The nice part is that the audio can be annotated within the page on which it originally lived rather than on some alternate page on the web
That would be ideal. Let’s figure out how to make it so. The Hypothesis client currently ignores fragment ids, but perhaps can make an exception in this important case.
I’m curious if the scheme may make putting all the smaller loose pieces together even easier, particularly for use within Hypothesis? and while keeping more of the original context in which the audio was found?