Turkish altyazı files, timed to the word.

Turkish has four i letters, not two: İ and i with dots, I and ı without, and they are different letters with different sounds. Any tool that uppercases your caption file using English rules will turn ı into I and i into I as well, collapsing the distinction — which is how İstanbul becomes Istanbul and neden becomes something no one wrote. If you restyle captions to all-caps in an editor, check that it is applying Turkish casing rules rather than the default.

Agglutination is the other pressure. Turkish stacks suffixes onto a stem until a single word carries what English would spell out in a clause, and words of twenty-five characters or more turn up in ordinary speech. A word like that fills a caption line by itself and cannot be hyphenated in the middle without confusing the reader, so the cue is built around it.

• Vowel harmony means the suffix vowels change with the stem, so the same ending is spelled several ways across one video, all of them correct.
• The verb arrives at the end of the sentence, so a cue that stops early leaves the reader waiting for the action.
• Loanwords keep the circumflex in a few spellings, such as kâğıt, where it marks a different vowel.
• Soft g, the ğ, never starts a word and lengthens the vowel before it; it belongs in the cue text as written.

No line in the downloaded SRT or VTT is tagged with a name — both formats are cue text and timings only. Voices are separated in the transcript on screen, and the JSON export from the same run carries it as data.