I can imagine a semantic metadata format in audio documents where the values will be the entry points to a new kind of exploration.

Cheers