I've been messing around with binaural recording ever since I made a pair with Radio Shack mic capsules and some ear-hook headphones. I used a cassette recorder at first, then recently, a ZOOM H1.

Fast forward to a year ago when this video from The Verge came out:
Hear New York City in 3D audio - YouTube

I then realized that my binaural microphones worked perfectly on Canon DSLR's as it provided mic power. My Canon HV-20 had not. I then looked at purchasing better binaural mics or soldering better mic elements.

Which brings me to the Andrea Electronics SB-205W binaural microphones which are being cleared out for ten dollars on Amazon. I guess people have no idea what they are and how it is being marketed.

Here is a video I did with the SB-205W this past Monday. Please listen with headphones.

Chinese New Year 2016 - Binaural Audio - 12 min. version NYC - YouTube

For ten bucks? Get a pair.. it is so much fun.

For the price, they certainly work. My biggest gripe, and to be honest it's not changed since we had a few vinyl releases in the 70s, is that the results never match the picture. The mega separation gives people talking the ability to go right through your head as they talk to themselves either side of the camera, but closing your eyes allows you to point to the drums, but when you open your eyes, it's somewhere else, and that's where the problem seems to be. With a locked off camera and static mics, the image in headphones is stable, but as soon as the camera moves left or right, the image shifts - dramatically. Worse is it doesn't edit very well. A cut produces a corresponding big jump to the new perspective and it's difficult to listen to. The quality of the mics seems pretty good. In fact, if the mics had been stationary, it would have been a very nice ambient recording - to me, it just works less well with attached pictures. Interesting though.

I can understand how the sound may not match the image– anything that spoils the illusion or takes the viewer out of the moment can be bad.

If I zoomed in, the sound doesn't match the fact it is closer to the action. Still for enthusiasts like myself, this is an interesting way to get an immersive/virtual reality type sound for a small investment.

I'm really intrigued as what is next–I see 3Dio selling a binaural mic with 4 pairs of ears and I assume with software plugins or playback software that automatically blends and fades for head tracking in Virtual Reality videos.

And recently the NY Times has a great feature with binaural sound and vertical video.

http://www.nytimes.com/interactive/2015/12/29/arts/design/sound-architecture.html

When I look at the waveform of that audio, it is severely flat-topped. Not exactly what I would call "accurate reproduction." I wonder where in the chain this occurred. (I'm guessing at the mic input stage.)

I also wonder how these mics would sound with some "normal" audio, e.g. speaking, solo guitar, choir, classical ensemble, etc.