Lucas Gonze is really on point with this thought:
Most of the doodads that people are dreaming up to help web audio and video to take off are beside the point. The challenge is making A/V more webby, not changing the web to accomodate A/V.
Going reductionist, if a human authorable URL scheme for specifying in/out points on continuous media became widely adopted, A/V would become first class on the Web.