# The Second DIHARD Diarization Challenge: Dataset, task, and baselines

## TL;DR

The second DIHARD diarization challenge introduces diverse datasets and evaluation tracks to advance speaker diarization robustness across various recording conditions and domains, providing benchmarks and baseline systems.

## Contribution

It presents a new challenge framework with multiple tracks, datasets, and baseline systems to improve diarization methods across diverse real-world scenarios.

## Key findings

- Baseline systems established for speech enhancement, activity detection, and diarization.
- Diverse datasets from multiple sources to test robustness.
- Evaluation metrics and challenge design to benchmark progress.

## Abstract

This paper introduces the second DIHARD challenge, the second in a series of speaker diarization challenges intended to improve the robustness of diarization systems to variation in recording equipment, noise conditions, and conversational domain. The challenge comprises four tracks evaluating diarization performance under two input conditions (single channel vs. multi-channel) and two segmentation conditions (diarization from a reference speech segmentation vs. diarization from scratch). In order to prevent participants from overtuning to a particular combination of recording conditions and conversational domain, recordings are drawn from a variety of sources ranging from read audiobooks to meeting speech, to child language acquisition recordings, to dinner parties, to web video. We describe the task and metrics, challenge design, datasets, and baseline systems for speech enhancement, speech activity detection, and diarization.