Most people approach this backwards. They open the two-hour recording, start scrubbing from the beginning, and give up around the twenty-minute mark. The recording is not the problem — the order of operations is.
Here is the process that works, whether you do it by hand or with a tool.
1. Start from the transcript, not the timeline
Reading is roughly ten times faster than watching. A timestamped transcript lets you scan an hour in a few minutes and mark the places where something actually happened: a strong claim, a story, a number, a disagreement, a clear answer to a question people ask.
If your video is already published somewhere with captions, you have a transcript for free. If not, transcribe it — this is the step that makes everything after it fast.
2. Pick moments that stand alone
The test for a clip is simple: would this make sense to someone who has never heard of you and did not watch the previous ninety seconds?
Moments that pass are usually one of these:
- A complete answer to a specific question
- A story with a beginning and an end
- A claim stated plainly, then justified in one line
- A disagreement between two people, where both sides are audible
Moments that fail: anything that opens with “so as I was saying”, anything whose punchline depends on a slide you cannot see, and anything where the interesting part arrives forty seconds in.
3. Cut the first three seconds ruthlessly
The opening decides whether the rest is watched. If the clip starts with a throat-clear, a “yeah, absolutely”, or a name-check, delete it and start on the first strong word. It is almost always two or three seconds later than you think.
4. Choose a length by format, not by feel
| Length | Works for | Watch out |
|---|---|---|
| 15-25s | A single punchy claim, a hook, a one-liner | Too short for a story |
| 30-45s | The default. One idea with a small payoff | Loses people if the hook is slow |
| 45-60s | Stories, demonstrations, arguments | Needs a genuine reason to keep going at 20s |
| 60-90s | Trailers and highlight reels | Only when the material earns it |
If you cannot decide, make it shorter. Nobody complains that a clip ended too soon.
5. Frame for a vertical screen, not by cropping the middle
A centre crop of a 16:9 shot is the single most common reason auto-generated clips look amateurish: people sit off-centre, they move, and half a head leaves the frame. Three better options:
- Track the speaker. The crop follows the person talking and re-picks them when the camera cuts.
- Split the frame. Two speakers stacked vertically, each cropped to their own face. This is the format most interview clips use, and there is a reason.
- Fit and blur. Keep the whole original frame, fill the space above and below with a blurred copy. Necessary for screen shares and slides, where cropping destroys the content.
6. Burn in the captions
Most short-form is watched with the sound off. Uploaded subtitle files get ignored or stripped often enough that the reliable answer is to render captions into the video itself. Keep them large, high-contrast, and clear of the bottom of the frame where every platform draws its own interface.
7. Batch everything
The reason clipping dies as a habit is that people do it one clip at a time: cut, style, export, upload, repeat. Do it in batches instead — pick ten moments, style them all identically, render them all at once, and you have two weeks of posts from one afternoon.
This whole process is what Soft Clipper automates: it reads the transcript, proposes the moments, tracks the speaker while reframing to 9:16, and burns in captions across the entire batch. But the process is the valuable part — it works with any tool, including a timeline and patience.