How to Add Subtitles to a Video

Two forces make captions mandatory rather than nice-to-have. First, silent autoplay: feeds on Instagram, TikTok, Facebook, and LinkedIn start videos muted, so the first few seconds have to communicate without sound or the viewer keeps scrolling. Captions are how a muted video earns a second of attention. Second, accessibility: captions are the difference between a video that everyone can follow and one that excludes anyone who can't hear it — and in many contexts (education, government, larger businesses) that's a legal requirement, not a courtesy.

There's a search and comprehension bonus too. Caption text is machine-readable, so it helps platforms understand and surface your video, and on-screen text measurably improves how much viewers retain — especially for jargon, names, and numbers that are easy to mishear.

There are exactly two delivery models, and choosing the wrong one is the most common caption mistake. Sidecar files are a separate subtitle file (SRT or VTT) that travels alongside the video; the player reads it and overlays the text, and the viewer can toggle it on or off. Burned-in captions are drawn permanently into the video pixels themselves — they're always visible and can never be turned off.

SRT (SubRip) is the most common subtitle format, and it's disarmingly simple — a plain text file you could type by hand. Each subtitle is a 'cue' with three parts: a sequence number, a start-and-end timestamp, and the line(s) of text. A blank line separates cues. A cue looks like this: a line with '1', then a line reading '00:00:01,000 --> 00:00:04,500', then 'Welcome back to the channel.' — followed by a blank line and cue '2'. Timestamps are hours:minutes:seconds,milliseconds, and the comma before the milliseconds is an SRT quirk worth remembering.

VTT (WebVTT) is a close cousin designed for the web: it starts with a 'WEBVTT' header, uses a period instead of a comma in timestamps, and supports a bit more styling. HTML5 video and most browsers prefer VTT; YouTube and desktop players happily take SRT. Because both are just text, you can open either in any text editor and fix a typo or nudge a timestamp without special software.

• YouTube — accepts uploaded SRT files directly and can also auto-generate captions you then correct. Sidecar is the norm here; viewers toggle captions with the CC button.
• Vimeo and most websites — sidecar SRT/VTT, the same toggleable model. On your own pages, HTML5 video reads a VTT track file.
• TikTok, Instagram Reels, YouTube Shorts — burned-in captions dominate. These platforms have their own auto-caption features, but creators overwhelmingly bake text into the video so it's always visible and survives re-uploads and downloads.
• LinkedIn and Facebook — accept SRT on native uploads, but because so much viewing is muted and re-shared, many creators burn in anyway for safety.

The pattern: horizontal, long-form, on-a-website video leans sidecar; vertical, short-form, social-feed video leans burned-in. When in doubt about a feed-based platform, burning in never hurts visibility.

Captioning by hand means playing a few seconds, typing what was said, marking the exact in and out timestamps, and repeating — for every line. A ten-minute video can take an hour or more, and the timing is fiddly to get right. This is why almost nobody types subtitles from scratch anymore.

AI transcription does the heavy lifting: a speech-to-text model listens to the audio and produces timed cues automatically, usually with word-level timestamps so the timing is already close. Modern models are strong on clear speech, but they're not perfect — they stumble on proper names, brand names, technical jargon, acronyms, and heavy background noise. That's exactly why the editable step matters.

FileMorf is built so your video doesn't have to leave your machine just to get captioned. The clever part: only the audio needs transcribing, and the audio is a fraction of the video's size.

1. 1

   Add your video — audio is extracted in the browser

   The tool pulls just the audio track out of your video locally, right in the browser. Your full video file stays on your device; only the small audio track goes anywhere.

2. 2

   Transcribe with AI

   The extracted audio is transcribed by an AI speech-to-text model at 1 credit per minute. A free account includes 6 credits per month, enough to caption several short clips, with more on paid plans.

3. 3

   Edit the caption text

   Review the generated cues and fix what the model missed — names, jargon, acronyms, numbers. The timestamps are already set from the word-level timing, so you're mostly correcting words, not re-timing.

4. 4

   Download SRT/VTT or preview as a native track

   Export a sidecar SRT or VTT file to upload alongside your video, and preview the captions as a native subtitle track right in the player to check them in context.

5. 5

   Optionally burn the captions in

   If you need always-on captions for a social feed, an optional server-side burn-in renders the text permanently into the video. Files processed on the server are auto-deleted after processing.