id	raw_text	normalized_text	speaker_id	split	gender	is_gold_transcript	accent