Originally Posted by Dartman
...It's a bit OT but seeing how you mentioned it it sure would be nice if engineers were allowed to master levels and audio for quality again rather then just louder then the next recording.
The DAT tape will immediately start to get problems if you go over Odb but will sound perfect at anything up to 0, no extra headroom like analog tape.
Sorry for drifting OT (just trying to help wajo with the sound Q in his original post) but I agree. With the death (this week) of analog TV and the eventual death of analog processing in TV stations and networks (not to mention the impending death of analog radio and implementation of its weak replacement, HD Radio) the tide will slowly turn and audio levels will finally all be the same and free of much compression, compared to now. But I am afraid the recording industry is too married to it for it to disappear completely.
Analog systems operated at the top of the window having 3-6 dB of headroom, because lower meant noise and hiss crept in and higher yielded "soft" clipping which was fairly benign and sounded OK.
Digital audio is very different. Noise and hiss are not factors. You can operate much lower in the transfer curve and have virtually the same outcome. Also, digital clipping is heinous, and can sound like a banshee being thrown off the digital cliff, resulting in sounds very unlike the original material. For those reasons, digital typically operates with 20 dB of headroom ("0" VU is calibrated to -20 DbFS).
I can't quite remember, but DAT may have operated as high as -12 DbFS. In non-live recording situations, pro audio guys typically operate closer to DbFS (above the recommended calibration) which can squeeze a bit more signal in and lower the relative level of quantization error noise (something much different from and much lower than analog "noise") in quiet passages, but 24-bit words and beyond make that technique less necessary and less effective.
DbFS (the "FS" means "full scale") means that any digital word attempted to be recorded at or above that level can only be "no 0's and all 1's", so is a brick-wall hard clip. There is no way to predict how digital systems will attempt to resolve a signal above DbFS, and typically they distort immediately and completely
(100%). No one wants or needs to ever hear what that sort of distortion sounds like.
So digital actually has more headroom than analog, technically speaking, but that headroom is an arbitrary choice chosen to be much greater (calibrated lower) than analog to prevent the dreaded digital clipping, and that calibration standard was able to be chosen relatively lower because lower-level signals will not incur noise in digital recordings. If not for that lower calibration, it could be said that digital systems actually have much less
headroom than analog, due to analog's soft clipping capability above calibration level which digital does not have.
(sorry, back OT now