Robust ASR based on clean speech models: an evaluation of missing data techniques for connected digit recognition in noise.

Jon Barker, Martin Cooke, Phil D. Green

Browse the full Interspeech paper archive.