They may be reception/transmission times or PCR times. But there is a recommendation that the PCR times be used, as these are piece-wise linear and the stts table then compacts sensibly. The big question is, do we have a packet structure that allows the documentation of a reception/transmission offset from the PCR time?
6.3.6Special issues for recording support?
de-hinting issues (error concealment)
6.3.7Track references
to associated media tracks (“de-hinting”, linking at authoring time)