Comparing maximum a posteriori vector quantization and Gaussian mixture models in speaker verification | IEEE Conference Publication | IEEE Xplore