For most web and social delivery, supply an SRT file as the simple, widely accepted default, and a WebVTT (VTT) file when you need positioning or styling on the web. Use burned-in captions only when the platform gives you no caption support or you need exact, guaranteed styling. Broadcast and some streaming deliveries use other formats, so always confirm the specification with the platform or distributor before you export.
The formats at a glance
| Format | Extension | Styling | Positioning | Typical use |
|---|---|---|---|---|
| SubRip | .srt | None | None | Simple, widely accepted subtitle exchange |
| WebVTT | .vtt | Limited | Yes | HTML5 video, web players, some platforms |
| TTML / DFXP | .ttml, .dfxp | Yes | Yes | Streaming and professional workflows |
| SCC | .scc | Broadcast style | Broadcast style | CEA-608 broadcast closed captions |
| EBU-STL | .stl | Broadcast style | Broadcast style | European broadcast |
| Burned-in | none (in the picture) | Full | Full | Social video, guaranteed look |
The styling and positioning columns above follow YouTube’s help documentation on supported caption files, which groups formats into basic, advanced and broadcast categories (YouTube Help).
SRT
SubRip is the simplest format. YouTube lists it as a basic format with no styling or positioning, and it must be plain UTF-8. It holds a numbered cue, a start and end time, and text. Its strength is compatibility: most editors, players and platforms accept it. Its weakness is control: you cannot reliably move a caption away from a lower third, or set colours.
Good for: social uploads, translations delivered as separate files, review copies.
WebVTT
WebVTT is a W3C text format for time-aligned data, mainly captions and subtitles. It is referenced from the HTML track element for video on the web. The specification says a file begins with the string WEBVTT, uses the text/vtt MIME type and UTF-8 encoding, and writes timestamps as hours (when needed), minutes, seconds and three-digit milliseconds, for example 00:11.000 --> 00:13.000 (W3C). Cue settings can set vertical text, line, position, size, alignment and regions. CSS can style cues using the ::cue selector.
Note that YouTube’s documentation describes its WebVTT support as having limited styling (bold, italic, underline) with positioning, and says class-based styling is not yet standardised on that platform. In other words, what the specification allows and what a given platform renders can differ.
Good for: your own website, accessible web video, cases where position matters.
TTML and DFXP
TTML supports richer styling and positioning and appears in professional and streaming workflows. YouTube lists TTML and DFXP as advanced formats and notes partial implementation. Delivery specifications for streaming services can be strict, so follow the provided document rather than a generic export.
Broadcast formats
SCC, EBU-STL and similar files are used for television-style captions. YouTube says it displays captions from these formats with the styling, colour and positioning they would have on TV. If a broadcaster asks for one, ask for their current technical specification and a validation tool.
Burned-in captions
Burned-in (open) captions are part of the video image. They always display, look exactly as designed, and survive platform changes. Trade-offs:
- Viewers cannot turn them off, change size or switch language.
- They are not machine-readable text, so search, translation and assistive technology cannot use them.
- A new language means a new render of the video.
- Placement must respect each platform’s interface overlays and aspect ratio. See reframing video for vertical, square and widescreen.
A common approach is to deliver both: a clean master with no captions plus separate caption files, and a burned-in version for platforms that need it.
Choosing a format: a short decision guide
- Is there a platform or broadcaster specification? Follow it exactly.
- Is it web video you control? Supply WebVTT (and SRT as a fallback).
- Is it a social platform with caption upload? Check the platform’s help page, and usually supply SRT.
- Do you need exact styling regardless of platform? Burn in captions, and keep a clean master.
- Do you need several languages? Keep separate caption files per language plus a clean master.
Delivery checklist
- UTF-8 encoding, especially for Chinese and other non-Latin scripts
- Timecodes that match the final edit, not an earlier cut
- Consistent frame rate between video and caption timing where relevant
- Reasonable line length and reading speed, checked on a phone
- Speaker labels and sound descriptions where captions are for accessibility
- File names that identify language and version
- A test play in the real player or platform before sign-off
For accessibility and reach, see captions and subtitles for video. For multilingual projects, see adapting video campaigns for different markets and AI lip-sync and dubbing.
Limits
Platform support changes. The details above reflect the sources listed at the time of writing (November 2026). Always confirm current requirements on the platform’s own help pages.
Where Radiance Impact fits
Radiance Impact supports localisation and adaptation of content for new markets, including preparing video for English and Chinese audiences. See localisation or contact us.
