All tests were done with my Metric Halo Mobile I/O 2882 interface and an Analogue Systems RS-95 VCO.
The original circuit
This one:

I measure 8V at the output when not plugged in to anything, and get about 7 octaves of calibration.
(You'll see therefore that I have a bit of a problem with the remaining tests - 7 octaves is about as much as I can hear, and 8V is not far off the supply rails of the modular, so I can't really expect much improvement in my own setup. However, the results should be useful for those with interfaces with less output.)
An even simpler version
It was pointed out to me that one of the diodes (pick one, doesn't matter) is redundant in the circuit above. To test that theory I removed the diode from the TRS ring and left it floating. Exactly the same results as above - so yes, you can do the same thing with just one diode.
Just as simple, double the voltage
It was also mentioned that connecting the TRS shield to the TS shield is a possible source of ground loops. Moreover, the above circuit isn't taking advantage of the balanced output's full voltage swing.
So, I tried a basic half-wave rectifier (see http://www.play-hookey.com/ac_theory/ps_rectifiers.html) using the TRS's T & R as inputs, leaving the shield unconnected. So like this image, but with the capacitor as above across the TS plug:

(Side note: from here on I did all my soldering at the TS plug end, leaving the TRS shield in place to actually be a shield down the length of the cable.)
Using this circuit I measured 16V on a disconnected lead, as expected. This is clearly a no-brainer improvement over the first circuit above.
Bridge rectifier
For completeness I also tried a bridge rectifier (same link, http://www.play-hookey.com/ac_theory/ps_rectifiers.html).

The results were the same as for the half-way rectifier - 16V.
Voltage doubler
Now this is a very interesting circuit: http://www.play-hookey.com/ac_theory/ps ... liers.html

I made that, and true enough, I measure 32V on a floating cable! Now that's a potentially dangerous amount of voltage to be putting into a modular, so I adjusted the 'Max Out' setting on Silent Way AC before proceeding with calibration tests.
As noted above, 32 octaves or even 12 octaves just isn't a feasible thing to hear or measure. I will say though that with this setup I measured the actual voltage on the cable when under load i.e. plugged into the VCO, and I was able to get that comfortably up to +12V.
So, for those who've been reporting only one or two octaves of calibration, do try this circuit - it's only one more capacitor than the first one, and really does double the voltage.
Let us know how it goes!