Fast likelihood computation for continuous-mixture densities using a tree-based nearest neighbor search.

Frank Seide

Browse the full Interspeech paper archive.