Auditory-based signal processing for noise suppression and robust speech recognition