YouTube has not published a specification of SBV. This page describes the files YouTube writes, as later tools document them. There is no signature or header; a file starts with its first caption.
One caption
| Line | Content | Notes |
|---|---|---|
| 1 | start,end | Two times separated by a comma, with no spaces. |
| 2 onward | Caption text | One or more lines. It ends at the first blank line. |
| last | A blank line | Separates this caption from the next. |
Unlike SRT there is no counter line, no --> and no styling or positioning.
Time format
A time is written H:MM:SS.mmm.
| Part | Digits | Meaning |
|---|---|---|
H | 1 or more | Hours, not zero padded |
MM | 2 | Minutes |
SS | 2 | Seconds |
.mmm | 3 | A full stop and milliseconds |
Worked example
0:00:01.400,0:00:06.770
Where are we going?
0:00:07.000,0:00:09.250
Somewhere
we have never been.
Not confirmed
Tools differ on minor points. One source adds a caption number before each time line, which does not match the files YouTube writes. The encoding of a file is not stated in it; downloads from YouTube Studio are normally UTF-8. A limit of two text lines is mentioned by converter sites, but is not published by YouTube.