Other than reconstructive D-A distortion, the whole point of the "reasonable" levels (NOT "low" - as this is how it's been done
since the advent of digital recording) is that EVERYTHING else other than the 1's and 0's themselves are designed to run at a particular level.
Although that level varies from piece to piece, it's almost always somewhere between -20 and -14dBFS when translated to digital.
Recording with the "meat" (not the peaks, but that's fine too) at around -14dBFS is how digital
was designed to run in the first place. That was the whole point. Being able to keep the sound relativley "pure" without having to smash anything.
Even 16-bit digital is going to blow away even really good analog tape as far as S/N is concerned. 24-bit has SO much usable headroom... Well over *twice* that of the analog tape that everyone is now
trying to emulate.
Can anyone give a technical description of ANY sonic benefits gained from recording tp peak at -12dBFS as opposed to -3dBFS?
Everything above - Ideal operating levels and reconstructive D-A distortion. Adding 10dB of unnecessary gain to every input (just to get a little more level) is permanently adding noise and distortion to the signal. On one signal, that might not be such a huge deal. When you put that noise and distortion on 20 or 30 tracks, it's a different story completely.
And again - I'm not pushing anything new here - This is how it's been done in every professional facility I've been in or have known of for over 20 years. Even back when 16-bit was king. Even back when most converters sucked. This is NORMAL. This is how it was all designed to work. The added detail of 24-bit was to *further* allow for this (attaining the most detail at nominal levels).
I'll give you the other part - Once everything is "in the box" then it really isn't going to matter where everything is (short of full-scale, of course). You could add gain until the peak hits -0.01dBFS without doing any damage to the audio - no matter what your D-A converters might think of it.
Which extends to the mastering phase - When mixes come in at those levels, most M.E.'s are going to turn them down considerably to
avoid reconstructive distortion anyway. Whereas, if mixes comes in at normal levels
(where the M.E.'s gear is also designed to run at) there isn't a reason for that.