Ought to work. But poorly.
Seems to me the savonius is taking up far too much of the swept area and interrupting the airflow between the darrieus blades during the highest-power part of their rotation.
I've seen a similar design where, instead of being proportioned like a pair of big juice cans the height of the darrieus, the savonius starter-rotors were proportioned like tunafish cans, one at the top, one at the bottom, leaving all of the "eggbeater" except the narrow part near the ends of the blades (which does about squat) open to the wind. THAT should have essentially the Darrieus' efficiency with the Savonius' stall torque to make it self-starting.
Also the savonius rotors in this one seem exceptionally wide. They have a max power TSR of about .8 while the darrieus is in the 6-7 range. So the darrieus rotor should be almost ten times the diameter of the savonius, or maybe 6-7 if the darrieus is supposed to spin up the Savonius until it just spins at a rate that makes it transparent to the wind. Looks to me like a bad mismatch.
(Then again maybe I haven't computed it right and the "transparent RPM" is much higher. Or the Savonius sucking out power in the middle helps to even out the vibration of the Darrieus' torque vs. angle.)
If I were doing one I'd go for either the tunafish-can proportions for the savonius starters or use a helical darrieus maybe with 3 blades), which is apparently self-starting due to always having a blade segment with forward lift.
