Powerset multi-class cross entropy loss for neural speaker diarization.

Alexis Plaquet Hervé Bredin

Published in: CoRR (2023)

Keyphrases

multi class
cross entropy
speaker diarization
log likelihood
network architecture
speech recognition
support vector machine
neural network
language modeling
maximum likelihood
pairwise
bayesian information criterion
broadcast news
ranking functions
evaluation metrics
language model
feature selection
data sets