Every device with more (then only one) audio sources must have some delay when switching from one audio source to another.
Some of them just switch sources, other are cross-fading (mixing) in some period, usually 1-2 seconds.
This delay varies from device to device and on some devices it can be changed, on others can not.

On the other side, some programs have delay (pause) between signal for changing audio source and sending real message, giving enough time to that device to actually switch audio source, so complete message could be reproduced.
And some programs do not have this delay (pause) before actual audio message, so first part of message could be missing in that situation (like with NextGen).

I do not have unit to test it and find working solution, but I think that there must be some way to "instruct" NextGen to put some delay (pause) before every message, so this kind of devices could have enough time to switch audio source to navigation and reproduce complete message.

Maybe some versions of NextGen developed for specific car head unit or some other device do have some kind of command somewhere in configuration doing this - inserting delay/silence before message.

This is only my thinking about problem, and maybe it could give some idea to someone how to find general solution for this problem.