ON-CAMERA DELIVERY BLOG / 8 MIN READ

How do I stop saying um and like on camera?

Filler words are a thinking noise, not a flaw, and they shrink with two moves: a change in how you talk, and a change in how you cut. Here is the full method, plus how to actually count the ones you have left.

≈2 /mingood filler rate
0.5spause that scares you
2places to fix it
0–100craft score

By Thomas, founder of CutScore · Updated June 2026

DELIVERY CHECK · talking_head.mp4
A presenter speaking into a microphone on camera, the moment where filler words like um and like either creep in or stay out of a take.
CRAFT SCORE
FIXES ADVISED
how your talking reads on camera
Fillers high · 7.4 "um/like" per min00:14
Pace on target · 152 wpm
Long um before the point · trim with a cut01:32
The 30-second answer To stop saying um and like on camera, fix it in two places. While you talk: slow down, plan and rehearse your first line, and when you feel the urge to fill a gap, pause instead. Most fillers are panic about a half-second of silence, so let the silence happen. While you edit: cut the leftover ums and likes with quick jump cuts. A working target is under roughly two filler words a minute. If counting them by ear sounds tedious, that is one of the things CutScore measures in one pass.
WHERE THE UMS COME FROM

First, the kind thing: filler words are not a sign you are bad at this. An um is the sound your brain makes while it picks the next word. You do it in normal conversation all day and nobody flinches, because they are doing it too. The camera is the problem. It adds a small, dumb pressure that speeds your mouth up while your plan stays the same speed, and the gap between the two fills with noise.

I have shipped videos where I counted my own ums afterward and physically winced. The pattern was always the same. The takes where I improvised every sentence were a swamp of "um, so, like, basically." The takes where I knew my first line cold were clean. The fillers were never about confidence in some deep sense. They were about whether I had decided what to say before I opened my mouth.

So the real enemy is not the word um. It is improvising live while a red light watches you. Kill the improvising and most of the fillers leave with it. The rest you remove in the edit. That two-part split is the whole method, and the next sections are just the details.

THE METHOD

Fix it while you talk, then fix it in the edit.

Two columns, two jobs. The left side is everything you change before the camera rolls. The right side is what you do with the ums that survive. You need both.

MoveWhenWhy it kills fillers
Plan the first linebefore you talkMost ums cluster in the opening, when your plan is the thinnest.
Slow your pacebefore you talkA slower mouth gives your brain time, so it stops reaching for a placeholder.
Pause, do not fillwhile you talkA silent beat reads as confident; an um reads as unsure. Swap one for the other.
Talk in shorter sentenceswhile you talkLong, winding sentences run out of plan halfway, and that gap fills with "like."
Do a second takewhile you talkThe second pass of any line is almost always cleaner than the first.
Cut the leftoversin the editA jump cut removes an um in one keystroke, no reshoot required.
Tighten the dead airin the editLong silences before a point read like the um you just deleted. Trim them too.
Count what is leftin the editAim for under about two filler words a minute, then stop fiddling.
The one that does the mostPlan the first line. The opening is where your ums are densest, and it is the part viewers use to decide whether to stay. Learn the first two sentences by heart and the whole take starts on solid ground.
STOP COUNTING UMS BY EAR

Tallying fillers by hand across a ten-minute video is miserable and you will lose count. CutScore counts them for you, marks each one with a timestamp, and tells you your rate per minute.

Join the waitlist
FIXING IT WHILE YOU TALK

Four habits that stop the ums at the source.

1. Pause instead of filling the gap

This is the single most useful thing on the page, so I will be blunt. An um is what you say because silence feels unbearable. It is not. A silent beat reads as someone who is in control and lets the point land. The trick is that a pause feels enormous to you and tiny to everyone else: the half-second you are terrified of is about half as long as it feels from the inside. Practice stopping your mouth completely when you lose the thread. Just stop. The next word arrives, the camera does not mind, and you have replaced an um with the most professional sound there is, which is nothing.

2. Slow down and plan the opening

Fillers spike when your mouth outruns your plan, so do two things. Slow your speaking pace a notch, which buys your brain the half-second it was trying to fill with "um." And rehearse your first two sentences until you can say them in your sleep. The opening is where ums are densest and where viewers decide whether to stay, so a clean start is worth more than a clean middle. You do not need a script for the whole thing. You need to know exactly how you begin.

An audio console with faders and meters, the kind of setup where a spoken take gets reviewed and the filler words become obvious on the timeline.
On a timeline, every um shows up as a little bump you can cut. Photo: Tima Miroshnichenko / Pexels.

3. Speak in shorter sentences

Long, winding sentences are filler factories. You commit to a thought, run out of plan halfway through, and the gap fills with "like" and "you know" while you scramble to finish. Short sentences give you frequent, natural stopping points. Each full stop is a clean place to breathe and decide what comes next, instead of trailing off into a placeholder. As a bonus, short sentences are easier for the viewer to follow, which helps you keep people watching. Say one idea. Stop. Say the next one.

4. Just do a second take

The cheapest fix in the building. The first time you say any line, you are half writing it; the second time, you already know where it lands, so it comes out cleaner with fewer ums. When a take goes sideways, do not push through and "fix it in the edit." Stop, reset, and say it again. Two clean takes give you options. One messy take gives you a problem. This is not perfectionism. It is the difference between editing and excavating.

RATHER SEE IT THAN READ IT?

Here is a real CutScore coaching report for an everyday talking-head video: filler rate, pace, the hook and more, scored, with timestamps and the exact fixes.

See a sample report
FIXING IT IN THE EDIT

Cut the ums that made it through.

No matter how clean your delivery gets, some fillers survive. The good news is the edit can erase them faster than you can say them. Three moves, in order.

1
1-MIN FIXDELIVERY
Jump-cut the obvious ones
An isolated "um" between two sentences is the easiest edit there is. Select it, delete it, and let the jump cut close the gap. Viewers are completely used to jump cuts in talking-head video, so the result reads as tight rather than choppy.
How Use your editor's waveform: an um is a small, rounded bump you can spot by eye, then cut.
2
EDITPACING
Trim the dead air, not just the words
A long silence right before a point lands the same way an um does: it signals you are still deciding. Once you have cut the fillers, tighten the gaps that are left so the speech moves. Do not flatten every pause, though. The deliberate ones are doing work.
How Keep intentional beats, cut the accidental ones. See pacing.
3
QUICKQC
Count what is left and stop
Once your filler rate is under roughly two a minute, you are done. Chasing literal zero makes you sound like a press release, and clipping out every single "you know" can leave the audio sounding spliced and unnatural. A few fillers are human. That is the target, not perfection.
How Watch the cut once and tally, or let CutScore count and timestamp them for you.
HOW MANY IS TOO MANY

What counts as too many filler words?

There is no official standard for this, so here is a working line I use. Around two filler words per minute is invisible. At that rate the ums hide behind your meaning and the speech sounds natural and human. Climb toward eight or ten a minute and the balance flips: viewers start hearing the fillers instead of your point, and the video reads as nervous even when you are not.

The goal is not zero. Zero sounds like a robot reading a teleprompter, and audiences trust a real person more than a flawless one. What you are after is few enough that nobody is counting. If you have ever rewatched your own footage and physically cringed at the third "um, so, basically" in one breath, you already know which side of the line you are on. The fix is the method above, and the measurement is the easy part.

How CutScore reads your delivery CutScore is an AI video quality coach, and delivery is one of the things it measures. It transcribes your speech, counts the fillers, and reports your rate per minute alongside your filler words, pacing and the strength of your hook, each one timestamped so you can jump straight to the moment. You get one 0 to 100 score, the evidence behind it, and a prioritised list of fixes, before anyone else sees the video. It judges the craft of the video itself, so it sits next to a growth tool rather than competing with one. More on the method and the standards.
QUESTIONS

Frequently asked.

You stop in two places: while you talk and while you edit. When you speak, slow down, plan the first line, and replace the urge to fill silence with an actual pause. Most ums are panic about a half-second gap, so let the gap exist. Then in the edit, cut the leftover fillers with jump cuts. Aim for under roughly two filler words a minute and the change is obvious to a viewer.
Because your mouth is faster than your plan. A filler is the sound your brain makes while it decides what comes next, and a camera adds pressure that speeds the whole thing up. You also say um more when you have not rehearsed the opening or when you are reading badly written notes. It is a thinking noise, not a character flaw, and it shrinks the moment you stop improvising every sentence.
A pause wins every time. A silent beat reads as confident and gives the viewer a second to absorb what you said. An um reads as uncertainty. The catch is that a pause feels enormous to you and tiny to everyone else, so the silence you are scared of is about half as long as it feels. Practice stopping instead of filling and the ums fall away on their own.
There is no official rule, but a useful working line is around two filler words a minute. Below that, nobody notices and your speech sounds natural. Climb toward eight or ten a minute and viewers start hearing the ums instead of your point. The goal is not zero, which sounds robotic, but few enough that the words disappear behind the meaning.
EARLY ACCESS

Stop wondering how you come across.

CutScore counts your fillers, checks your pace, and tells you exactly what to fix in your delivery, with the evidence to back it up. Join the waitlist for early access.

Join the waitlist