I've noticed similar activity myself using SDR-Radio (gotta love that recursive acronym style name) and the fact that the output from SDR-Radio is very low when FM Wide is chosen as the demodulation type; increasing all the volume controls in the path from start to finish don't help do much of anything in terms of the decoding performance either. I get a lot of decode errors of all kinds regardless of what protocol is in use (P25, DMR/MOTOTRBO, NXDN, etc) with SDR-Radio that I don't get using SDR# for the same purpose tuned to the same frequency and passing the baseband along to DSD+.
Maybe it has something to do with audio levels for FM Wide, I have no idea, all I know is that I don't even bother at this point using SDR-Radio for any digital mode monitoring whatsoever since (in my experience of using it) there's just something going on somewhere that wrecks the signal before it gets to DSD+, again at least for me.
When I fire up SDR# and DSD+ and hit Play, wham, near-perfect decodes of any protocol I'm capable of tuning in, period. Also, SDR-Radio seems to reverse the polarity of most every signal I try to decode with it; systems that I know for a fact are negative (-P25, -DMR, etc) that have zero issues with SDR# + DSD+ always end up being detected as the opposite if fed from SDR-Radio.
I like the program, I really do, and it does "sound better" than SDR# does by default, perhaps the reason it sounds better is part of the problem - there appears to be a lot of filtering going on that doesn't immediately appear to be anything we can totally disable to get a more pure (for lack of a better term) baseband signal for passing to DSD+ or DSD for the digital stuff.