Audio/video files how it is going to get categorized

Challenges of handling unstructured data with 5G. So, how audio/video data with vary in size going to get mapped with features, to process models