How to add subtitles to a video

The fastest way to add subtitles to a video is to let an app transcribe it and burn the captions into the picture. The alternatives are a desktop editor, YouTube’s automatic captions, or typing an SRT file by hand.

There are four realistic routes, and the right one depends less on the video than on where it is going and what you already have open.

First, decide: burned in or a separate file?

This choice comes before the method, and it drives everything after it.

Burned in means the text is drawn into the picture. It cannot be turned off, it looks identical everywhere, and it is visible immediately to someone scrolling with the sound off. This is what short-form video needs.

A separate file — SRT or VTT — is uploaded alongside the video. The viewer can toggle it, the platform can index and translate it, and assistive technology can read it. This is what YouTube and accessibility requirements want.

They are not really alternatives. On YouTube, do both.

Route 1: automatically, on your phone

Best when the video is already on your phone and going back out to social.

The app transcribes the audio, times each word, and lets you style the result before burning it in. Start to finish this is a couple of minutes, and no file ever moves to a computer.

What to check before exporting:

• The words. Automatic transcription is strong on clear speech and weak on names, slang and acronyms. Fix those; leave the rest.
• The position. Every platform draws its own interface over your video. Use a safe-zone guide so captions do not end up behind a username or a share button.
• The size. Four to six words on screen at once is the usual sweet spot — readable at arm’s length without covering the shot.

This is what Viten does. Free covers 5 captioned videos a month up to 60 seconds, and 5 exports a month at 720p with no watermark.

Route 2: in a desktop editor

Best when the video is already in a project on a computer.

Premiere, Final Cut and DaVinci Resolve all have speech-to-text built in now. You get a caption track you can style and burn in on export, or export as SRT.

Slower to get to a first result, but worth it when the video is long, has several speakers, or is part of a larger edit you are cutting anyway.

Route 3: let YouTube do it

Best when the video lives on YouTube and you only need accessibility and indexing.

YouTube generates automatic captions for uploads. They are free and reasonably accurate, but they are unstyled, off by default, and invisible to anyone who does not turn them on.

You can edit them in YouTube Studio, or upload your own SRT under Subtitles, which is more accurate and much faster than correcting theirs line by line.

For Shorts specifically, this is not enough on its own — a muted scroller will never see captions they have to enable. Burn them in as well.

Route 4: by hand

Best when you need exact control, or the audio is bad enough that transcription cannot help.

You write the SRT yourself in a text editor or a tool like Subtitle Edit or Aegisub. It is genuinely slow — several times the length of the video — and only worth it for short, high-stakes pieces, or for fixing a file that is nearly right.

The format is simple enough to hand-write:

1
00:00:01,000 --> 00:00:04,000
The first line of dialogue.

Which method should you pick?

Situation	Route
Phone video → TikTok, Reels, Shorts	Automatic, on your phone, burned in
Already editing on a desktop	Your editor’s speech-to-text
Long-form YouTube	Upload an SRT; burn in only if you want the look
YouTube Shorts	Both — burn in and upload the SRT
Accessibility requirement	A proper caption file, checked by a human
Bad audio, high stakes	By hand

What usually goes wrong?

Captioning before the final cut. Every trim afterwards throws off the timings from that point on. Caption last.

Ignoring the safe zone. Text that looked centred in the editor sits under the platform’s own UI once posted. Check against a guide.

Too many words on screen. A full sentence at once is unreadable at a glance and covers the shot. Fewer words, larger.

Trusting the transcript. It gets names and jargon wrong in ways that are obvious to your audience and invisible to you if you do not read it back.

Noisy audio. Clean the audio before transcribing. Accuracy tracks clarity closely, and it is much less work than correcting the output.