Skip to content

engineering

Caption timing for a script that is not Latin

Every subtitle tool assumes characters per second measured on English. Ge'ez packs a consonant and a vowel into one glyph, and the usual numbers are wrong by about half.

/7 min read/engineering

A subtitle line has to be on screen long enough to read and short enough to keep up. The industry rule of thumb is a character rate, and every number in circulation was measured on Latin text.

Ge'ez is a syllabary. One glyph carries a consonant and a vowel, so a line of Amharic has roughly half the glyph count of the English that says the same thing. Feed that into a Latin character rate and captions flash past.

What we measured

We timed real readers against real clips and landed on 8.5 glyphs per second as the comfortable rate for Amharic, with a minimum line duration that matters more than the rate does for short lines. Four assumptions we had started with turned out to be wrong, which is the normal outcome of measuring something instead of reasoning about it.

The same finding drives the measure in our design system: Amharic body text is set to a narrower column than English, because matching the character count of a Latin line produces something exhausting to read.

Two free hours of audio a month, and no card to start

Make one call, read what it cost in the response, and decide from there.