Abstract: Systems and methods for audience interaction in real-time multimedia applications are provided. In one embodiment, a method for a real-time video conference comprises, during the real-time video conference, receiving a media stream including audio and video from a remote computing system over a network, acquiring, video and audio of a user, detecting an event with a machine learning model in at least one of the video and the audio of the user, and transmitting an event detection message indicating the detected event and/or audio and video of the user to the remote computing system over the network. In this way, natural audience reactions may be automatically detected and shared.