Anyway, I started doing the math, and either I'm doing something wrong, or there might be something to it. Somebody double check my math?
The current balloon altitude record holder (BU60-1) reached 50 km, and had a volume of 60 000 m^3. The balloon in question had a mass of ~34,5 kg and the scientific payload + parachute of another ~5,5 kg.
If we can be generous and replace that science payload and parachute with wind turbine + battery of equal mass (total mass of 40kg) and keep the 50km altitude. To lift 40 kg you need 36 m^3 of hydrogen.
Potential energy of a 40kg object at 50km is 19.600 kJ. That's absolute maximum of energy available to extract.
On the other side, to generate 3,2 kg of hydrogen, at 50 kWh/kg, we need 512 kJ (160 kWh).
512kJ = 142Wh
Sure, back of the envelope calculation, but that's quite a margin.
The numbers do seem to stack up but there are a few pratical issues!

First problem:
=========
The air density at 50km is so low (0.08% ground level) that an enormous turbine would be needed to extract any useful energy. I tried a few cacluations:
To minimise the blade diameter the turbine would need to descend as fast as possible (power out is proportional to wind speed cubed). But this is limited by the maximum tip speed which needs to be less than the speed of sound.
Assuming a two blade prop, the optimum tip speed ratio (blade tip speed/wind speed) is around 6 (athough a single blade prop would likely be better for this application being lighter).
Arbitrarily limiting tip speed to 300m/s (Mach 0.88), the wind speed, ie. descent speed, should be <= 50m/s. At that speed the turbine would be losing potential energy at m x g x h = 40x50x9.8 = 19.6kW. Thus the wind energy input to the turbine is 19.6kW.
The kinetic energy in the wind passing throughy the turbine = 1/2 m v^2 where m is the mass of air through the turbine each second and v is the air velocity.
m = air density * blade swept area x wind velocity.
Thus power in = 1/2 x density x pi x r^2 x v^3
Therefore to capture 19.6kW of wind (input) power at 50m/s at 50km altitude the blade diameter would need to be 19.7m. That seems a bit unlikely, even for carbon fibre blades, given the available weight budget after the battery and generator mass.
However the air density improves dramatically at slightly lower altitudes - 4x greater at 40km and 18x at 30 km requiring blade diameters of 10m and 4.7m respectively. 4.7m may be achievable in 10kg or less (just guessing).
2nd problem:
========
The weight of the battery. The potential energy of 40kg at 50km is 5444Wh. The theoretical maximum efficiency of a wind turbine is 59% (Betz limit). Assuming an overall efficiency of 40%, allowing for drag, turbine, generator, battery charger and battery charging efficiency losses the battery would have to store 2178Wh.
Current Li-ion batteries have an energy density of less than 300Wh/kg meaning >= 7.3kg of batteries. That isn't unreasonable so the batteries might not be a problem after all. The charge rate isn't a problem either - descent time at 50m/s from 50km = 1000s, thus charge rate is only 3.6C which is modest for most batteries.
Third problem:
=========
Weight of generator. Given 40% efficiency the generator has to be around 9kW. Perhaps somene else can provide a reasonable weight for a state of the art 9kW generator?
Fourth problem:
=========
How much fuel and manpower would be required retrieving this thing (what should we call it? Balloony Mc. Balloon-Face?) and returning it to it's start point after spending time in the jetstream? Realistically some means of controlling the descent would be needed to return it reasonably close to it's launch position. Much heavier units, descending much faster would emeliorate this problem somewhat but this is likely the killer issue.
Fifth problem:
=========
What happens to all the hydrogen dumped at the top of the stratosphere? How much will escape into space rather than recombining with oxygen? How long would the oceans last if this system were to supply most of mankinds' energy needs? What would be the maximum altitude to ensure most of the hydrogen is recaptured in the atmosphere?
Scaling up will help considerably - 40kg is a tiny power station but ultimately there will be severe limits to the number of locations where governments will tolerate large masses pluming earthwards at 100mph+ when in control - and considerably faster when they aren't!

I like this idea - I reckon I could watch it all day! Lots of automation would be needed to capture the returning units and swap the batteries. Aviation in the vicinity might be somewhat hazardous however!
[EDIT] Forgot to add that as the air density increases as the unit descends the optimum rotor diameter and descent speed will change so the net efficiency of a (necessarily) fixed diameter rotor will likely vary considerably during it's descent.