WebRTC-QoE - A Dataset of Quality of Experience in Audio-Video Communications
This comprehensive dataset examines user satisfaction in WebRTC video calls under various network conditions. It captures subjective evaluations from 20 subjects across 15 different test conditions with controlled network impairments including delay (0/500/1000 ms), jitter (0/500 ms), and packet loss (0/15/30%).
Description
This comprehensive dataset examines user satisfaction in WebRTC video calls under various network conditions. It captures subjective evaluations from 20 subjects across 15 different test conditions with controlled network impairments including delay (0/500/1000 ms), jitter (0/500 ms), and packet loss (0/15/30%).
The 613.0 MB dataset combines objective metrics with subjective quality ratings using the Absolute Category Rating scale (1-5, Bad to Excellent), resulting in 300 ACR scores. The dataset includes four main components: subjective results (CSV, 4 KB), WebRTC internals session statistics (28.7 MB, 300 JSON files), facial expression features extracted using OpenFace (547 MB with gaze vectors, eye landmarks, and 35 action units), and speech features from the OpenSMILE toolkit (18.9 MB with 6,373 features per subject per condition).
This multimodal dataset provides a comprehensive perspective on conversational quality in WebRTC-based audiovisual telemeeting services, enabling research on the relationships between network conditions, facial expressions, speech characteristics, and perceived quality of experience. The research was conducted at the University of Cagliari and published in Computer Networks (2024).
Access
Openly available for download from Zenodo and IEEE DataPort
License
Creative Commons Attribution 4.0 International (CC-BY-4.0)