Rapid entrainment to spontaneous speech: A comparison of oscillator models


Oscillator models may be used for modeling synchrony between gestures and speech, or timing of backchanneling and turn-taking in dialogues. We find support for the hypothesis that oscillator networks can better predict rhythmic events on the syllable and foot level than single oscillators, but we do not find support for the hypothesis that phase resetting oscillators perform better that phase adapting oscillators. Overall, oscillators can be used to predict rhythmic events in speech, but higher level information needs to be integrated into such models to reach a satisfactory performance.

Back to Table of Contents