One cue follows one generated line
TTS Lines measures the audio made for each line. The SRT starts a cue when that clip begins and ends it when the clip ends. The pause after a line shifts the next cue forward.
For a concrete check, make two short lines, set a one-second pause after the first, then export the full recording and SRT. The gap in the audio should sit between the two cues.
Change the text before the final export
If you rewrite a line, generate that line again so the clip duration matches the new words. Then export the full recording and SRT together. A stale clip is not a reliable timing source for revised text.
Check long sentences in your video editor. One caption per line can be too much text on screen at once, even when its start and end times are correct.
Retime after video edits
The SRT matches the recording exported by TTS Lines. If you cut silence, move clips, or change their speed later, the caption timings will no longer match. Move or split the cues in the same editor where you changed the audio.
SRT is a useful starting point for a voiceover timeline, not a promise of word-level synchronization.