ReflectLive is a real-time, WebRTC-based communication system designed to help participants improve the way they present themselves and communicate to others. To accomplish this, the system analyzes a meeting’s audio and video streams to provide real-time feedback to participants about their own non-verbal communication behaviors.
Video teleconferencing allows people to connect and communicate across distance, but users are not always aware of their own non-verbal behaviors (such as how much they speak, interrupt others, look away from others, etc.). These types of behaviors can have a large influence on communication, and the limited view of speakers and their environment in video meetings makes these non-verbal behaviors particularly salient.
We designed and built ReflectLive, a system that analyzes the audio and video streams of a WebRTC-based video conference and provides real-time feedback (in the form dynamic data visualizations) about a speaker’s own speaking amount, eye gaze, interruptions, face position, and other non-verbal behaviors important for effective communication. ReflectLive runs in each user’s browser, maintaining privacy and reducing additional use of bandwidth. With ReflectLive, we investigate whether real-time feedback presented live to each user during the video meeting can help them adjust their communication behaviors to match their own expectations or goals, while at the same time minimizing distractions from the feedback itself.