Preprint
The Second DIHARD Diarization Challenge: Dataset, task, and baselines
Neville Ryant et al., arXiv (Interspeech 2019 challenge paper), 2019
Read the originalRead on October 17, 2026
- What it is
- The paper that defines the second DIHARD diarization challenge: its recordings from many domains, its task, its scoring rules and its baseline systems. It explains how the challenge's scoring differs from the scoring used in most earlier diarization papers.
- How much weight it carries
- Written by the challenge organizers, who have a stake in the scoring design they describe. The scoring argument is their reasoning, which is widely cited, not a measured result.
Facts taken from this source
- Most earlier diarization papers scored a modified DER that leaves out speech within 250 milliseconds of a speaker boundary and all overlapped speech.
“speech within 250 ms of reference boundaries and overlapped speech are excluded from scoring” (Section 1, Introduction (performance metrics))
- The second DIHARD challenge used DER as its primary metric, without a collar and without excluding overlapping speech.
“DER is adopted as the primary evaluation metric, but without use of collars or exclusion of overlapping speech.” (Section 1, Introduction (performance metrics))