rgraves said:
It seems very odd to me to record at -18 db, maybe that's because I'm doing 16 bit...it just seems that -18 db is way too low, I mean I wouldn't hardly have to turn my preamp to 1 or 2 out of 30.
We have to differentiate here between dBVU on the analog side and dBFS on the digital side. There really is not such thing as a "dB" without referring to some scale such as "VU" or "FS".
A very nice analogy is the term degrees when used to measure temperature. There is no suchthing as a "degree" (°) all by itself, it is only in reference to a particular scale such as Celsius or Farenheit. just as 18°C means something entirely different than 18°F, -18dBVU means something entirely different than -18dBFS.
VU is used mostly on the analog side whereas FS is used on the digital side. While there is no set-in-stone standard formula for converting dBVU to dBFS, the equipment manufacturers have settled on a de facto standard in A/D conversion that a 0dBVU analong signal going in should convert to about -18dBFS on the digital side.
As a matter of generl rote (there are exceptions), analog gear should be staged so that the signal is hovering somewhere around 0dBVU, with maybe the occasional peaks going over that by a few dB if you want to drive it hard and leep the S/N ratio down. (There are other complications like whether the VU meter is a fast peak reading meter or a slower averaging meter, but we'll ignore that for this basic post.) If one's gain staging is set correctly - and their converters are so calibrated - this would bring the signal in at somewhere around -18dBFS on the digital side, with the peaks going wherever they need to go; often as high as -9dBFS or -6dBFS if the signal is dynamic enough.
If you're recording at 24 or 32 bit - which you really should be - the extra headroom on the digital side is really something you want to keep for a few reasons. First, at those bit depths you have plenty of digital dynamic range, you really don't need to boost the signal to stay away from the noise floor. Second, you might want the headroom during mixing; if you need to raise the level of one track above another, you have that option without having to pull down the level of the other; this gets real important when you have a multitrack project in which you need to fit the tracks together without having to "domino effect" the volume on 10 other tracks just to fit one anemic track in at the top of the mix. Third is so that you have some headroom when summing the tracks down to a stereo mix. Chances are you still won't have the full headroom that you need, and you'll still have to throttle back the mix buss, but the severity of such throttling is kept to a minimum.
If you're recording at 16-bit, the "rules" change somewhat, and there you will find some folks saying that you need to record hot on the digital side because you don't have as much dynamic range to play with. This is a little bit debatable on both sides, I think. However, more to the point, I wouldn't recommend recording at 16-bit these days for any reason other than if your hardware or software limits you to that depth. If you have 24- or 32-bit capability, go with it; there's no reason not to, and your mixes will thank you for it.
jndietz said:
So those tracks we hear on the radio and on our CDs were recorded at about -18dB? Or are "professionally" recorded tracks different?
Most of those were probably recorded around 0dBVU if recorded analog, and -18dBFS if recorded digital. Don't forget, though, that after recording comes the mixing and the mastering. In those stages the volume is typically incrementally increased so that by the time it hits the CD, often it's peaking around 0dBFS. But it's not *recorded* that hot.
HTH somewhat clear up an otherwise admittedly very confusing topic,
G.
