SRT ↔ VTT Subtitle Converter

SRT and WebVTT describe the same thing — a list of cues, each with a start time, an end time and some text — and differ in details that are easy to get wrong by hand. SRT separates milliseconds with a comma and numbers every cue; VTT uses a full stop, opens with a WEBVTT line, makes the hour field optional and allows a named identifier before the timestamp. This converter auto-detects which one you pasted and writes the other, always emitting a full HH:MM:SS timestamp so the output is accepted by the strictest parsers.

Fractional seconds are normalised on the way in: a timestamp written 00:01:02.5 is read as 500 milliseconds, not 5, because the fraction is padded to three digits rather than parsed as an integer. Getting that backwards silently shifts every cue in a file by up to a second, which is the sort of error you only notice when the dialogue lands on the wrong shot.

What does not survive is everything VTT has that SRT has no place for. STYLE, NOTE and REGION blocks are skipped rather than carried across. Cue settings that follow the end timestamp — line:, align:, position: — are dropped, so a caption deliberately placed at the top of the frame to avoid burned-in text comes back centred at the bottom. Cue identifiers are read and not written out. Inline VTT markup can be kept or stripped with the toggle, which matters because <c.classname> and <v Speaker> mean nothing to most SRT players and will show up as literal angle brackets on screen if you leave them in.

One thing to check outside this tool: the file's encoding. WebVTT must be UTF-8, while SRT files in circulation are frequently Windows-1252, Shift-JIS or Big5 — a decades-old convention that no header records. If accented or non-Latin characters arrive as mojibake, the damage happened when the file was decoded, before any conversion, and re-saving as VTT will faithfully preserve the wrong characters. Everything here runs in the browser, and the result downloads as a .srt or .vtt file.