[LAU] master levels

Something has been bugging me... Ardour is doing 32-bit FP math to handle samples internally. And yet, when I change the volume on the master fader, even if I compensate for the volume change on my RME Multiface's physical volume knob for the headphone monitor, I'd almost swear I can hear some sort of change in the audio. I'm not using any post-fader plugins. I'm probably just fooling myself, but is it possible to change the perceived audio by just changing the output level over a 4-5db range when you've got a precision like Ardour is using, and you're listening through decent 24-bit converters? In a related question (which was the real reason I wanted to know), what kind of levels should we be shooting for in the master output anyway? Of course, in the old analog tape world, they used to shoot for zero, maybe a little hotter if you wanted some saturation. In digital, they say mix to about -3 or -4db. And mastering houses seem to want more headroom than that (-6 or 7db), even though in priciple, with 24-bit resolution, they could just adjust it themselves without changing it timbrally (true?). Why do they care what your maximum level is as long as it's 24-bit and it's not clipping? Commercial CD's these days seem to be mastered with the idea that if the level ever fell below -1db, that would be a sin against God. Suppose I'm just trying to produce a demo that can be burned to CD, uploaded to a streaming service, or given to a mastering house. Is there a different ideal level for those various uses? I suppose the question I'm wondering is, if you've got it just the way you like it at some particular master fader setting, is anything harmed by making it a little hotter before you export, especially if you're going to be putting a track on CD where it will be only 16-bit? Or should you just keep it where it is, even if you're wasting some headroom? -- + Brent A. Busby + "We've all heard that a million monkeys + Sr. UNIX Systems Admin + banging on a million typewriters will + University of Chicago + eventually reproduce the entire works of + James Franck Institute + Shakespeare. Now, thanks to the Internet, + Materials Research Ctr + we know this is not true." -Robert Wilensky