Most short-form video is a person talking to a camera. No B-roll, no crew, one take on a phone. The footage is rarely the problem. The editing is: a two-minute take full of pauses, restarts and filler words becomes a one-minute reel only after someone cuts it.
This guide is the editing pass we run on every talking-head clip, in order.
1. Cut the pauses first
Silence is the first thing viewers feel and the last thing they can name. A half-second pause between sentences is natural in conversation and fatal in a feed, where the next video is one thumb away.
Cut every pause longer than a few hundred milliseconds, but cut on word boundaries. A cut that lands mid-syllable clips a consonant and sounds broken even when the viewer cannot say why. Leave a little padding — around 80 milliseconds — on each side of speech so words breathe.
2. Remove the restarts and filler
After the silences go, listen for:
- Restarts: "So the — so the thing is…". Keep the clean take.
- Filler words: "um", "like", "you know". Remove most, not all; a few keep you sounding human.
- Throat clears and off-camera asides.
This is where a transcript helps. Reading the take as text, you see the repeated sentence instantly; scrubbing the waveform, you hunt for it.
3. Write the hook as a title
The first second decides whether anyone sees the second. Put the promise of the video on screen as a title on the first frame — not the topic ("Productivity tips") but the payoff ("The one thing that changed how I edit"). Keep it on screen for under three seconds.
4. Caption every word
Many people watch with the sound off, and even those with sound read along. Word-by-word captions, a few words at a time, centred or in the lower third, keep the eye on the video. Highlight the one word in each phrase that carries the meaning.
Our guide to captions that keep people watching goes deeper.
5. Add motion where the words land
A static frame for sixty seconds feels long. Add a slow punch-in zoom on the sentence that matters, a cut to a tighter frame at a new section, a title at each step of a list. Motion should follow meaning: if you cannot say why a zoom is there, remove it.
6. Music last, and quietly
Background music sets the mood but must never compete with speech. Keep it low under the voice — well below it — and let it rise only where nobody is talking.
7. Export for the platform
Vertical 9:16 at 1080×1920 and 30 frames per second works for Reels, TikTok and Shorts. Keep important text away from the bottom and right edges, where each app draws its own buttons.
Frequently asked questions
How long should a talking-head reel be?
As long as the idea needs and no longer. Most land between 20 and 60 seconds once pauses and restarts are cut; a two-minute take often becomes a one-minute reel.
Should I remove every filler word?
No. Remove most of them, but a few natural pauses and "so"s keep the delivery human. Cut anything that repeats or restarts a sentence.
Do I need captions if my audience watches with sound?
Yes. Captions help people follow along, keep attention on the video and make the reel accessible to viewers who are deaf or hard of hearing.