Identifying Speakers in Dialogue Transcripts: A Text-based Approach Using Pretrained Language Models