I agree. With really bad source material, there is only so much that you can do to make it better.

I've been in this situation before and I've got about 4 minuets while my tracks mixdown in CS6 trial so here it goes...

You may not be able to improve the 'quality' but you may be able to improve the intelligability of those voices. You could start with noise reduction in CS6 by selecting a section of noise and then opening the 'Noise Reduction' effect.  Set the 'Noise Print' options to the highest settings and then capture a noise print by clicking the 'Capture Noise Print' button.  Then click the 'Select Entire File' button.  Then it's just a matter of sliding the noise reduction % slider around to find a nice spot where the noise is tolerable around the voices.  In this case I would say around 35%.

Then you can use hiss reduction on light to remove some of the nasty hiss.  The next step would be a bit of equalization.  I like to use the 'Mastering' effect for this because I can also use the 'Loudness Maximizer' to fix the amplitude of the signal.  The EQ section is usually boosting the Low Shelf and the High Shelf by about 6dB.  The Low frequency at about 260Hz and the High frequency at about 2400Hz.  The mid or 'Peaking' I use to cut a bit of the center frequencies to eliminate some harshness... Usually the mid is about -3dB at about 400Hz. 

That usually clears it up a bit but the settings usually need a bit of adjusting to get right.  I suggest setting up the mastering rack after the initial noise reduction.  It's a shame that the noise reduction plug-in can't be inserted into the mastering rack for this though.  It runs so well on this computer even with the highest settings.

Hope that helps. Have to go now...

PS.  Great job to the developers on CS6. It's amazing and blazing fast here.   Mixdowns/Exports seem to be in real-time up until the last 30 seconds though.