A transcript can preserve the words spoken in a video without saying when each line should appear. A caption file needs that timing. If you paste an entire transcript into one caption, the text may be accurate while the viewing experience is almost unusable.
WebVTT is a plain-text format for timed tracks. Its cues associate text with intervals in the media. MDN’s WebVTT format guide describes the header, timestamps and cue structure. The following small example is invented for this article.
Read the two intervals first
WEBVTT
00:00:01.000 --> 00:00:03.500
The gate opens toward the garden.
00:00:04.000 --> 00:00:06.000
[bell rings]The first cue starts one second into the media and ends at three and a half seconds. The second begins at four seconds. The gap is part of the timing, not an instruction to join the two lines. The bracketed sound description is illustrative and should appear only if that sound actually matters in the real content.
Blank lines separate cues. The timestamp arrow is part of the file syntax. A text editor can save the file, but renaming an ordinary transcript with a .vtt extension does not supply the missing header and timing.
Test against the actual recording
Use a video player or publishing tool that supports the track, then play the relevant section at normal speed. Check whether the words appear as the speech happens, stay long enough to read and disappear when appropriate. Correct wording cannot compensate for captions that arrive after the speaker has moved on.
Now replay with the audio muted. Can you follow important speaker changes and relevant sounds? Do not add every background noise indiscriminately; the aim is access to meaningful information. A transcript and captions can complement each other because readers use them differently.
Keep editing and delivery connected
If you trim a section from the video after timing the captions, check the track again. Later cues may no longer line up. Store a note identifying the video version used for timing so a colleague does not pair the newest edit with an older caption file.
Finally, preview on a narrow screen. Long lines may wrap differently, and captions can cover important visual information. The small example above teaches the file’s structure; a complete accessible track still needs human review of wording, timing, speaker information and presentation. The finished result should help someone follow the video, not merely pass a file-extension check.
AI-generated editorial illustration; not a documentary photograph.