Adding controlled amount of noise to improve recognition of compressed and spectrally distorted speech.

Jan Nouza, Petr Cerva, Jan Silovsk

Browse the full ICASSP paper archive.