Create and export captions/subtitles

This episode needs burned-in English subtitles and a separate SRT file. Generate captions from the timeline for the sequence EP05_FINAL, proof the dialogue between 00:05:00 and 00:20:30 against the transcript, apply the broadcaster font and 42px size, export a timed SRT named EP05_EN.srt, and export an open-burned MP4 with captions visible. Check one-liners for overlap and correct line breaks.

Before we send to QC, make the captions easy to approve. Extract the subtitles into a side-by-side text list showing start/end times, speaker labels, and confidence hits where the auto-transcript differs from the transcript. Highlight instances longer than two lines, 42 characters per line, or overlapping closed captions. Suggest concise rewrites for any split mid-sentence.

I ran auto-captions on the documentary and two speakers swap names in a five-minute stretch. The producer will flag it. I don’t know if the issue is the reference transcript, poor audio, or the speech model. What’s the most likely cause and the fastest correction path that preserves timecodes and avoids redoing the whole caption file?

Every season I spend days fixing captions and resolving speaker IDs, which delays delivery and eats into QC time. I want a habit that stops late fixes and a preflight routine that reduces caption churn. What changes should I make in how we record, label speakers, and proof captions to cut turnaround by half?