Gien the number of folks who are unhappy with dialog levels in movies, at least comparing those levels to the loud stuff going on all around them, it will be interesting to see how these systems work. I doubt there's a logrithm that can reliably "pull out" the speech from an audio channel and just turn that up. That leads me to suspect that it will be simple volume level adjustments, perhaps along with a bit of EQ (IIRC, if you push up the levels where the consonants live (in the few thousand Hz range, I think) you can make speech more intlligable. So maybe there's a bit of that.
Also interesting to see if Yamaha can do something useful with a 2 channel input. The problem there, of cours, is if you EQ for speech you EQ everything else too which might make everything else sound icepicky. Although if the receiver would let you use the multichannel mix, it could theoretically work its magic on the center channel, uh, channel, then downmix to 2 channel and maybe get something done. It would require some futzing with the HDMI communication process since normally a receiver configured for 2 channel will (I think) "tell" the source device it's 2 channel, and the source will just send its stereo mix . . .
Obviously not heavily documented by the manufacturers, we'll need multiple boots on the ground reports to get a reasonable handle on whatever the process is. Heck maybe one of those REW wizards could look at the center output with and without dialog enhancement and tell us what gives . . .