Effect of speech bandwidth on speaker recognition performance using deep neural network (DNN).

Abstract

This work studies the general problem posed by the speaker recognition system in mobile environments with an effect of interest the bandwidth and speech coding. This is a comparative study on the performance of the speaker recognition system under different AMR Multi-Rate voice codecs and the bandwidths (NB narrowband, WB wideband and BEW wideband extension) based on the DNN approach.