Absolutely nothing, to be honest, as they're two different programs that - from all actual usage, at least in my experience - end up with completely different results in the long run, even down to the quality of the audio each produces from the exact same source signal(s).
We're all hoping that the develop of DSD+ continues doing so and also that new protocols like D-Star voice support as well as the big one - P25 Phase II - make an appearance in potential future versions. I've only seen D-Star used maybe twice here in my area, just doesn't seem to be all that popular but as with most "new" things, it takes time to get entrenched I guess.
I'm not 100% sure on this but, the command line you're showing for DMR + DSD usage and claiming the crappy voice quality may partly be from how you're doing it. You wrote this:
DMR: SDR# + DSD 1.6 -fr or -xr (windows binary from RR wiki) = crappy voice with I/Q and inverted I/Q)
DMR: HDSDR + DSD 1.6 -fr or -xr (windows binary from RR wiki) = crappy voice with I/Q and inverted I/Q)
and since there's no "or" switch for DSD I'm going to assume (watch out, could be trouble) that you're meaning that you used the -fr switch (to force DMR detection only) or
used the -xr switch (to invert the DMR signal). The problem is you are using one or the other when you need to be using both - that is, if your DMR signals are inverted, that is. The audio sample you provided sounds to me like it's an inverted signal. The proper command should be:
dsd.exe -fr -xr
which forces DMR detection and when found inverts it for proper (least that's the idea) decoding. It's not one or the other, it's both
. They both have to be active switches in use at the same time. You could also just use -xr as well I suppose by itself but then DSD won't be as efficient in detecting the DMR, could be an issue, I really don't know.
I could also be wrong here about what I'm reading in that post you just made, but that seems to be the issue to me based on the audio sample - I know it's a D-Star sample but it still sounds inverted, and it could even possibly be encrypted (no idea, just a possibility). If you're trying to specifically decode DMR with DSD then using the -fr switch is pretty much a given and should always be included; the -xr switch may or may not be necessary depending on the signal.
I'd say find something using DMR and use DSD+ to decode it with -fr and see what DSD+ shows for the signal: it'll either be +DMR or -DMR and then you'll know for certain which is in use. +DMR is a positive and doesn't require the -xr switch for DSD; -DMR does require the -xr switch.
Now as for the D-Star audio quality I can't even begin to say, I don't know how D-Star works much at all, whether or not it involves any inversion (which is what that sample sounds like to me), etc.
Also: would you care to share that DSD 1.7 build you made, 'cause I don't have the slightest clue about compiling stuff for Windows (I know it's a bit more complex than with Linux which I've used/compiled stuff on in the past but I don't use it for much of anything anymore). Would be interested to see how the D-Star decoding does actually work in DSD now with the dev build - even though it's barely used around here as I already noted, I'd still like to see how it works, if you're willing. Not sure if I'd need the cygwin compile you made as well but, I'm game for some testing.