Tamil subtitle & caption generator

Tamil hardly stacks consonants. Where Devanagari would join two consonants into one shape, Tamil writes each in full and puts a dot, the pulli, over the first to show it carries no vowel, as in க்க. The result is a line that sits low and stretches sideways, so a caption of four Tamil words can need noticeably more width than four English ones.

Vowel signs wrap around the consonant instead of piling above and below it. ெ, ே and ை are written in front of the letter they follow in speech, ொ, ோ and ௌ split into two parts on either side of it, and ு and ூ merge with the consonant into a new shape, which is why கு, மு and ரு each look different. Tamil needs less extra line height than Hindi, but it needs horizontal room.

A template whose font is Latin only draws its Tamil in Baloo Thambi 2, a heavy rounded face. Choosing a different Tamil face leaves any English words in the template's own font. Among the Tamil options:

Tamil is strongly diglossic: the language of conversation and the language of books and news differ in pronunciation and in word forms. A speaker says வந்துட்டேன் where formal writing has வந்துவிட்டேன், both meaning I have come. A Tamil-script caption that follows the speech keeps the speaker's voice; one rewritten into the formal register reads like a newspaper. Choose one approach per video and correct stray forms to match it.

Romanised Tamil, often called Tanglish, is how many viewers type Tamil on their phones, and it has habits of its own: ழ is commonly written zh, as in Tamizh, and long vowels are marked inconsistently, so a word may turn up with a doubled vowel or without one. An English translation loses a distinction Tamil makes whenever someone says we: நாம் includes the listener and நாங்கள் leaves them out, and English we cannot show which was meant.

Tamil script uses one letter for sounds English keeps apart: க covers both k and g, and த, ட and ப each serve for two sounds in the same way. Names and English words are where that shows, so compare them with how the person spells them, and watch ஜ, ஷ and ஸ, which some writers use in borrowed words and others replace with native Tamil letters.

The three l-sounds ல, ள and ழ, the two r-sounds ர and ற, and the n-letters ந, ன and ண are easy to swap in a transcript, and a swap changes the word. Tamil has digits of its own, but everyday writing uses Western digits, so a transcript that mixes the two needs one form throughout.