Fig. 4From: A simulation study on optimal scores for speaker recognitionThe EER and IDR results with homogeneous and non-homogeneous within-class distributions. The x-axis is the STD ω of the noise ξ added to the STD of the within-class distribution of each individual classBack to article page