Fast robust inverse transform speaker adapted training using diagonal transformations.

Hubert Jin, Spyridon Matsoukas, Richard M. Schwartz, Francis Kubala

Browse the full ICASSP paper archive.