Cutting a talking-head video on a timeline is slow. You scrub back and forth, listen for the start of a sentence, place a cut, listen again, and nudge it by a few frames. Most of the time you already know what you want to remove: a false start, a long pause, an “um” in the middle of a point.
Text-based editing lets you do that the way you would edit a document. You read the transcript, delete the words you don’t want, and the video is cut to match.
How it works
When capshak transcribes your video, every word gets its own timing. That is what makes text-based editing possible: each word in the transcript is tied to an exact moment in the video.
So when you select words in the transcript and delete them, capshak knows exactly which part of the video those words cover and cuts it out. The cut is padded slightly into the silence around the words, so the speaker isn’t clipped mid-syllable.
Three things in the transcript make cleanup fast:
- Pauses show as pills. Long silences appear between words as small pills. Click one to cut that pause.
- Filler words are underlined. Words like “um” and “uh” are marked, and you can cut all of them at once.
- Cut words stay visible. Deleted words are crossed out rather than gone, so you can see what you removed and restore it later.
Nothing is lost
Cutting in the transcript never deletes anything from your source video. If you cut too much, select the crossed-out words and restore them. The video comes back, and so do the captions for those words.
Every edit can also be undone and redone the usual way, so you can try a cut, watch it, and change your mind.
A step-by-step workflow for a talking-head clip
Here is a simple routine for a one- or two-minute clip where someone talks to the camera.
1. Upload and transcribe
Create a project, add your video and generate captions. Transcription is included in every plan. If the speaker uses names or technical terms, run AI proofreading first so the transcript is easier to read. You can give it a hint with those names.
2. Read it once, without cutting
Open the transcript and read it through. Mark in your head what the video is really about. This is the moment to decide what the opening line should be, because the first seconds matter most in short-form video.
3. Cut the big pieces first
Delete whole sentences that don’t serve the point: false starts, repeated explanations, tangents. Work from the top down. Big cuts first, details later.
4. Clean up pauses
Look for the pause pills. Cut the long ones that slow the pace. Leave a few natural breaths in, or the speaker starts to sound rushed.
5. Remove filler words
Cut all filler words at once, then skim the result. In most cases it sounds tighter. If one sounds unnatural without its “um”, restore just that one.
6. Watch it back
Play the edit from the start. Text editing is fast, but your ears are still the final judge. If a cut feels abrupt, restore a word or two around it.
7. Polish on the timeline
The transcript gets you a clean edit. The timeline is where you shape it:
- trim or split clips where you need a precise cut,
- speed up a slow section (0.25× to 4×, pitch kept) or hold a frame with freeze frame,
- add B-roll or pictures on overlay tracks,
- add music; it ducks automatically under speech,
- add text overlays, effects and transitions.
8. Style the captions and export
Pick a caption style, adjust the font and animation, reframe to 9:16 if needed, and export your MP4. Export is free and has no watermark.
Tips for better results
- Record with pauses between thoughts. Clear gaps make cuts cleaner.
- Don’t over-tighten. Removing every breath can make a natural speaker sound robotic.
- Cut whole phrases. Cutting one word out of the middle of a phrase is more likely to sound jumpy than cutting the whole phrase.
- Fix the transcript early. Proofreading before you cut makes the text easier to scan.
Try text-based editing
The fastest way to understand it is to try it on one of your own clips. Create a free account to get started, or read more about text-based editing first.