A combination of discriminative and maximum likelihood techniques for noise robust speech recognition.

Kari Laurila, Marcel Vasilache, Olli Viikki

Browse the full ICASSP paper archive.