Sounds very similar: my GPS mouse is a Navisys GR-300 with a SiRF III chip and a USB to Serial bridge (very easy for gpsd!).
Ha, you have no idea how many times I've thought of doing that to improve the signal - haven't quite got there yet. Here in Canada I get WAAS/DGPS signal too, so it's not been too much of a problem, but when I go back to the UK (dense area) I may struggle - we'll see
GPSDrive!! That's the only one I haven't tried yet - looks like it uses festival for speech output, which I've used before so no problems there. I've switched off speech support in Navit as it was driving me mental - note also that the speech support in Navit is for routing (directions), so tells you to go into this street, then that street, then the third exit etc, etc. It doesn't provide useful directions 'off-map' to a single waypoint (i.e. ahead and left a bit, 100 metres), which by the sounds of it GPSDrive can do. I'll have to give it a go when I get home (and can bash my wi-fi into life).
As far as I know TangoGPS doesn't have speech support, and the only real reason I use it is that Navit is primarily designed as a sat-nav program, to route from A to B, whilst TangoGPS is a pure GPS program like the handheld GPSr's.
One question though (open to anyone) - I mentioned in my previous post that I used TangoGPS because it seemed more responsive. That's rubbish, because both TangoGPS and Navit take data from gpsd. I've noticed that I need to walk about 4.5km/h before gpsd starts registering my movement, after which I can slow down a little (but not much), before it thinks I've stopped again. This is not helpful when scaling steep, crumbly cliffs etc, where horizontal speed is not very high. Is this gpsd's problem, or a problem with the receiver (USB GPS mouse)? Do the handheld's suffer from this too? I'd be very interested to know!!
Cheers!!