Kernelized Rényi distance for speaker recognition

TitleKernelized Rényi distance for speaker recognition
Publication TypeConference Papers
Year of Publication2010
AuthorsVasan Srinivasan B, Duraiswami R, Zotkin DN
Conference NameAcoustics Speech and Signal Processing (ICASSP), 2010 IEEE International Conference on
Date Published2010/03//
Keywords#x0301;nyi, approach;input, distance;reference, entropy;graphical, equipment;entropy;speaker, graphic, identification;speaker, processor;information, Re, recognition;, recognition;speaker, signals;kernelized, signals;speaker, theoretic, verification;computer

Speaker recognition systems classify a test signal as a speaker or an imposter by evaluating a matching score between input and reference signals. We propose a new information theoretic approach for computation of the matching score using the Re #x0301;nyi entropy. The proposed entropic distance, the Kernelized Re #x0301;nyi distance (KRD), is formulated in a non-parametric way and the resulting measure is efficiently evaluated in a parallelized fashion on a graphical processor. The distance is then adapted as a scoring function and its performance compared with other popular scoring approaches in a speaker identification and speaker verification framework.