The ear part is completely impossible--there is no way to build anything over the ears that gives more than IIRC 32db of noise rejection (it varies somewhat by frequency)--the problem is that's what goes through the head instead.
I don't believe the microphone part could be done, either--even if you accepted a system that put the mic at a truly fixed position relative to your head you would still have sound reflecting off your face. There are also theoretical limits on focusing that mean as you get a narrower angle of observation the object doing it must be bigger. I don't know how much effect this has on microphones. (Which is why radio astronomers like to play with two dishes on opposite sides of the world--they get a much more accurate picture when their "receiver" is 8000 miles across. It's still a major undertaking to play the same games at visual frequencies, though.)