A recent study by Polka, Rvachew, and Molnar (2008) found that 6- to 8-month-old infants do not discriminate a simple native consonant-vowel contrast when familiarized to it in the presence of distraction noise (i.e., recordings of crickets and birds chirping), even when testing was conducted in quiet. Because the distraction noise did not overlap with the phonemes' frequencies, failure to encode the familiarization phoneme could be due more to a disruption in infant attention than to direct masking effects. Given that infants learn speech under natural conditions involving noise and distraction, it is important to identify factors that may 'protect' their speech perception under non-ideal listening conditions. The current study investigated three possible factors: speech register, face information, and speaker gender. Six-month-old infants watched a video of a female speaker producing a native phoneme in either an adult-directed or infant-directed manner accompanied by the same background noise as in Polka et al. (2008). After habituation, infants were tested with alternating trials of the familiar phoneme and a novel phoneme in quiet. Phoneme discrimination was measured by recording infants' heart rate and looking times during familiar and novel trials. Discrimination was poor in infants who viewed a female speaker using adult-directed speech but was significantly improved (as seen in both dependent measures of attention) when the female speaker used infant-directed speech. Results indicate that common factors in the typical environment of infants can promote speech perception abilities in noise.