speaker identification, determining which character said each utterance in
literary text, benefits many downstream tasks. Most existing approaches use
expert-defined rules or rule-based features to directly approach this task, but
these approaches come with significant drawbacks, such