I'd like to repeat a tip or concern here that I've mentioned before in other threads. The gate of a MOSFET exhibits large and variable capacitances. I think that the gate should not be connected directly to the relatively high-impedance plate. The resulting RC time “constant” creates significant phase shift in the audio band. The problem is that the RC time “constant” really isn’t a constant at all. The capacitance varies significantly as voltages swing on the MOSFET. This creates the potential for frequency / phase intermodulation distortion which is pretty ugly. Furthermore, the large value of capacitance simply loads down the plate at higher frequencies, not good even if it were constant. I would suggest driving the MOSFET gate from a very low impedance source to shorten the time constant as much as possible and reduce (but not eliminate, sadly) the amount of phase shift in the audio band. Perhaps a cathode follower intermediary stage would be best (being careful of stability with the capacitive load).