mexca
Capture emotion expressions from video, audio, and text with a single pipeline.
A machine learning approach
We laugh many times every day, but not all laughs sound the same. In this project,we tried to understand whether howpeople laugh is different depending on what is making them laugh. For example, do we giggle at jokes but belly laugh when tickled? We measured the acoustics (e.g., pitch, loudness) of laughs produced in four kinds of real-life situations: being tickled, watching something funny, witnessing someone else’s misfortune, and verbal jokes. Machine learning analyses revealedthat being tickled-laughter is acousticallydifferent from laughs that we makein other kinds of situations. Further experimentsshowed that human listenerscan tell that people laughing when being tickled sounds different than other laughs. Laughs made when being tickled also sound less controlled than other laughs. This study shows that being tickledlaughter is a distinct kind of laughter, potentially reflecting its ancient evolutionary roots; many animals, including chimpanzees and rats, laugh when being tickled too. This project also highlights how new computational methods can help shed light on common everyday behaviors, like laughter.
Vocalisations of Disgust across Behavioural Contexts: A Computational Analysis
Multimodal Emotion Expression Capture Amsterdam
Capture emotion expressions from video, audio, and text with a single pipeline.