Analysis-by-synthesis feature estimation for robust automatic speech recognition using spectral masks.

Michael I. Mandel, Arun Narayanan

Browse the full ICASSP paper archive.