Speaker identification (diarisation) in transcripts

Users recording meetings or interviews want each speaker automatically labelled (Speaker 1, Speaker 2) rather than a single block of text. Otter.ai users cite speaker separation accuracy as the top complaint.