Multimodal Machine Learning Can Predict Videoconference Fluidity and Enjoyment

TL;DR

This paper demonstrates that multimodal machine learning models using audio, facial, and body motion data can accurately predict negative experiences and conversational issues in videoconferencing, aiding in improving user experience.

Contribution

It introduces a multimodal machine learning approach to predict videoconference fluidity and enjoyment, highlighting the effectiveness of audio-video features in identifying negative user experiences.

Findings

Models achieved ROC-AUC up to 0.87 on hold-out data.

Domain-general audio features are most critical for predictions.

Multimodal signals can identify rare negative experience moments.

Abstract

Videoconferencing is now a frequent mode of communication in both professional and informal settings, yet it often lacks the fluidity and enjoyment of in-person conversation. This study leverages multimodal machine learning to predict moments of negative experience in videoconferencing. We sampled thousands of short clips from the RoomReader corpus, extracting audio embeddings, facial actions, and body motion features to train models for identifying low conversational fluidity, low enjoyment, and classifying conversational events (backchanneling, interruption, or gap). Our best models achieved an ROC-AUC of up to 0.87 on hold-out videoconference sessions, with domain-general audio features proving most critical. This work demonstrates that multimodal audio-video signals can effectively predict high-level subjective conversational outcomes. In addition, this is a contribution to research on videoconferencing user experience by showing that multimodal machine learning can be used to identify rare moments of negative user experience for further study or mitigation.