Paste a timed SRT, WebVTT or transcript file and get clean readable text with cue numbers, timestamps and markup removed.
Subtitle files are built for machines, not reading: every few words are interrupted by a cue number and a pair of timecodes. This tool strips those lines, removes standalone timecodes at the start of transcript lines, and deletes HTML-style tags such as font or bold markup left in caption text. With paragraph joining on, caption fragments are merged and a new paragraph starts at each speaker label and after longer completed sentences, which turns a caption file into something that can be read, quoted or pasted into notes. Speaker labels in brackets or in the Name: form can be kept or removed. The cleaning is pattern-based, so a quick read-through is still worthwhile for files with unusual layouts.
Which formats does this accept? SRT, WebVTT, and plain timed transcripts where each line starts with a time such as 00:01:23 or [00:01:23]. Cue numbers, arrow timestamp lines, the WEBVTT header and NOTE, STYLE and REGION blocks are all removed.
What does joining into paragraphs do? Caption lines are usually fragments of one sentence. Joining merges them with spaces into flowing paragraphs, starting a new paragraph at each speaker label. Turning it off keeps one cleaned line per original caption line instead.
Are speaker labels always detected? Labels in the form [Speaker 1] or a short capitalised name followed by a colon at the start of a line are handled. Unusual label styles may survive the cleaning, in which case they can be removed with a quick find-and-delete in a text editor.