The paper introduces a virtual audience framework for online conferences, addressing synchronization issues and providing real-time audience feedback without audio transmission. Future work includes using deep neural networks for local laughter detection and exploring advanced audio synthesis techniques for improved virtual audience sound.