The first weeks of life for preterm babies are critical for bonding and attachment. Early vocal contact, as a family based intervention, engages mothers to speak and to sing to their preterm infants in the incubators. This study tested the emotional and smiling content of the mother's speaking and singing voice in a context of early contact with her newborn preterm infant in the neonatal intensive care unit (NICU), and it examines which acoustic parameters are associated with emotion attributed to the mother's voice. The main hypothesis is that the emotional content of maternal speech and song, when directed to their preterm infants placed in incubators, is modulated by the infants' behaviors. Thirty-two blocks of vocal extracts in the presence and absence of preterm infants' displays and in the 5 s preceding each display were presented to 31 adult naïve listeners who were asked to rate the degree of emotion and smile in the mother's voice. The present results show that when infants open the eyes or smile, the maternal voice is perceived as more emotional and more smiling than it is in the absence of any facial display. This effect is particularly evident in the maternal speech. The maternal voice is rated as more smiling in the presence of the infant's smile than when preterm infants opened the eyes. The main acoustical features of the infant-directed voice—mean pitch and perceived sound pressure level—are positively associated with smiling attribution while only the mean pitch is associated with emotional intensity. These findings extends prior evidences by showing that maternal ID speech and songs, even in at-risk conditions such as prematurity, are not only related to preterm infant behavior, but also bear emotional content. Early face-to-face interactions between mothers and preterm infants need to be encouraged in the NICU and investigated, especially neural correlates, to assess the impact of emotional infant-directed voices on preterm infants' brain development.