I don't believe there is very much difference between Ge and Si diodes, apart from the voltage where they turn on. They both operate on the same exponential conduction law. The first part of the following wiki link describes this fairly well: http://en.wikipedia.org/wiki/Diode_modelling
I've attached an image of the essential part here.
The formula is the same for both. The difference is that the parameter Is varies considerably between them. All this difference does is shift the point at which conduction occurs (0.2 volts for germanium vs 0.7 for silicon as you know). Notice that the value for germanium's Is is 3 orders of magnitude higher
I should try some LTspice experiments on a Germanium type to demonstrate that. What I'd expect to find is that the V_DC1 voltages for the graphs will just shift downward. In other words, instead of no change from 0 to 1 volts (of V_DC1) as it did for silicon, I would expect that changes in frequency response start to occur at some point below 1 volt in V_DC1. This would simply be the result of germanium diodes conducting sooner.
Conversely, if you were to use LEDs instead, the dead zone would be considerably higher than 1 volt for V_DC1 (due to their much higher turn-on voltages).
So in this particular application, I believe there is little to be gained from using different diodes. It will just move your sweet spot for the control signal (which is adjusted out by calibration).