Speaker identification using online, frame dependent, and diffusive variance adaptationScott Axelrod2002ICASSP 2002