That sounds pretty good for ProVoice. One problem, which I think occurred about midway, is on the user side: Many users talk too loud into the radio / the ADC gain on the radios is too hot. ProVoice doesn't like that. When I hear traffic having that problem, parts of the correspondence is unintelligible as if the volume over the clipping point is cutoff (but it doesn't actually clip in the playback) and so whatever they were saying is just a blend of loud. You know they are saying something, but you can't differentiate it. I confirmed this by comparing the P25 to ProVoice patched traffic (on OP25, the hot gain will actually clip).
Sounds like the feed provider passed along some good advice. I mostly agree with leaving the tuning off. I experience too much variability between calls for the fine tune to be worthwhile. If you had the source audio recorded and there was something you really wanted to try to decode better, the fine tune would be good for that purpose. However, leaving it set all the time will help some of the traffic decode better, but the traffic not matching the configuration will probably decode worse.
DSD+ plus is pretty sensitive. Having too hot of source audio is worse than "too quiet".
If you have drift correct and squelch set in UniTrunker, it will handle getting to a near perfect drift correction. You just need to get it close.
I have several varieties of RTLs (except the mini), and I haven't noticed subpar performance from any of them. I have several there were less than $10.
