Their spline fit algorithm. Meridian doesn't really tell you what it uses for algorithms, but the Wadia website describes what it does to the signal when over/up-sampling which, from what I know of spline fits and 3D graphics (and the little I had read about them in audio), is the best way fill the empty upsample space. You are effectively creating more realistic higher resolution on the sample axis of signal than with other methods around.

I found after little time with a Meridian G08 demo model, that vocals often seemed too sunken back behind the music, too soft. Not with albums that are dedicated to vocals, like Norah Jones (but even these seemed a bit too soft), but band albums like: Radiohead, Smashing Pumpkins, The Doors, Floyd. Unfortunately, I listen to that stuff a lot. Classical and instrumental pieces sounded phenomenal on it, and the music itself was always balanced, even in recordings where vocals sunk back behind the music. I like the vocals to be a little more distinguished from the rest of the music. Perhaps it didn't match well with the rest of my equipment.

I decided not to get either for now, and traded in my B&W 7 series 5.1 speaker setup to upgrade to a pair of 802Ds (plus some extra cash...). I've been using my Denon DVD-5910 since then for all sources, and can't complain. The change in speakers with the Denon DVD-5910 made a world of difference over the Meridian paired with my 703s. I am content for the time being. I'm thinking of saving up for dCS components one at a time and using the Denon as a transport. That or the Levinson 390S or the EMM Labs transport and DAC. When I get to the point where I have saved up enough, I'll start demoing those.