All of them! In remote proceedings, channel separation and the identification of all speakers is automatic. In hybrid and live environments, multi-speaker diarization is used to identify each of the speakers.