User Scene With Privacy Preserving Component Replacements

In some implementations, the disclosed systems and methods can replace components descriptive of private real-world objects with generic components that obscure details of the private real-world objects, such as the objects' structure and semantic information. In some implementations, the disclosed systems and methods can convert image and/or sensor data into pose data that can be used by a recipient device to animate an avatar of the sender while the audio stream is being played.