Understanding when learning begins is critical for identifying the factors that shape both the developmental course and the function of information acquisition. Until recently, sufficient development of the neural substrates for any sort of vocal learning to begin in songbirds was thought to be reached well after hatching. New research shows that embryonic gene activation and the outcome of vocal learning can be modulated by sound exposure in ovo. We tested whether avian embryos across lineages differ in their auditory response strength and sound learning in ovo, which we studied in vocal learning (Maluridae, Geospizidae) and vocal non-learning (Phasianidae, Spheniscidae) taxa. While measuring heart rate in ovo, we exposed embryos to (1) conspecific or heterospecific vocalisations, to determine their response strength, and (2) conspecific vocalisations repeatedly, to quantify cardiac habituation, a form of non-associative learning. Response strength towards conspecific vocalisations was greater in two species with vocal production learning compared to two species without. Response patterns consistent with non-associative auditory learning occurred in all species. Our results demonstrate a capacity to perceive and learn to recognise sounds in ovo, as evidenced by habituation, even in species that were previously assumed to have little, if any, vocal production learning.