ABSTRACT Infants prefer infant‐directed (ID) speech. Concerning talking faces, previous research showed that 3‐ and 5‐month‐olds prefer faces that produce native ID than native adult‐directed (AD) speech, regardless of background speech being ID, AD or silent. Here, we explored whether infants also show a preference for non‐native ID speech. We presented 3‐ and 6‐month‐old infants with pairs of talking faces, one producing non‐native ID speech and the other non‐native AD speech, either in silence (Experiment 1) or accompanied by non‐native ID or AD background speech (Experiment 2). Results from Experiment 1 showed an overall preference for the silent ID talking faces across both age groups, suggesting a reliance on cross‐linguistic, potentially universal cues for this preference. However, Experiment 2 showed that preference for ID faces was disrupted at 3 months when auditory speech was present (ID or AD). At 6 months, infants maintained a preference for ID talking faces, but only when accompanied by ID speech. These findings show that auditory non‐native speech interferes with infants' processing of ID talking faces. They also suggest that by 6 months, infants start associating ID features from faces and voices irrespective of language familiarity, suggesting that infants' ID preference may be universal and amodal.