Identifying speaker roles in a streaming environment

An apparatus comprising:

• at least one processor;
• a memory coupled to the at least one processor;
• a stream analyzer residing in the memory and executed by the at least one processor that monitors multiple real-time transport protocol (RTP) streams associated with a conversation having speakers with unknown speaker roles;
• the stream analyzer further comprising a streams controller residing in the memory and executed by the at least one processor that runs diaritization on the RTP streams to identify speakers and runs transcription on the RTP streams to output text corresponding to speech of the speakers;
• wherein the streams controller uses the speakers identified by the diaritization and runs text analytics on the output text of the transcription to identify speakers in the RTP streams by their role;
• where there is only a single speaker on each RTP stream, the streams controller associates a speaker in each RTP stream with a synchronization source identifier (SSRC) and after the speakers are identified and associated with an SSRC, discontinues diaritization of the RTP streams to preserve computer resources while continuing to identify the speakers in the RTP streams by their associated SSRC and outputs the speaker role to provide additional services.