Fig. 3From: Masked multi-center angular margin loss for language recognitionExperimental results with different numbers of centers (K) under four different scale factors (r)Back to article page