Best Recording Levels

I was trying to explain metering and levels on another board recently. I may as well repost that here .. if only to find out if I have misunderstood the nature of these things (yes, there is some generalisation in the following, simply for clarity).

..............

Regarding levels:

The key here is that your analogue gear and your digital meters measure different things on different scales.

The analogue gear will generally use a VU meter. VU stands for Volume Unit. It is a measure of input voltage and gives an RMS (Root Mean Square) reading .. or an average. It measures an average. This is similar to the way we hear sound. If a snare drum goes CRACK quickly, the meter will move and will start to fall again quickly. It may never register the absolute loudest signal on the meters needle. However, use a constant sound (like a sine wave) and the meter will stay in the same place.

OK. A preamp is designed to raise a signal to "line level". This is generally +4dbu or -10dbu depending on the kit (pro or amateur). 0VU will be calibrated to represent line level having been reached.

Of course, what we are recording isn't like a sine wave. We will have peaks that go over 0VU and troughs that are well below (including silence!). Analogue tape had plenty of allowance for peaks before nasty distortion happened.

How about digital?

Well, we go from the preamp to our AD converters (analogue/digital). These expect a line level input. They are looking for something that is around 0VU, but how should they represent that digitally?

Well, in the digital world we need a dynamic range to work with. Each "bit" we use can represent 6db. 16 bits gave us a range of 96db. 24 bits now gives us a range of 144db (huge amounts to play with!).

If we set 0VU to be the maximum, that would make no sense. What would happen to the peaks? We couldn't store them physically as a number!

Most converters are calibrated to set a signal at line level to be -18dbfs (this isn't always true.. there is some variation, but its close enough).

OK, what is this "dbfs" thing? Well, the fs stands for "full scale". It is an absolute number, and absolute volume and is used for storing the volume of a signal so that it can be reproduced by a machine. This is very different from an average level as a guide for a human listener. It's a completely different scale from the one used in the analogue world.

The key here is that 0VU is calibrated to around -18dbfs.

BUT.. remember that 0VU was an average (RMS) reading. So we can target an RMS of around -18dbfs. In practice, this means that there will be peaks above that and troughs (and silence) below that. What we will see on the meters is something that bounces around maybe -12ish and peaks at around -6ish.

That will leave us some headroom to play with compressors, reverb, EQ etc.

I hope this hasn't confused.
 
No, you're right. In fact, about 5 minutes before your post I was going to edit mine and say "-18, or whatever your equipment is calibrated to...", but didn't bother.

Funny thing is, my TASCAM 2488 is calibrated to -16, so I don't even know why I generalized with -18. I think that number came up more often in this thread, so it seemed to be the arbitrary number we were throwing around.
I think you missed his point. he was saying that in order for it to mean anything, you have to specify the scales. ie. -18dbfs = 0dbVU or +4dbU = 0dbvu or +18dbVU = 0dbFS
 
I think you missed his point. he was saying that in order for it to mean anything, you have to specify the scales. ie. -18dbfs = 0dbVU or +4dbU = 0dbvu or +18dbVU = 0dbFS
I got his point. And I was throwing out the same numbers that were being thrown around in this thread, without specifying scales and relationships....My point is that alot of people don't understand their digital meters and think that the levels they should be resording at are too low, when they're not.
 
I was trying to explain metering and levels on another board recently. I may as well repost that here
Very nice post, Freddie. Covers most of the basics pretty well.
Most converters are calibrated to set a signal at line level to be -18dbfs (this isn't always true.. there is some variation, but its close enough).
I won't disagree with the thrust of this statement, but let me add some detail to this poiint...

There is no one "standard" unfortunately from converting dBVU to dBFS (more accurately, it's actually converting from dBu, but never mind that for now). Rather there are at least a half-dozen different "standards", no one of which ever became de-facto.

The "EBU Standard" set the conversion level to -14dBFS. This standard was develpoed when 16-bit digital bit depth was common. A lot of "older" digital gear inteneded for the European market has this calibration.

The "DAT standard" set the conversion to -15dBFS. Anybody still using a DAT or ADAT machine or a converter that follows this standard and is using the onboard converters is using this calibration level (this is as true for XT/20 as standard ADAT). In fact Alesis is still using metering on some models of gear that has a special color or marking at the -15dBFS level to indicate this classic conversion level.

The -18dBFS level commonly referred to on these boards was pushed by Ampex and SMPTE, and a lot of (but not all) gear bought in the US from around '01 or '02 up until about one or two years ago has been calibrated to -18dBFS. This is also favored among most engineers over the earlier standards because it is the standard that provides the most peak headroom (18 decibels) on the digital side of all of them up until this point.

There seems to be a trend in the past year or two - especially in the hardware DAW/DAW controller/digital mixer segment, but not limited to that segement - to increase the digital available headroom even higher than the SMPTE standard. There seems to be an increasing amount of new gear who's converters are claibrated to -20dBFS as a new competitive standard, with a few even going as low as -22dBFS.

And anything in between is fair game to the engineers (as Rami's Tascam at -16 can testify to.) But a dB or two amongst friends is not that big of a deal (until you hit the digital ceiling), so in casual conversation, -18dBFS is a good common and average level to cite.

Now all we gotta do is get the producers to stop the RMS wars and actually take advantage of these wonderfully dynamic calibrations :rolleyes: .
Farview said:
in order for it to mean anything, you have to specify the scales.
Exactly. The way I like to put it is that "decibels" are kind of like "degrees". The value means nothing unless you know if you are talking about degrees Centegrade or Farenheit (or even Kelvin). When you say it is 30 degrees outside, it is either fairly chilly out or swelteringly hot; theres no way of knowing unless you put a °C or °F at the end of the number. Same thing with decibels. "0dB" can be way too hot, slightly hot, ideal or even slightly low, depending on which dB scale you're talking about (and whether you're talking about analog or digital.)

G.
 
Last edited:
I record, so my meters on the mixing desk hit 0 (sometimes) , this is compareable to a level of -18 dbfs.

Dirk


Keep in mind that this is equivalent to peaking at -18dbfs, rather than averaging -18dbfs which is what most people would reccomend (the -18 number being somewhat of a variable).

So far the most important thing I have heard is this, who cares about the digital meter? Take proper care and make proper use of your analog front end, and in the digital realm the most you will have to do is glance over every once and a while and make sure you have not clipped your converters.
 
Sorry for the delay in response! It's been a busy holiday...

Turning the track fader up about +9 to +12db (obviously not letting it clip) should let you hear it just as if it were normalized.

I've tried that. It's just not enough to accurately hear the lows. The highs are even a little touchy there. And mixing with the faders that high is a bit tough.

Turn up your monitors. Your recording level and you playback volume have nothing to do with each other.

My Wharfie's are cranked. When mixing to -12 with a track recorded at -18 I can't hear the lows (drums and bass) well enough. And THAT is where I have the most problems getting a good, full sounding mix. I can't seem to get a mix to sound full UNLESS there is a comp across the bus. The lows always sound dead as hell.

There is also a zoom control for the waveform so you can see it.

Never thought of the zoom... ;)

Who cares? Since when does the waveform - Especially on one individual track mean much of anything anyway?

I only mention the wavform being non-existant because literally EVERY example I see that shows a track, whether it is a pic or a vid, the wavform is VERY visible, where mine look like little dashes. So I figured that the examples show the way it is SUPPOSED to look. NOW, I wonder if there is something extremely wrong in my software setup somewhere.

On top of that, I've not seen a DAW that can't vertically zoom in on a waveform in over a decade.

Well, maybe the pics and vids that I've seen were zoomed in on?

Ok, I talked with Glen by phone and we went over my setup and we both came to the conclusion that something is wrong somewhere. We just couldn't figure out WHERE!!!!! I recorded a test track using a keyboard at -18. On playback, the actual levels were right around -30. Hmmmmm...
 
Ok, I talked with Glen by phone and we went over my setup and we both came to the conclusion that something is wrong somewhere. We just couldn't figure out WHERE!!!!! I recorded a test track using a keyboard at -18. On playback, the actual levels were right around -30. Hmmmmm...
Just to detail that a bit more, I had 7 (Jeff) record a solid steady organ sound from his keyboard through his 2496 into his GTPro3 so that he'd get a fairly solid and regular RMS level with which to set his digital recording level.

Over the phone we went through all the gain stages on the digital side - including the Windows mixer, the control panel driver for his 2496, and his DAW software (GTPro 3) - and Jeff tells me he had all recording and playback sliders (track and master bus) set for unity gain.

The metering in GTPro 3 was set to show a range of 40dB at peak speed, and it showed him taking in and recordig a steady -18 to -20 dBFS from his keyboard through the 2496. The problem was his playback level was very quiet and registered on his meters at -30dBFS.

The question I have is what funk might be happening or set in GTPro 3 that would cause that discrepency in supposed record levels and playback levels?

I advised Jeff to update the drivers on his 2496 as well as check with some 2496 owners on a couple of driver settings that I didn't quite recognize - which I still recommend he do - but I can offhand see how that would cause the symptoms in GTPro that he is experiencing.

Anyone with any ideas or expereince with GTPro 3 that might be able to help?

G.
 
Just to detail that a bit more, I had 7 (Jeff) record a solid steady organ sound from his keyboard through his 2496 into his GTPro3 so that he'd get a fairly solid and regular RMS level with which to set his digital recording level.

Over the phone we went through all the gain stages on the digital side - including the Windows mixer, the control panel driver for his 2496, and his DAW software (GTPro 3) - and Jeff tells me he had all recording and playback sliders (track and master bus) set for unity gain.

The metering in GTPro 3 was set to show a range of 40dB at peak speed, and it showed him taking in and recordig a steady -18 to -20 dBFS from his keyboard through the 2496. The problem was his playback level was very quiet and registered on his meters at -30dBFS.

The question I have is what funk might be happening or set in GTPro 3 that would cause that discrepency in supposed record levels and playback levels?

I advised Jeff to update the drivers on his 2496 as well as check with some 2496 owners on a couple of driver settings that I didn't quite recognize - which I still recommend he do - but I can offhand see how that would cause the symptoms in GTPro that he is experiencing.

Anyone with any ideas or expereince with GTPro 3 that might be able to help?

G.

Much thanks for the help on the phone AND with the above clarifications!!!!! I grabbed the new driver but I haven't had a chance to install it yet. I'll let you know what happens.

Thanks again, G.!!!!
 
is there a chance that some sort of panning rules are going on here? Though -18 is a lot more than the -6 I usually see.

See what happens in reaper, itll only take a second to download,k then you can configure the RMS meter in the master and look at peak and RMS at the same time

Otherwise stick a metering plugin inside gtp and see if you can find where in the chain it gets silly

Is there a master output meter in that app?
 
is there a chance that some sort of panning rules are going on here? Though -18 is a lot more than the -6 I usually see.

No panning rules. The test that Glen and I did was straight up center.

See what happens in reaper, itll only take a second to download,k then you can configure the RMS meter in the master and look at peak and RMS at the same time

I recently moved out in the country and they barely have electricity here let alone high speed Internet. I hate dialup but believe it or not it's the ONLY thing available here so downloading anything takes about 3 days IF it downloads at all. Whenever I try to download one of Rami's songs in the clinic I get about 30 seconds into the download and then it says, "Download Complete." It gives me about 15 seconds of the song.

Otherwise stick a metering plugin inside gtp and see if you can find where in the chain it gets silly

Is there a master output meter in that app?

Yes there are master output meters.

Thanks!
 
I had a band in yesterday, and was getting an average level 0f -18 when tracking. which was working well.

Never had any RED LIGHT problems, and the volume's of the recorded tracks we're pretty good. (Nice and full sounding - and the wave's looked big on the screen).

So, if when recording -18 is what I am shooting for, what level do I want to average when I am mixing down to stereo?

Usually, when I am mixing, (say, around 8 to 16 tracks) my master channel is hovering around -3. This of course, is probably too hot, as it frequently clips, soon as a rogue snare hit, or cymbal is just a bit louder. But I mix everything like this, as it's just where it ends up, due t how it sounds too me.

I don't want to be involved in the "loudness" war, but at the same time, I want nice big and professional sounding tracks.

Long story short....... what level should I be averaging when mixing down ?
 
Anywhere between -24 and -20dB(FS)RMS is generally a nice place to be.

Generally...

And assuming "big" means "big" and not "loud like a CD off the shelf" then you're golden. What's considered a "normal" mix level isn't the same level as what it's (again, "generally") going to be after the mastering phase. It's to allow the mastering engineer a little headroom. If you have a mix that's riding around -24dBRMS and it's peaking (naturally, without excessive compression or any 2-buss limiting) at -3 or -4dBFS, I don't think anyone could ask for a more "ideal" level.

Generally... For the average "commercial" sounding, pop/rock/rap/HH/etc., mix.

OF COURSE, lower is fine also. Nothing wrong with a mix that peaks at -6, -10, -15 or even a little lower for that matter. In 24-bit, there's a lot of room and no shame in using it.
 
Have you tried mixing the tracks that you recorded at -18dbfs? With only 16 tracks, I would think that you would be hard pressed to clip the master buss. On average, depending on the songs arangement, you should end up with something that comes in around -18dbfs RMS or so. It just happens. When you record at the proper levels, everything just kind of works out from there.
 
and it begins


*grabs popcorn*


from what I hear and understand if it's at around -10 ish it's not too bad. I guess it depends if you're sending it away to get mastered.. In which case I wouldn't worry about making it loud.. In fact I'd imagine it would be better to leave as much headroom as possible for the mastering engineer
 
On average, depending on the songs arangement, you should end up with something that comes in around -18dbfs RMS or so. It just happens. When you record at the proper levels, everything just kind of works out from there.
I agree with this. The only thing I worry about with levels is to make sure I don't accidentally push the mix into clipping on the peaks.

If things wind up getting a bit too hot in that regard, I'll either pull the overall fader on some individual tracks down, or - more likely if I have a ton of tracks - use either wave editing or automation to pull down some individual track peaks a couple of dB just to avoid the clipping on the mix.

I just let the rest of the levels naturally fall where they may. For me, the numbers do wind up being pretty much where John and Jay say, with RMS levels usually somehwere between -17 and -22 dBFS, depending upon content and arrangement. Any boosting I may want to do from there will wait for mastering.

G.
 
and it begins


*grabs popcorn*


from what I hear and understand if it's at around -10 ish it's not too bad. I guess it depends if you're sending it away to get mastered.. In which case I wouldn't worry about making it loud.. In fact I'd imagine it would be better to leave as much headroom as possible for the mastering engineer
Here is where a lot of people get confused: The peak level of a mix (or a track) doesn't matter as long as it doesn't clip. The RMS or average level is what is the big concern. It doesn't help that all DAWs are equiped with peak meters only, but that is where the confusion starts.
 
Here is where a lot of people get confused: The peak level of a mix (or a track) doesn't matter as long as it doesn't clip. The RMS or average level is what is the big concern. It doesn't help that all DAWs are equiped with peak meters only, but that is where the confusion starts.

Thanks for the good info guys.

Not sure what metereing REAPER uses on the master bus, but I'll take a look in the manual to be sure I'm following your advice properly.

(Apples and apples, not apples and oranges).
 
Back
Top