Video Exports

How to Deliver Closed Captions as SRT or WebVTT

Export one caption timeline to SRT and WebVTT, then verify the words, timing and controls in the intended players.

Lock the video cut before exporting SRT and WebVTT from the same caption timeline. An added two-second slate delays the spoken lines by two seconds while the exported cues retain their old times.

Listen with the video as you edit the text. Capture meaningful sound effects and identification where speech alone leaves the viewer guessing. The W3C caption guidance explains why captions include non-speech information needed to follow the action. Use an audible door latch as a cue if it changes the scene, without adding every background sound in the room.

Export the same cue in both forms

YouTube lists plain UTF-8 SubRip among its basic caption formats. Its documentation also lists WebVTT with limited styling support. Plain cues make a practical shared baseline. In this original sample, a single spoken line and a sound cue begin at 4.2 seconds and end at 7.6 seconds.

SRT sample

1
00:00:04,200 --> 00:00:07,600
Sam: Ready to roll?
[Door clicks shut]

WebVTT sample

WEBVTT

00:00:04.200 --> 00:00:07.600
Sam: Ready to roll?
[Door clicks shut]

The WebVTT specification defines the opening header and decimal point in these timestamps. SRT uses a numeric cue index and a comma before milliseconds in the sample. Keep a blank line between successive cues. Produce both versions from the same edited timeline so future changes in spelling and timing do not have to be reconciled across two files.

For an HTML video player that uses native text tracks, link the WebVTT file as a caption track and provide its language. In an upload workflow, select the matching language for the SRT. If the captions stay blank when you expect a cue, start by toggling the track on. Play through 4.2 seconds in the sample and leave it running until the cue has cleared at 7.6 seconds. If it stays blank, check the request for the caption file and confirm that the player selected the language of the file you exported.

Watch the rendered captions

Compare the first spoken cue, a cue at a cut with overlapping sound, and the final caption on both players. Turn captions on and off and test on a small screen. For dialogue, check who is talking when it could be ambiguous. For sound descriptions, check that the event occurs while the cue is visible. Read accented names and punctuation as the player draws them. A text editor can show them correctly while a receiving player interprets the file differently. The file should carry the final cut's version and the language in its name, so a later export cannot be mistaken for the caption of an earlier edit.

Keep the language and cut on the caption timeline

Store the cue list beside the final edit and keep its language and intended end time with it. If the timeline has a slate at the start, decide whether that slate belongs in the copy you will upload. The sample cue beginning at 4.2 seconds needs to begin at 6.2 seconds if the new file starts two seconds earlier with an added slate and nothing else changes. Shift the shared caption timeline with the edit and export both formats again. If an insert changes only a later section, check timing on both sides of that cut instead of applying a single offset from the beginning.

Check the words while listening to the actual delivery file. A cue with a person's name should use the approved spelling. Keep speaker labels when they help identify a voice that is off camera. Describe a sound where missing it would make a reaction or change in the scene confusing. In the example, the door click explains a move that otherwise has no visible cue. Choose compact words that fit comfortably with the necessary dialogue.

Give each cue realistic reading time. Watch short and long lines at normal playback speed. Two overlapping speakers may require splitting or rephrasing the displayed text, according to the platform and captioning style you use. A title overlay or a player control can interfere with default caption placement on small screens. Verify the chosen style in the delivery player before assuming it renders the same way as your editor.

Reveal a format mistake with a small test

Open the SRT and VTT exports in a plain editor first. Check the identifying header for VTT and compare seconds and milliseconds for both copies. In the example, the punctuation between seconds and milliseconds is the only timing notation change. Check line breaks, the blank line between cues and any punctuation you need to preserve. When adding another line, keep it within the cue or create its own cue with the intended timing.

What you observe What to check next
There is no caption option Track attachment, its language and whether the file request succeeded
Captions start late by a stable amount Compare the cut, slate and first dialogue cue with the export timeline
Captions start on time and drift after an edit Find the first insert or trim that separated video and cue timing
Text appears but styling is absent Verify which cue styling the actual upload or player supports

Use a different check for a missing track and a blank cue. If you can toggle the intended language but the cue is blank at 4.2 seconds, inspect the file and its syntax. If you never see the language, check the configuration of the track itself. For a hosted player, confirm that the web request serves your current file and that it is reachable from the player's page. For an uploaded subtitle, check whether the platform lists the right file language and shows it after processing.

Deliver both forms with a matching name

Give the video, SRT and VTT a common revision label. Record which language is expected in each track and which file goes to which destination. Watch the first and last cues in both players after the export. Include a frame range around a fast edit where a small shift would be obvious. If your player changes the line breaks or colour, confirm the text remains readable at the intended size and setting.

If another editor handles distribution, have them use the delivery record without opening the edit project. They should find the right captions and select the expected language from the player. The final delivery folder guide shows where to put that manifest entry. When the cut changes, repeat a timing check from the new exported file, including the last cue where older timing problems tend to reveal themselves.

Sources