SDBench provides a reproducible 13-dataset benchmark for speaker diarization, and its companion SpeakerKit achieves a claimed 9.6x speedup over Pyannote v3.1 with comparable DER.
Figure 4 breaks down DER into 3 components: (i) Missed Detection, (ii) False Alarm, and (iii) Confusion
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SD 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
SDBench: A Comprehensive Benchmark Suite for Speaker Diarization
SDBench provides a reproducible 13-dataset benchmark for speaker diarization, and its companion SpeakerKit achieves a claimed 9.6x speedup over Pyannote v3.1 with comparable DER.