I think what Spridge is trying to ask, is which happens first - WAN QoS prioritization or VPN encapsulation.
If the VoIP traffic is encapsulated inside VPN packets, is the WAN interface still able to spot that its voice and prioritize it according?
I would also have to ask whether each voice packet is encapsulated (which would be adding a huge overhead)?
Sadly I don't get to play VPN.