Quote:
That isn't true. Blindfolded, you can pinpoint location even at great distance by simply turning your head a little.
I'd say angular resolution of human hearing is a few degrees and does not improve with distance. It's worse for nearby sounds. Source: http://www.dspguide.com/ch22/1.htm
Changing head location indeed can improve it a few times, still not that good for faraway sources.
I'd hazard a guess of one degree resolution for a highly trained listener in perfect conditions. At 3 meters it'd give ~5 cm lateral.
Try it in a double-blind trial some day - I bet you won't fare nearly as well. (e.g. ABX) I'd love to see some papers on this topic.
It might be easier to pinpoint really diffuse sound sources due to diffraction pattern - but you won't gain anything by using speakers, in fact the diffuse room reflections will mess up this information.
Anechoic room would score highly in this regard, though it'd be decreed as unnatural by most listeners.
--
Quote:
I'm not sure if this is correct, but I suspect that the biggest difference between speakers and headphones is the presentation of dynamics. Speakers put out soft and loud into the room pretty much the way it's recorded, but headphones seem to either boost the quietest stuff or swallow the dynamic peaks. Whichever way it works, the end result is that the small details like inhales of breath between the lyrics of a song are accentuated. It could also boost the spacial cues in the ring offs.
I know that volume level in headphones can be deceiving. With cans, it's easy to get too loud. With speakers, the sound pressure slams you and room starts to vibrate if you get too loud.
Not nearly as much as you'd think - the main issue with almost all headphones is that they sound nonlinear, typically in the way that amplifies some bands of high frequencies. Most quiet sounds fit in these bands, esp. the ones you've mentioned.
Conversely, louder sounds are low frequency and many headphones are at least somewhat deficient in this area, either in amount or decay.
Equalization can help tremendously with both of these, thus improving localization a lot.
Another thing contributing to this effect is the lack of reverberation - the quiet sounds are less obscured.
I prefer to calibrate the loudness to the speech - if it sounds like normal talk, it's not overly loud or too quiet. (40-60 dB range)