Subtitle Editor

Good subtitles are invisible: viewers absorb them without noticing the craft behind them. Getting there is mostly about timing and restraint. A caption should appear the moment speech starts and disappear on the natural pause — never linger into silence, and never vanish before the viewer has finished reading. As a rule of thumb, keep each cue on screen between 1 and 7 seconds. Shorter than a second and nobody can read it; longer than 7 seconds and viewers re-read it, which breaks immersion.

Reading speed: the CPS limit

Professional subtitlers measure density in characters per second (CPS). Most adult viewers read comfortably at 17 to 20 characters per second, and children's content should stay closer to 12–15. This editor flags any cue that exceeds 20 CPS so you can split it in two. When a line of dialogue is too dense, don't just stretch the timing — split the cue at a natural phrase break and give each half its own time slot. Two short cues always beat one crammed cue.

Line length and line breaks

Keep every line under 42 characters and never use more than two lines per cue. Break lines at phrase boundaries ("I went to the store / to buy some milk") rather than mid-phrase, and keep the top line shorter than the bottom one when possible — it forms a pyramid shape that the eye scans fastest. Avoid hyphenating words across lines, and put one speaker per cue; if two people talk in the same time slot, use a dash for the second speaker.

SRT vs VTT: which should you export?

SRT (SubRip) is the universal workhorse: YouTube, Vimeo, VLC, Premiere Pro and DaVinci Resolve all accept it, and its format is dead simple — a number, a time range, and text. VTT (WebVTT) is the web-native format, designed for the HTML5 <track> element, and it supports extras SRT can't do: text styling, cue positioning and chapter metadata. If you're uploading to YouTube or a desktop editor, export SRT. If you're embedding captions on your own website, export VTT. Both are plain text, both are generated locally here, and you can export both from the same cues with one click each.

A practical workflow

Transcribe first, time second. Get all the words down without worrying about timing, then play your video and set in/out points against the actual speech. If the whole file drifts — for example the video has a few seconds of intro the captions don't account for — use the "shift all" control instead of fixing cues one by one. Finally, run the preview end to end: watch for captions that flash by too fast, overlap each other, or sit on screen during silence. Fix those, export, and upload with confidence.