Probabilistic Linear Discriminant Analysis of i-Vector Posterior Distributions

Learn probabilistic linear discriminant analysis techniques applied to i-vector posterior distributions in this 37-minute seminar presented by Sandro Cumani from Brno University of Technology at Johns Hopkins University's Center for Language and Speech Processing. Explore advanced statistical methods for speaker recognition and speech processing, focusing on how probabilistic approaches can enhance the discriminative power of i-vector representations through posterior distribution analysis. Gain insights into the mathematical foundations of linear discriminant analysis in the context of modern speaker verification systems and understand how uncertainty modeling in i-vector space can improve classification performance in speech and language processing applications.