like others have pointed out, this makes automatic detection easier not harder if true. Just cut the segments that move around, audio analysis and even transcripts (they don’t have to be good! just good enough to identify missing segment!) are pretty mature.