Computing vocal entrainment: A signal-derived PCA-based quantification scheme with application to affect analysis in married couple interactions

► We quantify vocal entrainment with unsupervised computational framework. ► We propose to measure vocal entrainment using automatically extracted vocal cues. ► We utilize principal component analysis as the framework for computing entrainment. ► We demonstrate the validity of this quantification scheme using psychologically inspired statistical tests. ► We explore the usefulness of this quantification scheme in studying affective states in real distressed couples' interactions.