Evaluating The Performance Of Using Speaker Diarization For Speech Separation Of In-Person Role-Play Dialogues