Sometimes you do not need a converter, you need to change three words and nudge a cue. Installing a full subtitling suite for that is overkill. This SRT editor online gives every cue its own text box and its own start and end time, so you can see and fix exactly what is wrong. Open an existing SRT or basic VTT file, or start from an empty file and build subtitles line by line.
What this SRT editor online is good at
It is built for corrections rather than for authoring a feature film from scratch. Typos, mis-heard names, punctuation, cues that flash by too fast to read, a missing line, a duplicated line, a subtitle that should have been two — these are all a few seconds of work here. Because every cue is exposed as editable text with editable times, you never have to guess what the tool is doing to your file. What you type is what gets written.
Writing subtitles people can actually read
Two rules cover most readability problems. First, give a cue enough time on screen: roughly one second minimum, and longer for longer lines — a cue under 500 ms usually reads as a flicker. Second, keep lines short. Around 42 characters per line and no more than two lines per cue is the widely used convention, because anything more forces viewers to read instead of watch. If a sentence will not fit, split it across two cues at a natural clause break rather than cramming it into one.
A quick pass for auto-generated captions
Machine captions are usually accurate on common words and wrong in predictable places, so a fixed order of checks is faster than reading the file top to bottom. Start with proper nouns: names of people, companies, and places are what speech recognition gets wrong most, and they are what viewers notice. Next, add punctuation and sentence capitalisation, which many caption engines omit entirely. Then look for cues that are far too short — auto-captioners often emit one cue per phrase — and merge neighbouring fragments into complete sentences. Finally, scan for repeated words at cue boundaries, a common artefact where a phrase is emitted twice.