Persian text in subtitles: letters, digits and half-spaces

Persian subtitles can look right and still be wrong underneath. Four small things decide whether the text is searchable, readable and consistent. Here is each one, how to check it and how to fix it.

1. Persian letters, not Arabic ones

Two Persian letters have Arabic look-alikes with different character codes.

They come from keyboards and old encodings, such as Windows-1256, that only had the Arabic forms. In many fonts they look almost the same, but the Arabic yeh keeps two dots below at the end of a word and the Persian one does not. The cost is real: a search for کتاب will not find كتاب, and spell checkers flag the word. Check: search your file for ي and ك. Any hit is a candidate. Fix: replace them with ی and ک. The SRT fixer can do it for the whole file.

2. Digits: pick one style and keep it

Persian text can use Western digits (0 to 9), Persian digits (۰۱۲۳۴۵۶۷۸۹, U+06F0 to U+06F9), or the Arabic-Indic digits (٠١٢٣٤٥٦٧٨٩, U+0660 to U+0669). The Persian and Arabic-Indic sets look the same for most numbers but differ for four, five and six: ۴ ۵ ۶ in Persian, ٤ ٥ ٦ in Arabic. Mixing styles in one video looks careless. Persian subtitles usually use Persian digits, and Western digits are common in technical content. Choose one.

3. The half-space

Persian joins letters inside a word, and some words are written as two parts that sit close together without joining. The invisible character between them is the zero-width non-joiner, U+200C, called the half-space (نیم‌فاصله). It belongs in:

• verbs with the prefix می or نمی: می‌شود, نمی‌دانم;
• the plural ending ها: کتاب‌ها;
• many compound words and endings: خانه‌ام, دانش‌آموز.

With a full space the word falls into two words. With nothing, the parts join and the word looks wrong. Check: read a few lines with می verbs and ها plurals. If you see می شود with a wide gap or میشود joined, the half-space is missing. Persian keyboard layouts include a way to type it, and many text editors can insert U+200C from its code.

4. Punctuation and quotation marks

Persian has its own comma (،), semicolon (؛) and question mark (؟), and quotes are usually written with guillemets: «این». The full stop is the ordinary one. Because the full stop, the exclamation mark and the quotes are neutral characters, they can jump to the wrong end of a line in a renderer that starts left to right. See why the full stop lands on the wrong side.

A note on spoken and written Persian

Spoken Persian is not written Persian. People say می‌شه and write می‌شود. Subtitles for a casual video can follow the speech, and subtitles for a lecture can use the written form. What matters is that you choose, and keep it the same across the video.

The checklist

1. No ي or ك in the text.
2. One style of digits.
3. Half-spaces in می‌ verbs and ‌ها plurals.
4. Each line wrapped in a right-to-left embedding, or ending with a right-to-left mark.
5. A font that has the Persian letters.

When you dub into Persian with Dublio, the translator is asked for correct half-spaces and natural spoken Persian, the subtitles are laid out right to left, and you can read and edit every line before the voice is made. See dubbing into Persian.