Subtitle spec check

Check every cue for line length, on-screen duration and reading speed against platform subtitle specs, and list the ones that need attention.

Accepts .srt and .vtt up to 5 MB. Everything runs locally, and you can download a text report at the same time.

The file is never uploaded — everything runs locally in your browser

Want your subtitles to pass the first time?

The full version breaks sentences for the target language and controls line length as it generates the subtitles, so the burned-in video is ready to publish without rework.

Sign up free and get trial credits

FAQ

Where do these thresholds come from?

They are the conservative end of the YouTube and Netflix subtitle specs: Latin up to 42 characters per line and at most 2 lines; Chinese up to 18 characters per line; reading speed up to 20 characters per second for Latin and 9 for Chinese; each cue on screen for 1 to 7 seconds. These are industry-wide thresholds, not ones we invented.

What happens if the reading speed is too high?

The cue disappears before viewers can finish reading it, so it may as well not be there. This is the easiest item to overlook and the one that hurts completion the most — machine-translated subtitles especially tend to balloon in length because of literal, word-for-word output, and the reading speed goes over the limit with it.

Does the check modify my file?

No. It only reads the content and produces a new report; not a single character of the original file is touched. If you want to fix things, pair it with “Clean up an SRT file” to clear overlaps and duplicate lines, or edit the subtitle text directly.

Why do Chinese and English use different thresholds?

Chinese characters carry far more information per character than Latin letters, so the same character count reads much faster in Chinese — one Chinese character is roughly equal to two Latin characters of information. Judging Chinese by the 42-character English limit would let clearly overloaded subtitles pass, so the two are assessed separately.