How to stop saying um: what the research says, and what to do on camera
Why we say um, how much it actually hurts, the technique that reduced filler sounds in studies, and why on video you do not have to stop them all while recording.
Cover image: AI-generatedThe most reliable way to stop saying “um” is to replace it with a silent pause. Studies of public speakers found that a short training in two steps, first noticing each “um” as it happens and then pausing instead, cut filler sounds sharply, often after a single session. You do not need to get to zero: a few fillers a minute do not hurt how listeners rate you. And if you speak on video, you can remove the rest in the edit.
Why we say um
“Um” and “uh” are not random noise. Psycholinguists Herbert Clark and Jean Fox Tree, analysing large collections of recorded English conversation, found that speakers use them as signals: they announce that a delay is coming while the speaker plans what to say next. After “um”, the pause that follows was on average about twice as long as after “uh” (UC Santa Cruz).
So an “um” usually means your mouth is ahead of your thinking. That points to the fix: the problem is not the sound but the gap it fills.
How much does saying um hurt?
Less than most speaking advice suggests, until it becomes frequent.
A 2024 study in the Journal of Applied Behavior Analysis had listeners rate speeches with filler sounds and words at 0, 2, 5 and 12 per minute. High rates, especially of filler sounds such as “um” and “uh” at 12 per minute, lowered ratings across most categories. At 5 per minute, perceived effectiveness was not harmed (summary of Laske et al., 2024).
Context matters too. In a relaxed Instagram video, a filler can sound natural, even relatable. In a keynote or an investor pitch, the same filler draws more attention, so the bar is higher.
The technique that works: notice, then pause
The method with the best evidence comes from behavioural research on public speaking, often called habit reversal. Studies with small groups of speakers used two steps:
- Awareness training. You learn to notice each filler as you say it, at first by listening to recordings of yourself and marking every “um”, then live, while speaking.
- A competing response. Each time you feel an “um” coming, you close your mouth and pause silently instead.
In the early studies, speakers’ filler rates dropped close to zero after one short session (USF thesis), and a later study using a brief awareness-plus-silent-pause routine reported that every participant cut disfluencies by more than half after one lesson (summary). Not every replication was as clean, and the studies were small, so treat the numbers as encouraging rather than settled.
You can run the same routine alone:
- Record yourself speaking for two minutes about something you know well.
- Count the fillers while listening back. Write down the number per minute.
- Speak again on a new topic, and every time you feel an “um” coming, stop and breathe instead.
- Record and count again. Repeat over a few days.
Other habits that reduce fillers
- Know your first sentence. Most fillers cluster at the start of a thought. If the first words are decided, the rest follows more easily.
- Use bullet points, not a script. Reading a script sounds flat; speaking without any structure produces long searches for the next point, filled with “um”. A few key words per section is the middle ground.
- Slow down slightly. Fillers appear when speech outruns planning. A slower pace gives your thinking time to keep up.
- Allow the silence. A pause of a second feels long to you and normal to the listener. Speakers who are comfortable with silence need fillers less.
- Watch “so”, “like” and “you know” too. They do the same job as “um” and often replace it once you have trained the sound away.
On camera, you do not have to fix it while recording
If you record video, there is a second option: say what you want to say, fillers and all, and remove the ones that hurt in the edit. That is not cheating. Every polished talking-head video you have watched was edited.
Trying hard not to say “um” while the camera runs often makes delivery worse: stiffer, slower, more self-conscious. Many creators record relaxed, pause instead of restarting when they lose the thread, and clean up afterwards. Our guide on how to talk to camera covers the recording side.
Do not remove every single one in the edit, though. A video with every filler stripped out can sound oddly breathless. Some fillers work as punctuation, signalling a turn or a thought being formed. Strike the ones that stall and keep the ones that carry; the reasoning is in filler words are not the enemy.
How to remove ums from a video
In a timeline editor, you find each “um” by listening, then cut it out by hand. Several desktop tools can detect and remove filler words automatically; how to remove filler words from a video compares them.
On a phone, slicecut makes them visible instead. It transcribes your take, so every “um” sits in the text where you said it. Tap the ones you want gone and the cut lands right before the next word; tap again to bring one back. You decide which stay. The pause dial then evens out the gaps, so the edit does not sound choppy.
The short version
- Fillers are signals that you are planning. Planning better reduces them.
- A few per minute do no harm. Aim for fewer, not zero.
- Train yourself to notice each one and pause instead.
- On video, record relaxed and remove the distracting ones in the edit.