Aligning Speakers: Evaluating and Visualizing Text-based Diarization Using Efficient Multiple Sequence Alignment (Extended Version)

TL;DR

This paper introduces new evaluation metrics and an alignment algorithm for text-based speaker diarization, enabling more comprehensive error analysis and visualization to improve dialogue system development.

Contribution

It proposes two novel metrics for utterance- and word-level evaluation and a multiple sequence alignment algorithm supporting high-dimensional token alignment.

Findings

New metrics capture more error types in SD evaluation.

Alignment algorithm effectively handles multiple sequences and high-dimensional data.

Tools facilitate better analysis and visualization of diarization errors.

Abstract

This paper presents a novel evaluation approach to text-based speaker diarization (SD), tackling the limitations of traditional metrics that do not account for any contextual information in text. Two new metrics are proposed, Text-based Diarization Error Rate and Diarization F1, which perform utterance- and word-level evaluations by aligning tokens in reference and hypothesis transcripts. Our metrics encompass more types of errors compared to existing ones, allowing us to make a more comprehensive analysis in SD. To align tokens, a multiple sequence alignment algorithm is introduced that supports multiple sequences in the reference while handling high-dimensional alignment to the hypothesis using dynamic programming. Our work is packaged into two tools, align4d providing an API for our alignment algorithm and TranscribeView for visualizing and evaluating SD errors, which can greatly aid in the creation of high-quality data, fostering the advancement of dialogue systems.