This is a topic that's been on my mind for a while now...

I read this blog post by John Scrip about proper recording levels:
Proper Audio Recording Levels

Basically, it says that the analog signal going into your converters should be around 0 VU = +4 dBU for optimal sound quality, because this is the level analog equipment has been designed to work at. And it makes sense to me.

On the other hand, saturation and analog distortion is all the rage these days. People buy preamps for the specific way they can saturate and distort an input signal. They overdrive their outboard compressors to get some "analog dirt". The result of this is analog levels of 20 dBU or more.

How does this fit together?

I have to add that I personally love distortion. Subtle distortion. But I still want my mixes to sound open and airy. Listen to some of The National's albums (recorded by Peter Katis) to see what I am after.

Is it a matter of finding a balance between "hot" input signals and "normal" ones? Which instruments should I record "hot", which not? E.g. kick drum, snare, overheads, ac. guitars, e guitars, bass, vocals?

P.S. This is not about digital levels and ITB gain staging. All I want to know applies to mixing OTB as well.

Can't you just use your outboard gear to get the specific sound you're after and then attenuate back to line level before going back to your converter/console?
When I want to overdrive my preamp I just turn the input on my converter down a bit...

And for the question which signals you should record hot for a specific effect, that's more down to taste than to convention, no?

The fact that you have to be far from the range it was designed to work in is telling me that you are purposely using the equipment in a way it was not designed to, because you like the resulting effect it has on the signal.
That's how it all fits together.

Sure, my problem is not that the signals are too hot for the converters. I am using an attenuator device for that.

Thing is, individually, I like a bit of saturation / distortion on almost everything: kick, snare, overheads, bass, guitar, vocals...So basically if I record a rock band I end up recording every instrument hot. Now I am asking myself if this could be the reason my mixes tend to sound a bit tight and dense. They don't breathe.

Another thing is, in a recording situation, I really don't hear a difference between a +20 dBu signal from the preamp that is attenuated by 16 dB and a +4 dBu signal (RMS). Should I go for the +4 dBu signal in this case?

Would you say, as a rule of thumb, that I should either record a signal at nominal level (+4 dBu) or, if I want the preamp distortion, at, like, +26 dBu, but not in between?

Or should I just forget about analog levels and only make sure the converters are not clipped?

If you don't hear the difference betwee a +20 dbu and the +4 dbu I would always go with the latter. Besides, if you don't hear any effect at +20 dbu I would suggest that the preamp maybe isn't suited for the sound you're trying to get?

About the +26 dbu distorted signal, I would only record it that way if I'm absolutely sure it's going to work that way in the mix... it's the same as with compression, you can always add later if needed.

That's a good point.

So what's the preferences of people here on this forum? Are there any instruments that you regularly record with some preamp distortion (e.g. snare drums) or instruments that you always want to record as clean as possible?

I'm talking indie / rock music here. Obviously this isn't about classical music.

I love to run pre amps hot, but generally nowhere near 20dB over 0VU.

It's possible you are damaging your signal by attenuating it post pre. Doing lots of gain adjustments in the analogue domain will degrade a signal. For a good clean signal (even if it is a distorted sound) 'Unity gain' should be the thing to shoot for.
What that means is that you gain the signal you want to record once, and every other part of the electronic chain simply passes that signal straight through without introducing any imperfections (namely unwanted distortion or noise) by amplifying or attenuating that signal again.
Does that make sense?

The generally accepted (and AES) standard of 0VU = -18dBFS seems to work pretty well for most recorded music.
If you were to record a sine wave or oboe or any other fairly steady state signal that sits at around 0VU it will go to your DAW sitting around -18, which is a very useable level in 24bit digi land.

The thing that throws many people is that the VU measurement was designed as an "average" level that will not react to fast transients. Unlike your DAW meters that are almost certainly peak meters that read every transient.
So if you put a heavy Kick drum into your pre it may well read 0VU on your analogue gear, because that is it's average energy level and the VU meters will miss the transients, but when you look at your DAW that sucker is hitting -6dBFS.
What that actually means is you are hitting your pre amp way above (+12dB or more) it's optimum operating level.

Thankfully most good pres sound great when you start spanking them. Most decent new analogue gear has 20dB (the really good shit may have 24dB) or so of headroom.
You will start to slightly round off the transients and hopefully bring in some very pleasant sounding harmonics and distortion.

I hope my little rant makes at least some sense.
Short version would probably read: Calibrate 1kHz tone from gear at 0VU to DAW to read -18dbFS. Aim for -6dBFS peaks on transient signals.

Yes it does. It doesn't always work though. When you want the preamp saturation you need to run the signal hot and attenuate it afterwards so that it's not too loud in the mix.

Anyway, I think what I want to achieve with preamp distortion probably works best with percussive signals, because with percussion, the peaks are much higher than the RMS value. This means I can, as you say, round off the peaks without damaging the signal as a whole.

I think I'll try to record more steady signals like distorted guitars at much lower levels than I used to. Maybe this will improve the transparency of my mixes.

Assuming you are working ITB could you not just attenuate at the fader?
I do know exactly what you mean though.
The whole Unity Gain thing is worth thinking about as 'best practice', but in the reality of a session it's not always practical.

I would definitely agree with you re: distorted guitars. Adding additional saturation to an already driven signal can often choke a sound and kill the harmonics.

Drums, Bass, vocals (sometimes) and synths (sometimes) are ripe for a little dirtying up on the way in, in my book.

Having the time to experiment with things like finding the sweetest spot on a pre-amp is a true luxury these days though.

I follow of most of this stuff and have been tracking levels into Sonar at around -12db for some time now. However, I am unclear about levels during mixing. In the article you linked he said that even after the signal has been tracked at the correct level you will need to attenuate the signal further.

"Now, after all the other tracks are recorded, ALL of them need to be attenuated by 12, maybe 15dB or more so the mix doesn't clip."

The tracks I have going in at about -12db (on Sonar's meter) don't seem to be clipping and I have to crank the volume on my monitors to get a good listening volume. Should I be tracking at an even lower level and possibly attenuating so much after the signal is tracked?! Seems a little much to me.

No, not at all.

I'm pretty sure Mr Scrip is talking about people that slam the levels. So you with your -12 should be completely exempt from that paragraph .

A good rule of thumb if you are recording and working entirely inside a computer would be, try your very best to not touch the master fader (in any software, PT, Logic, Cubase, whatever).

If you have to push it up to get your song near to 0 you are probably not getting enough level on your recordings, if you have to pull it down (a far more common situation) then you could probably ease off a bit.

I equate transparency with "clean". When I got my motu interface modded by BLA, my mixes got much more transparent. That's because cheap A/D units distort the closer the recording level gets to zero.

It seems to me that distorted tracks take up more space in a mix. For example, recording background vocals with a gnarly mic/pre combo is going to make them more difficult to sink back into a mix. Recording them clean is going to leave you more space in the mix for other elements. You can always run mix elements through a plugin if you want distortion, but you can't undo distortion once it's been recorded. You can also dirty up a mix in the mastering stage.

I'd like to add something to this thread that I don't remember seeing mentioned in any of these "type" of threads. A daw isn't necessarily out of head room once you hit 0 on the scale a lot if not most have scaled their meters so you still have a little bit of headroom to account for things such as inter sample peaks. The reason I bring this up is because not all convertors have scalable metering, therefore I still believe that if you are going to calibrate your analogue equipment using a signal from your daw it is important to measure the signal coming out to make sure it is +4dBU/1.23 volts RMS. If your converters don't have scalable metering I'd ignore the meters on them and your daw (unless they are clipping) and focus more on the meters on your analogue equipment (provided they're there in the first place if not just keep your levels at a reasonable level on the way in and you should be fine, I'd also keep note at what level your converters meters are reading in mind for reference).

I also like how (I think it was) Bob Katz has said that +4dBU even if stated as the calibration level to be used for a specific piece of equipment that does not mean it is the actual optimum operating level for the equipment in question. Instead he (If it was Bob in the first place) suggests that each piece of equipment be calibrated by ear to the optimum calibration level (and desired amount of headroom). Sounds like a great method. I don't personally do this myself as it would be time very consuming. I just use the standard reference and if a certain bit of kit doesn't like being hit so hot, turn down the output of the preceding piece until it cleans up. Unless I like what it's doing.

I record with the PSP Vintagemeter in all the tracks and master section. Set It at -18rms=0vu. The most peaks are in the -12/-8 region. If a track peak (like percusion and drums) go up -6db I turn down the gain. All time I see the master meter for a 0vu mix... The peaks are at -9/-6db.... Before I record at -15dbfs peaks...I found the sound clear but small, now I have a strong sound with a more "analog" footprint.

sweet, I thought I was missing something. just one quick question that might be dumb. does Reaper have the same metering issues as the major DAWs? those guys making it seem to be tailoring it to the gearslutz crowd and if this is such a big issue to recordists and engineers in general, can't it be corrected with the software?

The only gear I would purposely hit hard is tape and tube gear, especially vintage. I've never driven a solid state pre hard and said 'oh hey this sounds good'. Energetic drums or vocals sometimes sound good to me pushed hard. But to fair once I clipped the ADA inputs on all the drums of a take (a case of yeah i'll play quietly until you actually start recording scenario) and they actually sounded quite good, but the ADA was the soft limiting type.