Recording levels and how they vary with bit rate/sample freq

Posted: Wed Jul 08, 2009 11:45 pm

I was wondering what the standard db level to record at was and whether it varies with bit rate and sample frequency? I've always tried to record at the highest level possible with out any distortion (around -6db); but i just read an article stating that at 24/192 it's best to have you recording peak at -12db Any ideas?

James

Posted: Thu Jul 09, 2009 1:42 am

There is no standard db rate to record at. Keep your tracking and mixing levels LOW, and then you can just raise up your master if you need it louder. You can always draw in volume envelopes on any obnoxious peaks, or even apply the dreader limiter sparingly: a little bit of limiting goes a very long way. And remember, kids, just say no to loudness maximizer plugins.

Posted: Thu Jul 09, 2009 1:45 am

welcome

actually some would argue there is a standard
but I get why you answered the way you did

no changes for rate or bit depth

-6dbFS for a high peak is ok all things considered
( don't throw a good take away if you go over this once or twice )
-12dbFS is a very sensible general purpose target

some old schoolers would have 0dbVU set to -18 or -20dbFS depending on your place of schooling
but
don't get too excited here if you don't have to interface with analog gear or live in an existing established world like a TV or Radio station etc

Posted: Thu Jul 09, 2009 4:23 am

In theory it would be best to record each track with the maximum peak at 0dBfs. This yields the maximum signal to noise ratio. Of course, just hitting that target without overshooting is quite a trick, so we usually shoot for a lower level to give ourselves a safety net. While there may be standard in industries where people collaborate, if you are working on your own it seems to me that the choice of the level to track is a matter of comfort level and and cost/benefit calculation. How much danger of clipping is there on a particular source vs. how low a signal to noise ratio is acceptable. So the choice depends on the source (organ and distorted guitars have pretty well defined maximum volumes - drums and vocals have large transients) and on the bit depth (a lot more signal to noise ratio to spare with 24 bits than with 16) and on your knowledge and skill (as you get better you are more able to predict the level of peaks and can safely track closer to saturation). Somewhere between -12dBFS and -6 usually works for me at 24bits.

Posted: Thu Jul 09, 2009 12:52 pm

I came from many years of analog tape. We frequently peaked the meters out depending upon sound source. VU meters are not Peak meters and we all knew what our headroom was. Sometimes we wanted to saturate tape. For its color, its sound or because we knew we were going to be stacking up tracks. Digital is different and so most digital meters reflect total peak value as opposed to the average/averaging VU type meters. In the land of digital, there are those of us that still nearly treat digital as if it was analog. Some minor peaks of transients, will be clipped. This is not 100% bad depending upon what it is you are clipping. And I frequently stay within the land of 16-bit even though I can and do record at 24-bit. Just because 24-bit provides for 140 DB of total dynamic range, there is no preamp that can match this digital specification. Everything hovers around 96 to 110 DB, average. So my feeling is, even with quality converters, you still want to maintain your levels as high as possible because the input electronics really just can't cut the mustard at lower than average levels. So don't clip vocals nor any legoto instruments. And don't make it a point of clipping drums unless that's the sound you are going for.

Learn how to pack it up when cutting tracks
Ms. Remy Ann David

Posted: Fri Jul 10, 2009 3:54 am

NCdan, I disagree. In the world of both analogue and digital, if you lower your signal level at one gain stage (e.g. the faders) and then make up for it at the next stage (e.g. the master fader) you're going to have a horrible signal to noise raitio. The reason that a mixer is marked at 0dB is because it's at unity gain, which means there's no cut or boost. If you try and keep most of faders around this level and keep your master fader low, then you'll get a better SNR because each track is operating well above it's noise floor without clipping.

If you have nice things, then this becomes easier, because the noise floor is lower.

But I can't afford nice things...

This applies at every gain stage, your instruments, mic, then your preamp, eq's, compressors etc. If the output volume of all of these components run at their highest level without distorting, then you should have a decent sounding mix, in terms of volume, noise and distortion. When it comes to recording, what i always try and achieve is the highest signal without clipping from the start, because then i don't need to push any part of my signal flow too hard to get more volume.

Posted: Fri Jul 10, 2009 8:02 am

A few things to keep in mind-

Signals in the digital realm will theoretically sound the same at any dB level. (Bearing in mind that the relationship between signal and noise will alter how we perceive that signal).

Changing the bit depth shouldn't change how you treat your peaks.

Changing Sampling Rates will have no impact on your noise and peak levels.

Personally, when I record to digital, I don't shoot for a specific or "standardized" peak level. Instead, I just make sure it's not clipping and I get it as high as I can without it clipping. However, depending upon what I'm recording, I may give myself plenty of headroom per channel.

For example, with classical or with "sparse" music (such as freeform jazz) I try to make sure my levels are such that when all of the tracks are summed, the master bus output peaks below 0dBFS without having to alter faders. The main reason for this - you lose resolution when you start messing with digital faders. However, if I do need to mess with levels, I'd certainly prefer to lower them than to raise them! Also, I do like to mix and sum in the analog realm, but in many cases, time or budget do not allow this.

Of course, if you're leaving yourself -6dBFS headroom in a dense mix, your master bus will be clipping anyway, so you're likely to have to lower the individual channels anyway - again, far preferred to raising them after the fact.

But...to answer your original question - no. There is no specific standard and it generally shouldn't vary with either sample rate or bit depth. Of course, one additional note - you obviously have a LOT higher Signal to Noise Ratio (SNR) with 24 bit than you do with 16. This means, as long as your preamps are staged properly and you're even 20dB lower than a traditional 16 bit signal, you're still likely to have a less noisy recording when all the levels are finally adjusted.

Cheers-
J.

Posted: Tue Jul 14, 2009 1:13 am

Quote:

NCdan, I disagree. In the world of both analogue and digital, if you lower your signal level at one gain stage (e.g. the faders) and then make up for it at the next stage (e.g. the master fader) you're going to have a horrible signal to noise raitio.

You'd have to record pretty darn low with really crappy equipment for noise to be an issue when you raise up your master. It seems to be trendy to have absolutely, positively no noise whatsoever in the recording. I really don't buy in to that philosophy; noise can be a good thing sometimes, because it makes things sound more realistic. Of course, nasty hissing due to crappy equipment is never a good thing, but I think you get my point. Whatever level you record at, just make sure you don't clip unless you want to. I'm not an advocate of recording waaaaay below 0, but my peaks usually end up around -4 to -6.

This topic sponsored by: