The technical problem of generating a video feed that represents a user who is party to a video call in a manner that invokes the sense of visual presence of the caller without communicating the live video of the user is addressed by configuring a video calling system to include an avatar call platform. The avatar call platform is configured to generate and display a video of an animated figure that represents a caller during a call, in that the face of the animated figure moves in a way that matches what the caller is saying during the call.