[jonmrodriguez]

> Binaural means a recording with two microphones spaced at the distance of a human head.

I don't think that's enough. It's about the effect of the shape of the head and ears on the propagation of sound waves. For example, the definition of binaural in the page that you linked to says:

> the head and ear structure affect the way sound waves are picked up

[anigbrowl]

Correct. I don't have one myself, but for high-end acoustic work people actually use dummy heads, sometimes with ballistic gel inside, to get a very good approximation of the sound that actually reaches the ears.

For less perfect results you could just set up measurement microphones in the right position and then convolve the result with a Head-Related Transfer Function.

Here's one: https://www.neumann.com/?lang=en&id=current_microphones&cid=...

They cost about $10,000 unfortunately. I'd love to have one for film use but you don't really get the effect except when listening with headphones so it's a pointless.

Also, for film we totally fake it and pan all the dialog into the center, on both stereo and surround mixes. Otherwise it would sound like the room was flipping around every time the camera cut back and forth between two actors during a dialog scene, which would get very disorienting for theatrical viewers. So when I build the audio for a dialog scene in post-production, I aim for an approximation of what you would hear if you were floating in the middle of the conversation just above people's heads. What you're hearing when you watch a movie bears little or no relation to the actual acoustics of wherever the scene was shot unless it was very quiet indeed, but is rather a composite of 10-20 different recordings.

[rawe]

There are mics that use a real human head like http://www.ohrwurmaudio.eu/ohrwurm-3-kundenbeispiele.html

[amelius]

> For less perfect results you could just set up measurement microphones in the right position and then convolve the result with a Head-Related Transfer Function.

This sounds interesting. Could you get to near-perfection by simply increasing the number of microphones used? Also, could using cheap microphones be compensated by using more of them?

[TheOtherHobbes]

>Could you get to near-perfection by simply increasing the number of microphones used?

No. Ears are (kind of) point sinks. Which is why dummy head stereo works so well on headphones - it's literally recording the sound that would usually go into your ears, as opposed to normal stereo, which records something that's usefully but rather distantly related to what goes into your ears.

Two microphones are enough for that.

A side point is that everyone has a slightly different HRTF, so it would be interesting to hear what you'd get with a neutral point sink recording convolved with your HRTF.

>Also, could using cheap microphones be compensated by using more of them?

No again. Cheap microphones add non-linear distortions and have a limited and inaccurate frequency response. If you use more cheap microphones you just get more of the same.