The example links in the original post are dead, just going on general recommendation here..
I find it useful to start by listening to each source individually, getting each to sound good and natural on its own first - balance, EQ, clean up, whatever. Do that before trying to combine things. Determine which will be the foundation of the recording. I tend to go with the basic acoustic room basis of the recording and build from that. For AUD pair + SBD feeds that will usually be the AUD pair as foundation. Listen to that for a while on its own after basic balance, EQ and clean up and ask yourself what might make it better. Determine if the other sources can provide that missing element or elements. Bring that other stuff in and up slowly and play with timing, balance, panning. Determine if the addition(s) is really making things better or not based primarily on what the foundation on its own was missing.. rather than listening primarily to the attributes of the new addition. This is something of a mental listening trick. Your brain will naturally be hyper focused on identifying and hearing the addition most clearly, almost ignoring the foundation to do so. Resist that and keep the foundation as the primary focus of attention, with the addition subtlety modifying it. If you decide the addition really isn't helping with what the foundation needs, leave it out. Don't feel bad about not using everything.
If the addition is helping, now go back and work on how the addition fits with the foundation. Alter things like EQ from each part being "good and natural on its own" to whatever works so that things fit together most beneficially. Mute the extra stuff regularly while listening, in various combinations if there is more than one additional source, to make sure the additions are doing what is needed. Double check how much level you really need from each so as not to over-compensate or try to pull out what just isn't there.
Difference between sources is good. That's the only way one of them might add what might be missing in the foundation, and the addition generally works best when each element can stand on its own in isolation first, prior to any special tailoring. Rather than trying to brute-force match the average spectrum of one to the other, its best to work to get them both sounding decent by ear as a starting point, then carefully work the combination from there.