Thousands of dollars on speakers and a clean listening environment with one sweet spot. Or a few hundred on headphones and an amp.<p>I know which I picked. The author doesn't include listening environment in the analysis. It is insightful to point out that the degree of separation provided by headphones is extreme and there is a conflict there between how audio is mastered and how it is delivered. Not sure how much of that can be fixed at the point of delivery.<p>When I seriously did multiplayer FPS gaming I always used headphones. The original Unreal engine had excellent spatial positioning, but didn't seem to implement much in the way of occlusion. Felt good being accused of using a radar.
Nitpick: often we refer to "omnidirectional" bass because, given that subwoofers are usually much smaller than bass wavelengths (around 16 ft at 60 Hz), they have an omnidirectional radiating pattern.<p>Live sound is usually highly artifical, with the heavy use of compression, EQ, effects, and a very exagerrated bass. For instance, all reverb is usually digital, because instruments and singers have to be close-miked to avoid feedback. Also, past a certain venue size, it's better to mike guitar amplifiers rather than make them louder or add more of them, since their radiation pattern gets very narrow at high frequencies, resulting in a "pick to the forehead" effect for the people standing directly on-axis and a muffled sound for the rest of the audience.
The author doesn't really address one of the most important mechanisms that human ears use to determine direction: the delay between a sound wave reaching the "near" and "far" ear.<p>You can try this with headphones, by playing an identical recording in both ears, but delaying one by about 1 millisecond (I set this up back in college on my BeBox).<p>You'll perceive the sound as coming entirely from the non-delayed ear, at least until you take that side of the headphones off, at which point you can once again perceive that it's also playing in the delayed ear.
Certain recordings seem to have too much stereo separation on headphones and they sound unnatural (this effect is discussed in the article). There are a variety of analog and digital methods for adding in a small amount of "crossfeed"[1] to reduce this effect and make headphone listening more comfortable.<p>[1]
> To make a truly “natural” live recording, it might be necessary to use two microphones mounted on opposite sides of a roundish, human-head-sized, sound-dampening object.<p>Not exactly, you will want 3d-modeled earlobes as only the reflections, delays and echos caused by them allow proper omnidirectional (front-behind) hearing.<p>edit: also, sound-dampening the head, e.g. making it out of foam, will diminish a natural reception as a part of our sound reception is that soundwaves from one ear get transmitted by our head bones and "material", as well as through vibration via the ground and the bones of the legs. A truly natural recording is very difficult.