Two ways to get the text, one review process
Importing an existing file saves transcription but inherits its errors and its timings. Generating from audio gives you timings that already match this edit but sends unfamiliar words through a recogniser. Either way the review pass is the same: names, numbers, then line breaks.
