Abstract
Since recorded audio material is played, navigating relevantly through it is a key expectation. This paper provides a formalism to introduce exible navigation systems based
on sets of annotations applying to the same audio object. It
aims to build web interfaces to explore audio in time, robust
for large data-sets and long files. Introducing the concept of
weights applied to annotations, it specifies a parameterized
version of the functionality next/previous and presents an
effective implementation.