I think I already mentioned earlier in the thread looking at lead ins/outs that don't appear faded, 70 db is about the best recording dynamic range I can come up with. That is a total measure though. One in the 3-5 khz range is 10-12 db lower.
Well, then. You have a good point. Let's start from the noise floor of the atmosphere, which is in the neighborhood of 6 to 8 dB SPL flat weighted 20-20k.
You can't get rid of that, unless you get rid of the atmosphere, and there are both physics and biological difficulties in that route.
If you have an average loudness orchestra, even in the quietest, no-audience venue in existence, maybe you get 84dB or so of dynamic range. Of course, we are not taking into account the human sensitivity to the noise floor (many years ago, with a well-known mastering guy, I did a perceptual SNR measurement in ERB's, but it never got submitted for publication thanks to a failing AT&T).
But sound is an interesting beast. At 120dB, sound transmission is nonlinear enough to measure distortions over the natural noise floor. at 140dB, any presumption of linearity (along with any inner hair cells in your cochlea
) goes POOF out the door. Of course, this is frequency sensitive.
If you need some evidence to that, air pressure at sea level is approximately 101kpa, and at 10,000 feet is approximately 70kpa. That's a signal of 101-70= 31kpa.
SPL reference is 0.0002 pa (note, not kpa). So 31000 (pa)/.00o02 (pa) = 1550000000 amplitude ratio. That's 183dB SPL, which is beyond earsplitting. Of course, that's not a problem because the spectrum is in microHz or so.
So, there are a variety of issues to be considered, but air is a limited medium, as is an atmosphere