For general information, on a recent switch to a new DMR server, an observation was made that DVSwitch does not seem to cope with replicated voice packets.
The effect of the replicated packets is that the outgoing audio stream to DVSM sounds slow.
Whilst it is not desirable to have this situation in the network, my understanding is that the G4KLX software does deal with this by looking at a sequence number in the packets and drops replicated packets and replaces them with silence.
The cause of the replication has been fixed on the network in question, but it highlighted that it is also desirable to drop the replicated packets at the client side if possible.
I am not sure if the feature to drop can only be provided for the homebrew DMR protocol.