Guide6 September 2026·6 min read

How to add subtitles to a screen recording on a Mac

Most people watch the first few seconds of your demo with the sound off. Here is what macOS gives you (nothing), the three ways people actually add subtitles, and how to correct the words as text, where deleting a sentence cuts it out of the video.

Musthaq Ahamad

Building Prequel

Part of The complete guide to screen recording on Mac

Most people who open your screen recording will watch the first few seconds of it with the sound off. On a phone, in an office, in a feed. The video plays, nobody hears a word of your narration, and the decision to keep watching is made before they ever turn it on.

Subtitles are what you have in those seconds. They are also the difference between a demo somebody can follow on a train and one they close.

This is how to add them to a screen recording on a Mac: what macOS gives you on its own, the three ways people actually do it, and the way we built into Prequel, where the words are transcribed while you record and land in the editor as a paragraph you can type into.

What macOS gives you

Nothing, is the short answer.

QuickTime Player and the Cmd+Shift+5 toolbar record a screen and a microphone. Neither transcribes anything, and neither has any way to put a word on the frame. Finder shows you subtitles that are already inside a file; it does not make them.

So every route below is a third-party one. They differ in where the transcription happens, and in what you can do to the words afterwards.

The three ways people do it

Type them by hand in a video editor. Final Cut, Premiere, DaVinci Resolve and CapCut can all put text on a frame, and Resolve and Premiere will transcribe for you. This is the most control and by far the most work: a three-minute demo is a few hundred words, and every one of them has to land on the right frame.

Upload it somewhere. YouTube captions anything you upload. Descript, Veed and Kapwing transcribe on their servers and hand back a styled video. This is the least work, and the cost is that your unreleased product goes to a server you do not own, in exchange for a round trip you wait on.

Record with something that transcribes. The recorder already has the microphone track and the timeline, so the words can be made locally the moment you stop and dropped straight onto the take. No upload, no waiting, and the subtitles are part of the edit rather than a layer bolted on at the end.

That last one is what Prequel does, so the rest of this is how it works.

How to add subtitles in Prequel

1. Record with a microphone

Turn the microphone on before you record. The transcription is made from that track, so a take with no microphone has nothing to caption. The Captions section does not even appear in the editor for one.

2. Stop, and they are already there

The editor opens on the take you just made and starts transcribing it in the background. There is no button, because there is nothing to decide: it runs on your Mac, it costs nothing, and a button everyone presses every time is a step in the way of the thing they came for.

It uses Apple's own on-device speech recognition, in your Mac's language. Nothing is uploaded, and it works with the Wi-Fi off.

3. Pick a look

Open Captions in the inspector down the right-hand side.

Six looks. Subtitle is a plain pill; Highlight lights the word being spoken; Pop shows one large word at a time; Outline drops the plate and rings the letters instead; Band runs a bar the full width of the frame; and Blur in brings each word into focus on the moment it is said, over a line that fills up as you speak.

Under them: how big the words are, where they sit, how far from the edge, how many lines a caption may fill, and the colour a lit word takes. Every one of these belongs to the clip you have selected, so a caption can be large over an introduction and small over a dense middle without splitting the video in two.

4. Fix what it misheard

This is the part worth the words. Press Edit captions at the top of the section and the transcript opens as one paragraph.

Type into it. Correct a product name it spelled wrong, drop a filler word, add the word it missed. Every word you do not touch keeps the timing the recogniser gave it, so a correction never drags the rest of the sentence out of sync with your voice.

Reset puts the original wording back if you go too far. Your corrections live with the project, so re-transcribing never overwrites them and reopening the recording tomorrow still has them.

5. Delete a sentence to cut it

Select a stretch of the transcript and the timeline lights up under it. The band across the strip is the footage those words were spoken over.

Press delete, and that stretch leaves the video. Not the subtitle, the footage: the clip is cut where the sentence started and again where it ended, and everything after it closes up.

This is the fastest way we know to tighten a screen recording. Read the transcript, delete the false start, delete the sentence where you lost your thread, and watch the take get shorter. Cmd+Z puts any of it back.

Where the subtitles end up

Burned into the exported video. Prequel draws the captions into the frame, the same way it draws the zoom and the camera, so the file you hand over has them in it and nothing needs to be enabled by whoever watches it.

There is no .srt or .vtt file alongside. That is the right trade for a demo on a landing page or in a launch post, where a sidecar file is one more thing to lose and most players will not show it. It is the wrong trade if you need captions a viewer can switch off, or text that a search engine can read. For that, upload the video somewhere that generates its own.

Four things that make subtitles read better

Keep them to one line. Two lines of caption over a screen recording covers the thing you are pointing at. One line, and let it break more often.

Put them where your interface is not. The default sits them near the bottom edge, which is usually empty in a screen recording. If your app has a toolbar down there, move them to the top.

Do not caption every "um". Filler sounds are dropped automatically, and if one survives, delete the word in the transcript.

Read it before you export. On-device recognition is good and it is not perfect, and the words it gets wrong are almost always the ones that matter: your product's name, the feature you are launching. That is a thirty-second read of the transcript and it is the single highest-value thing on this page.

Is it worth it?

For anything going in front of a customer, yes. A demo watched on mute with subtitles is still a demo. Without them it is a silent screen recording of a product nobody can hear you explain.

Prequel records the screen, the camera and system audio on a Mac, then opens an editor where the zoom is already placed, the frame is already composed and the subtitles are already written. $29 once, or $9 a month.