So, the two hosts are in the same subnet and L2 domain, correct?
Can you ping from either paging host or even ping either paging host from the two switches? If, at least the latter, you might try pinging both paging hosts from both switches.
From what you've described so far, the fact you don't see any huge ping latency jump between switches, nor believe there's any issue with the paging host <> IE switch connections, nothing jumps out to account for the delay you describe the paging devices detect between themselves.
What @Mancunian suggested earlier, i.e. using QoS to insure other traffic doesn't delay the RTP traffic, is a good suggestion, whether such congestion is the problem or not. What he specifically suggested, for a QoS policy, should be accepted on most current Cisco ISRs, but from skimming the IE3300 Configuration Guide's QoS section, much of it appears unsupported (unsurprisingly for a switch). Something equivalent could be done if you knew the port numbers being used. (Again, insufficient information to say whether this would be a "fix" or not, but still, possibly worth doing.)
BTW, any idea what the bandwidth need is for this paging app?
