An Automatic Child-Directed Speech Detector for the Study of Child Language Development.
Soroush Vosoughi, Deb Kumar Roy · 2012
In this paper, we present an automatic child-directed speech de-tection system to be used in the study of child language devel-opment. Child-directed speech (CDS) is speech that is directed by caregivers towards infants. It is not uncommon for corpora used in child language development studies to have a combina-tion of CDS and non-CDS. As the size of the corpora used in these studies grow, manual annotation of CDS becomes imprac-tical. Our automatic CDS detector addresses this issue. The focus of this paper is to propose and evaluate differ-ent sets of features for the detection of CDS, using several off-the-shelf classifiers. First, we look at the performance of a set of acoustic features. We continue by combining these acoustic features with several linguistic and eventually contextual fea-tures. Using the full set of features, our CDS detector was able to correctly identify CDS with an accuracy of.88 and F1 score of.87 using Naive Bayes. Index Terms: motherese, automatic, child-directed speech, infant-directed speech, adult-directed speech, prosody, lan-