NTT speaker diarization system for CHiME-7: multi-domain, multi-microphone End-to-end and vector clustering diarization.
Naohiro TawaraMarc DelcroixAtsushi AndoAtsunori OgawaPublished in: CoRR (2023)
Keyphrases
- speaker diarization
- end to end
- multi domain
- bayesian information criterion
- speech recognition
- cross domain
- clustering algorithm
- domain specific
- broadcast news
- congestion control
- k means
- model selection
- multi modal
- unsupervised learning
- selection criterion
- mixture model
- neural network
- gaussian mixture model
- feature vectors
- probabilistic model
- speaker verification
- decision trees