Title: Remarks on Optimal Scores for Speaker Recognition

Abstract: In this article, we first establish the theory of optimal scores for speaker recognition. Our analysis shows that the minimum Bayes risk (MBR) decisions for both the speaker identification and speaker verification tasks can be based on a normalized likelihood (NL). When the underlying generative model is a linear Gaussian, the NL score is mathematically equivalent to the PLDA likelihood ratio, and the empirical scores based on cosine distance and Euclidean distance can be seen as approximations of this linear Gaussian NL score under some conditions. We discuss a number of properties of the NL score and perform a simple simulation experiment to demonstrate the properties of the NL score.