I think I only want to use this technique on the SBD channels and just let the AUD be there a couple dB down for the ambience factor (the SBD mix was pretty good).
I suggest mixing the AUD and SBD to taste first, then applying parallel compression to that stereo mix to bring up the details and softer parts. Doing it that way keeps the process simple and also makes it much easier to hear what the parallel compression is doing to the final sound. It's often the best aproach regardless.
Of course it's possible to process only the SBD or only the AUD, but mentally juggling what effect processing one or the other alone before combining them will have is not as obvious as one may imagine. If you have the time and inclination, try it all three ways and make 3 mixes to compare: 1) EQ each source and mix 1st, then play with applying parallel comp to the resulting mix; 2) EQ each and parallel comp only the SBD before mixing the sources; 3) EQ each and parallel comp only the AUD, then mix. The difference between them may not be what you expect.
Actually if using
standard compression on one of the two sources, and not compressing the other, you end up with a form of parallel compression on the end product once mixed. By performing parallel compression on just one source before mixing you effectively parallel your
parallel compression. In otherwords, the relationship between things get complicated. It' might sound fine, but it gets harder to draw clear conclusions about why it does and exactly what is going on.
I think what I'm going for with the compression is during the quiet parts I want the compressed track to be higher levels than the original and during the loud parts I want the original uncompressed track to be higher. Overall so the highest levels aren't touched but the lowest levels are brought up a bit?
This is exactly what parallel compression applied to the stereo mix does.
Also I think I saw this in a previous post but should I also be limiting the compressed track?
Parallel compression is 'bottom up' and does nothing to the peaks of the uncompressed paralleled material. That's the whole point really. You may want to control the peaks of the final mix, or the components making up that mix, but that is a seperate issue.
What bdasilva is doing might be more accurately called 'parallel limiting'. Its attractive since he has extra tracks available on the recorder that would otherwise go unused. By doing that he eliminates the work in the DAW of setting up the compressor on the paralleled tracks, at the expense of flexibility in dialing in the compression characteristics. Because the compression applied with the parallel technique needs to be pretty heavy to be effective, and because the technique does a pretty good job of obscuring the compression artifacts of standard compression in general, pushing the limiter on the recorder hard on those extra tracks is an effective way to do this 'quick and dirty' in the field.