It is not clear whether you have made experiments with the JMP2 connected or not. Another point worth mentioning is that according to the first schematic, you are apparently performing a speaker feedback path for the signal ground through the 100K resistors whereas it would be expected to be done to the chassis ground, instead (or rather, is it really necessary?). In addition, the microphone's reference track bellow J4 does not help much to equalize this potential among each device, it seems like floating, not connected to anything else; think of drawing a plane there connected to all devices referenced by signal ground. BTW, is the MIC cable shielded ?