Pretraining an encoder to identify multiple speakers from fully overlapped mixtures yields accurate local diarization without simulated conversational data.
The method is storage-friendly, simulation-agnostic, and outperformed diarization-based pre- training, with further gains from additional DIA pretraining
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
eess.AS 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Pretraining Multi-Speaker Identification for Neural Speaker Diarization
Pretraining an encoder to identify multiple speakers from fully overlapped mixtures yields accurate local diarization without simulated conversational data.