DIHARD 3
Speaker diarization error rate (full)
speechstatic
Definition
- Version
- not in the notes
- Category
- speech
- Maintainer
- not in the notes
- Status
- static
- Home page
- not in the notes
Sources
- Top 7 Speaker Diarization APIs and Models in 2026 - pyannoteAI Speaker Intelligence and Diarization pyannote.ai https://www.pyannote.ai/blog/top-7-speaker-diarization-apis-and-models-in-2026
Scores
| Model | Score | Who ran it | Setting | View or population | Date | Source |
|---|---|---|---|---|---|---|
 pyannote Community-1 | 20.2 DER % | vendor-reported | full, no collar, overlap scored | – | 2026 | pyannote.ai |
Where this comes from
Scores from different settings, harnesses or dates are not directly comparable; the row says which. Sorted by score, highest first, as a reading aid, not a ranking.
To cite this page: Project AURION, AI Use-Cases Research & Innovation, "DIHARD 3", https://aurion.semai.id/benchmarks/dihard-3/, records of October 2026.