Fig. 9From: Learning domain-heterogeneous speaker recognition systems with personalized continual federated learningComparing the embedding plot between the PFL and centralized trainingBack to article page