Honest pacing
Cue durations come from word count at your chosen speaking rate: estimation, clearly labelled.
A transcript knows what was said; it does not know when. Real caption timing comes from the audio, which is why professional workflows align captions with speech-recognition tooling. What this converter offers is the honest middle path: statistically sound timing from speaking rate, ready in one paste, labelled as the estimate it is.
Most narrators speak between 140 and 170 words per minute. At a stable pace, generated cues can land within one or two seconds of the audio. Each timestamp in the source becomes a fixed anchor, and the converter adjusts the cues between anchors.
Caption guidelines use about 42 characters per line and two lines per cue. Line breaks should follow grammatical boundaries. The converter applies these rules while it builds the cues, then runs the same checks as the caption validator.
Use the output as a strong first draft: import it into your editor, nudge the cues that drift, and ship. For a file that must be frame-accurate without manual work, transcribe from audio instead: then come back here when all you have is text.