As part of support for encrypted media, the RTP packetizer and depacketizer of senders and receivers have to be able to:

a) know that it's pointless to try to look inside the packet for codec-dependent data
b) know what rules to apply when packetizing and depacketizing the frames

The architecturally cleanest solution would be to ask the SDP negotiation machinery to negotiate a PT value to use with the encrypted and packetized frames, and have the ability to retrieve the resulting PT value so that when enqueueing a frame, the processor can set a payload type that elicits the expected behavior.