut. The steps are the same in each:

1. Import the video and zoom into the timeline so you can see the audio waveform.
2. Play to just before the filler and split the clip there.
3. Move to just after it, where the next word starts, and split again.
4. Select the small piece between the two splits and delete it.
5. Repeat for every filler, restart and long pause, then watch the whole thing once more.

It works, and it's free. The catch is time: a two-minute video with twenty fillers means forty splits, each placed by ear. Cut a few frames too early and you clip the next word; too late and half an "um" stays in.

Method 2: Edit the transcript on a computer

Transcript-based editors turn your speech into text, and deleting words from the text deletes them from the video. Descript ("Remove filler words") and Adobe Premiere Pro's text-based editing can both find filler words and remove them in one go, and both can cut pauses. Check the language: Descript's filler detection works on English transcripts only.

They're powerful, and worth it if you already edit long videos at a desk. For a video you filmed on your phone and want to post from your phone, they add steps: move the file to a computer, learn the editor, export, move it back.

Method 3: Let VCUTS do it on your iPhone

VCUTS is an iPhone app that removes ums, pauses, false starts and repeated takes from talking-head videos automatically. It's built for one job, the step before you post:

1. Open VCUTS and pick the video from your Photos.
2. In seconds you get the clean take: the ums, uhs, false starts, earlier takes, off-script remarks and long pauses are folded away.
3. Read through it. Tap any fold to hear what was cut, and tap again to bring it back. Nothing is ever lost.
4. Pick how tight the pauses should be (Light, Natural or Tight), add captions if you like, and save to Photos or share it straight to TikTok, Instagram or CapCut.

If you say "like" on purpose, tell it: "Keep 'like', it's how I talk." It works in more than 90 languages and recognizes which one you speak, so the French "euh" and the Spanish "este" go too (here's a list by language).

Which method should you use?

Wondering what the editors you already have can do? See which iPhone apps can remove filler words.

How to make the cuts invisible

Whichever way you cut, these keep the result smooth:

• Cut in the silence, not in the sound. Place each cut in the quiet just before the next word, so no half-syllables survive.
• Shorten pauses rather than deleting them. A short breath between sentences sounds natural; zero gap sounds like a machine. More in how to remove silence without it sounding choppy.
• Hide the jump. When your head moves between two cuts, a small zoom-in, a caption change or a cutaway makes it feel deliberate. See jump cuts.
• Watch it once at full speed with sound. You'll catch the one cut that clips a word.

Say fewer ums in the first place

Editing is the fast fix, but a few habits leave less to cut: pause silently instead of filling the gap, talk from bullet points instead of a word-for-word script, and when you stumble, keep recording and say the line again. The earlier take is easy to cut later. More in how to stop saying "um" on camera and how to edit out mistakes and retakes.