it might be possible to build a self exciting system where as the rotor of a synchronous motor is fed from a stator coil. given that the torque of a motor is proportional to the flux density squared, and the flux density would be proportional to the rpm, this could actually match the cubic power curve of the wind, the problem being that you need a square relationship between the coils to get the current to flow into a constant voltage load.
1 turn might be good for full load, but you would need 9 turns of wire for 1/3rd the windspeed.
the existing ratios of 1, 1.73, 2, 3.46 from a typical 12 wire motor should be able to cover this ratio if some additional help is provided in the form of an auto transformer or rotor current control is provided.
for the low end of windspeed and rpm, a dc-dc boost converter is the obvious choice, since it only needs to operate for a fraction of the peak power. since the flux density of the entire system will be proportional to the rpm, there should not be any transformer start up problems, so an auto transformer for the higher voltage/rpm/wind speed is probably the most efficient way to transform the voltage and match the load.