RE: SDR# : Do make sure and turn off the filtering. Do experiment with filter width (even though filtering is off). There are sweet spots for better decode. I usually move around between 12.5 and 16 khz.
RE: TRBO : Not necessarily. If you tune to an active TRBO freq and it CONSTANTLY says you have voice frames, then you need to invert. That means if you were using the -xr switch, don't, and that if you weren't using the -xr switch, do. I'm not sure a privacy-enabled TRBO signal will even produce any audio.
I always preach about CPUs. If you're going to use DSD (especially along with SDR software and VAC), it will perform much better on a beefy machine (I5/I7 with hyperthreading). So basically a multicore hyperthreading processor of recent vintage will give you the best performance. A Pentium D 2.8 HT doesn't cut it. A Core 2 Duo T5750 doesn't cut it. Anything less than those two certainly doesn't cut it. Basically when CPU maxes out during decode, you end up with a lot of choppiness. Hey, I encourage anyone who wants to run anything less to do so, but they shouldn't expect much.
Mike
WRONG. I run exactly what you say, a Pentium D 2.8 and it decodes fine, consistently, for even 1 minute of voice. My Core i7-2720QM with 12GB or RAM doesn't do any better job.
Yes, privacy DMR does produce audio, but garbled none-the-less.