How to Edit a Short Vertical Video of Someone Speaking to Keep Viewers Engaged
You've got a minute of someone talking straight into a phone. The take is fine, but if you post it as it is, a lot of people will scroll past before the second sentence. That's not because of what's being said. It's because of how a raw take plays: a slow start, gaps between the sentences, no captions, and the same picture the whole time.
You can fix all of that in Premiere Pro with a few shortcuts, the Text panel, and a second video track.
Start with the first sentence
Make a vertical sequence, 1080 by 1920, and drop the clip in. Then, before you touch anything else, find the first sentence worth hearing.
Almost every take starts with a warm-up, something like "Hi, so, today I want to talk about...", and nobody needs that. Play through until the person says something that would make a stranger curious, and cut everything before it. If the best opening line comes later in the take, it's fine to start there.
If you only fix one thing in the whole video, fix this.
Where to cut
Cut between thoughts, not in the middle of them. When someone stumbles and says a sentence twice, keep the second try. It's almost always the better one, because by then they know what they want to say.
Three shortcuts make this a lot faster. Ctrl+K (Cmd+K on a Mac) adds a cut wherever the playhead is. Put the playhead where the good part starts, press Q, and Premiere removes everything between the previous cut and the playhead and closes the gap. W does the same in the other direction, up to the next cut.
Watch the edges of words. Zoom into the waveform and cut a couple of frames before the first sound, not right on it, or the first letter gets clipped. And let the last sound of a word finish before you cut.
What to do with the pauses
People talking to a camera leave gaps everywhere, between sentences and while they think of the next word. You don't notice them while filming, but on playback they feel long, and that's usually when people scroll away.
Text-Based Editing is handy for finding them. Transcribe the clip in the Text panel, and Premiere marks the pauses in the transcript, so you can jump straight to the long ones instead of scrubbing through the whole clip. It even offers to delete all of them in one go.
Don't do that. If every gap goes to zero, the words run into each other, the breaths disappear, and it stops sounding like a real person talking. Shorten the pauses instead of deleting them, and leave a small breath between sentences. Keep a slightly longer pause in a couple of places on purpose: after a question, and right before the line the whole video is about.
Every cut inside one continuous shot makes the head jump a little. People are used to that on short videos, so you don't have to hide every jump. If one bothers you, cover it with b-roll rather than zooming in, because a picture that keeps changing size every few seconds is tiring to watch.
Captions
Lots of people watch with the sound off, so for the first few seconds the captions are the only way they know what's being said.
In the Text panel, use Create captions on the transcript. Before you click OK, set it to a single line and a short maximum length, so you get three to six words on screen at a time instead of a whole sentence. Then read every caption once. Speech-to-text still gets names, product names, and numbers wrong, and those are usually the words that matter most.
Style the whole caption track at once. Big, bold letters with a dark outline or a dark box behind them stay readable over a bright window or a white wall. Then think about where they sit. TikTok, Reels and Shorts put their own text along the bottom and a column of buttons down the right. Captions at the very bottom end up under the app's text, so a safer spot is a little below the middle, away from the right edge and not over the mouth.
B-roll
B-roll is anything that isn't the person talking, like the product, the screen they're describing, or a street. Put it on V2, above the talking clip, so the voice keeps going underneath.
Start it on the word it shows, not before. If they say "the invoice", the invoice appears on "invoice".
Keep each clip to one to three seconds, long enough to see what it is.
Come back to the face for the important lines: the main point, a personal moment, the ask at the end. People trust a face more than stock footage.
Don't cover more than about half the video. It's still a video of someone talking.
It doesn't always have to be full screen. Scale the b-roll down into a card in the top half, and the face stays visible underneath. And check the captions over every clip, because white letters that were fine over a dark shirt can disappear over a bright sky.
A headline for the first few seconds
Put one short line near the top of the frame for the first few seconds that says what the video is about, something like "What I'd check before signing a lease". Someone scrolling with the sound off reads it before they've heard a word and decides whether to stay. Make sure it doesn't sit on top of the captions.
Check it on a phone
Export at 1080 by 1920 and watch it on your phone twice. With the sound off, check that you can follow it from the captions alone. With the sound on, listen for chopped words and gaps that still feel long. Both are easy to miss on a big monitor.
If you'd rather not do it by hand
Full disclosure: I work on Cutroom, a tool that does this kind of edit for you. You upload one clip of up to three minutes and get back a vertical video with captions, b-roll, and a headline at the start, which you can write yourself or pick from a few it suggests. On its faster styles, Punch and Native, it also shortens the pauses. The calmer ones, Straight cut and Flow, keep the full take as it was said.
But nothing above needs a tool like that. Open on a strong first line, cut between thoughts, shorten the pauses without getting rid of them, and make the captions and b-roll easy to take in on a small screen.
Comments