Fundamental frequency of child-directed speech using automatic speech recognition

It has long been recognized that speech directed to infants and toddlers - child-directed speech (CDS) or motherese - is characterized by a collection of features including hyperarticulation, a distinctive lexicon, reduced structural complexity, and increased fundamental frequency (f₀). This paper examines f₀ in nearly 500 hours of recordings from 33 families, showing that mothers, but not fathers, distinguish CDS with consistently higher f₀. We also found that, compared with parents of typically-developing children, parents of children who are hard-of-hearing do not differ in production of f₀ to their children. Results are relevant for improved ASR application and better understanding of mothers' and fathers' speech to their children.