Techniques are described for Artificial Intelligence (AI) based deep fakes for a participant view in the meeting. A deep fake video may be built from poor quality and noisy/low-light video. A participant can eliminate others in the video that the participant did not intend to see but are in the camera field of view. Quality of video from poor quality cameras can also be eliminated. Moreover, a deep fake algorithm is described to build a high quality hyper realistic video from a low quality/no video scenario from the audio. This can be achieved by capturing only the face metadata on the meeting client using previously learned AI models and sending it to the cloud. The meeting server uses the AI based deep fake to create a high-resolution video from a low-resolution video using AI algorithms previously learned from previously shared videos/metadata from the user. The apparent location of the participant may also be changed using deep fakes (e.g., bedroom to work room, car to office room, etc.), in order to provide an optimal meeting experience for the remote participant.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.