How to edit a talking-head video
By Anteras · Updated
A talking-head video is one person speaking to the camera. Editing one comes down to seven jobs, best done in this order: cut the pauses, remove the filler sounds, clean the sound, fit the frame to 9:16, add captions, add a hook for the first seconds, and add one visual accent. This guide covers each job, the setting to start with, and how to tell when you have gone too far.
The short version
| Job | Start with | Change it when |
|---|---|---|
| Cut pauses | Natural: pauses over 0.6 seconds | It still drags: go to Fast (0.4 s). It sounds breathless: go back |
| Remove filler sounds | On: um, uh, erm | A real word was heard as a filler: correct that word |
| Clean the sound | On | Your voice sounds thin: switch it off and compare |
| Frame to 9:16 | Centre crop if you are in the middle | Two people or a screen: use blurred sides |
| Captions | Three words at a time, middle of the frame | Your hands or a product are there: move them to the top |
| Hook title | The promise of the video in a few words | The first sentence already says it: leave it off |
| One accent | A word behind you at the start | The video is calm or serious: use none |
Before you edit: three things to get right while filming
Editing cannot fix everything. Three choices at recording time decide how good the result can be.
- Film vertically. A sideways video can be made vertical afterwards, but you lose the edges or get blurred bars.
- Get close to the microphone. A phone at arm's length in a quiet room beats any noise filter applied later.
- Leave space above your head. Text and cards placed behind you need somewhere to go. If your hair touches the top of the frame, there is no room.
1. Cut the pauses
Pauses are the biggest difference between a raw take and a finished one. Software finds them by listening for stretches where the sound drops to near silence for longer than a set time, then cuts them out and leaves a small margin so no word is clipped. In Anteras there are three paces: Natural cuts pauses over 0.6 seconds, Fast over 0.4 seconds and Extra fast over 0.25 seconds.
Start with Natural. Explanations and stories need a breath between sentences. Fast suits lists and tips. Extra fast gives the clipped rhythm of very short videos and gets tiring past a minute. The longer guide: how to remove silences from a video automatically. The tool: remove silences.
2. Remove the filler sounds
Um, uh and erm can be found in the transcript and cut like pauses. Be careful with anything beyond those sounds. Words such as “like” and “you know” are often part of the sentence, and cutting them automatically leaves holes in what you said. Anteras removes only the sounds, with a separate list for English, German, Spanish and French. See remove filler words and the guide on removing ums and filler words.
3. Clean the sound
Three steps, in this order: remove the low rumble (below about 80 Hz), turn down steady background noise such as a fan, then set the loudness. Around −14 LUFS is a common level for social video: loud enough next to other videos, without clipping. In Anteras this is one switch: clean audio.
Listen once with the cleaning on and once with it off. If your voice sounds thin or watery, the recording was too noisy for the filter, and the untouched sound is the better of the two.
4. Fit the frame to 9:16
Reels, TikTok and Shorts are vertical: 9:16, exported at 1080 by 1920. If you filmed sideways there are two ways to fit: a centre crop, which fills the frame and loses the edges, or blurred sides, which keeps the whole picture and fills the rest with a soft copy of it. Use the crop when you are in the middle and alone; use blurred sides for two people or a screen. See resize a video for Reels.
5. Add captions
Captions let people follow you with the sound off and give the eye something that moves. Three choices matter more than the font.
- How many words at once. Two to four. Fewer is punchier and harder to read at speed; more starts to look like subtitles on a film.
- Where they sit. The middle of the frame is safest: the apps lay their own caption and buttons over the bottom. Move them to the top if your hands or a product are in the way.
- How loud the style is. A bold style when the captions carry the video. A quiet one when your face or a graphic should.
Then read them through once. Names, brands and numbers are what speech recognition gets wrong, and one wrong name is what viewers comment on. In Anteras you tap the word and type the right one; its timing stays. See add captions and the guide to caption styles for Reels.
6. Add a hook for the first seconds
The first seconds decide whether someone stays. A hook title is a large line of text over the opening that says what the video will give the viewer: the question it answers, the mistake it fixes, the result it shows. Write it last, when you know what the video actually delivers. If your first spoken sentence already does this job, a title on top of it is noise; leave it off. More in how to write a hook.
7. Add one accent, not five
An accent is a single visual moment that makes the video look made rather than recorded. The ones that suit a talking video use a cut-out of you: a word placed behind your head, a card that lists your points as you say them, or the room dimmed so you stand out. Anteras allows up to three per video. One at the start is usually enough. See text behind you and the guide on putting text behind you in a video.
Why the order matters
Cuts change the length of the video, so everything timed to your words has to come after them. Add captions first and cut pauses afterwards, and the captions drift away from the speech. The same goes for a hook title or an effect set to appear at a certain second. Cut first, then frame, then put things on top.
In Anteras you do not have to keep the order yourself: you switch the jobs on in any order and they are applied in the right one.
Doing all of it in Anteras
- Upload the take. It is transcribed with the timing of every word.
- Switch on Cut pauses and Remove filler words, and pick a pace.
- Switch on Clean audio. If the video is landscape, choose blurred sides or centre crop.
- Pick a caption style, then check the Words list for wrong names.
- Add a hook title and, if you want one, an effect.
- Watch the free preview. Change what bothers you and preview again.
- Export in 1080p.
The free plan covers 3 exports a month, videos up to 3 minutes, 1080p, no watermark. Previews do not count as exports.
What this guide does not cover
Videos with several camera angles, B-roll, music beds and long interviews need a timeline editor, and Anteras is not one. The comparison pages say which tools do those jobs.
Questions
What is a talking-head video?
A video in which one person speaks directly to the camera, usually framed from the chest up. Most educational Reels, TikToks and Shorts are talking-head videos.
In what order should I edit a talking-head video?
Cut pauses and filler sounds first, then clean the sound and set the frame, then add captions, the hook title and any effect. Anything timed to your words has to come after the cuts.
How many words should captions show at once?
Two to four. Three is a good starting point for a talking video on a phone screen.
Do I need a script?
No, but know your first sentence and your last one. Pauses and filler sounds in the middle can be cut; a weak opening cannot be edited into a strong one.
Can I edit a talking-head video for free?
Yes. Anteras's free plan includes 3 exports a month, videos up to 3 minutes, 1080p, no watermark.