It takes quite some time and effort to derive the most accurate transcription possible from the CVR.
Once finalised by the working team, the transcript may be matched up to the FDR traces using the synchronised time code.
I would also expect there to be a lot of tedious manual cross-checking between the CVR and FDR to ensure data consistency. For example, ensure that a particular sound, e.g. from a setting of some switch or control, corresponds to a change in the corresponding FDR trace at exactly the same time. If not, the data might be unreliable. Add the ACARS messages, and possible recordings from QAR and internal memories of the instruments and FADECs, and the worst case time required for all this consistency cross-checking may be quite long.
On the CVR,
I'd say the priority starts about 15-20 minutes prior, to see what the crew were discussing and doing as they approached the time where things began to go wrong.
Preferably even before. IIRC the CVR records the last 2 hours. That requirement is not there by coincidence. The Swiss cheese theory says each accident has multiple contributing factors, some of which may have occurred long before the !!!!! hit the f*n.
***
Can anyone provide a description of the parameters recorded by the A330 DFDR? That would be interesting to study while waiting for the analysis results.