Speaker detection (also called diarization) automatically attributes speech to different speakers in your transcript, labelling each one separately (Speaker 1, Speaker 2, and so on) instead of producing a single undivided block of text.