View Single Post
Old 11-11-2010, 01:25 PM   #73
Human being with feelings
Join Date: Aug 2006
Posts: 2,012

Originally Posted by jnif View Post
Does this guideline to record at -24...-12 dBFS level apply only to 24-bit recording?
What is the recommended level in 16-bit recording?
The recommended level is to switch to 24-bit recording.

More seriously, with 16 bit you have a much smaller needle to thread. The same principles apply, but you're closer to the noise floor. With 16-bit it's time to start considering applying analog compression, filtering, etc, before converting to digital, in order to make the best use of the "space" you have. More sensibly, though, it's time to switch to 24-bit recording.

Another thing that still puzzles me is the input level calibration. Based on the Homestudioguide article linked in post #67, it seems that after you have calibrated your mic/instrument inputs in the analog mixer you are safe to use the analog headroom (don't adjust input gain but use faders). And when you use all that headroom and record the mixer's otput to DAW you will get -2 dBFS peak level in your DAW. According to the Homestudioguide article that is perfectly fine.
But reading this thread I understand that there can be potential distortion problems in the DAW input (MOTU 828mkii in the article) if recorded at -2 dBFS level.
So, the Homestudioguide article did not tell the whole story and following the advice in that article could lead to bad recording quality in the DAW.
Did I understand this one right?
Once again, the more obsessive-compulsive and autistic we start to get about rules and recipes and "correct" levels, the more all this stuff disintegrates into "if it sounds good, it is good" (and vice versa). Which is absolutely correct, but of mitigated usefulness when you're trying to record 12 tracks of drums in the same room the drummer is playing in, while also trying to monitor for subtle distortion on the bottom snare mic. Or when you're trying to detect a little bit of edginess on an acoustic guitar track while simultaneously playing the guitar and signing to a headphone mix of backing tracks with your voice resonating in your skull and the guitar vibrating your chest cavity and so on.

The point of the suggestion that prompted this thread is not to tell you at what input level you will start to get analog clipping or "inter-sample" digital overs... I have no idea, because I don't know what you're recording or what kind of equipment it's going through or anything like that.

With respect to the specific question about whether the home recording guy might still be experiencing clipping, there's no way to know from here, but a couple of points are relevant:

1. Very brief, instantaneously clipping of peak transients is basically inaudible and irrelevant. One or two samples one track in a 48-track project that stray into overs are not going to ruin a take.

2. Analog systems frequently handle such instantaneous overs quite elegantly, without distorting. Think of analog as like rubber, digital as like glass-- it IS possible to break or permanently distort analog by trying to pass something too-large through it, but you can get away with a lot more if you don't push your luck too much. So even if he occasionally gets a transient past the "optimum performance" range of the system, it's not necessarily going to ruin the take (at least not within the analog realm-- converting to digital is another story).

3. Very few kinds of source tracks in typical pop/rock production are likely to have peaks that are much more than 14 or so decibels above the steady-state "average" (we're speaking in pretty fuzzy terms here, so take with a grain of salt). IOW, not very many people who have an average daily checking account balance of, say, $10,000 are likely to have to deal with recurring periods where they unexpectedly have $100,000 in their account for a day or two. So if he's recording at a max "VU meter" level of zero dB, where zero on the VU corresponds to -20dBFS or whatever, the problem of what happens to +18dB peaks is likely to be an infrequent, if not entirely theoretical one.

This suggestion is not a "rule for good sound".

If what you care about is your own recordings, the best thing you can do is to suss out your own gain-staging with a day or two of careful experimentation. Put on a pair of good headphones, Record different instruments as hot as you can and then at softer levels and see whether and where the "edginess" or "brittleness" starts to recede, and then record at levels slightly lower than that. And then do the same thing with all your plugins and processors, and every other piece of equipment.

Or, make life easier and just record a little bit lower than common sense tells you is "safe", and chances are you won't have many problems, and that, if you do, they will be infrequent and not too severe.
yep is offline   Reply With Quote