Guide to Levels in Digital Audio

[Lagerfeldt]

Some of the most repeated questions on this and other forums are questions about levels in digital audio.

I've written this practical guide to address most of those questions and also some of the myths and paranoia. The article requires that you already have a fairly good understanding of audio, so it doesn't deal with the most basic questions.

Unless otherwise stated, this article deals with levels in digital audio, ITB mixing and floating point DAWs. This includes sequencers such as Logic Pro, Cubase/Nuendo, Pro Tools LE (not HD), FL Studio, Ableton Live and Digital Performer.

Download "Levels in Digital Audio" PDF
http://www.popmusic.dk/download/pdf/...ital-audio.pdf

It can also be found on my link page here
Links | Audio Articles, Technical Tips, Download Presets

Contents
· Word Explanations
· Bit Depth when Recording
· A Word about Meter Values
· Levels when Recording
· Debunking some Myths
· Levels Matter with some Plug-ins
· How to Avoid Overloading a Plug-in
· How to Avoid Overloading a Bus or Master
· Levels when Processing with Plug-ins
· Levels when Outputting from a D/A Converter
· Mix Output Levels
· Normalizing
· Related Articles

[DistortingJack]

I concur.

[Ethan Winer]

Originally Posted by Lagerfeldt

I've written this practical guide to address most of those questions and also some of the myths and paranoia.

Excellent!

--Ethan

[Lagerfeldt]

Since I've tried very hard to be as objective and scientific as possible (while focusing on the practical side of things) it's not a black & white thing.

Some of the myths are just that: myths. Bogus claims about how audio works in floating point DAWs. These claims can be tested and disproved by doing null tests.

On the other hand there are valid reasons for using lower levels when dealing with specific plug-ins and D/A conversion.

It's all explained in more detail in the PDF.

[Boschen]

Thanks!
Learned some new stuff in the debunking section! - (lossless scale in DAW summing engine)
Saving for permanent reference.

[hpsounds]

Thanks !!!

[Old Goat]

Thanks, Holger, very informative.

[Alécio Costa]

nice! very high S/N ratio!

[mdwhitaker]

Very clear manual. Thanks.

[nonstatic]

fantastic, thanks very much.

[Lagerfeldt]

Originally Posted by mdwhitaker

Do you ever find that this level is too low for some plugins (mostly vintage models ie UA, Waves) ...

No, you can almost always use makeup gain at some natural occuring stage, if you really require more level. And if not, then simply insert a gain/trim utility plug-in and boost the level. This takes place in floating point.

I assume these vintage modeling plugs were designed with a specific optimal input in mind but I've never fully understood what the plugins are "expecting" to see. -12, -6, -3, etc.

They're not necessarily "expecting" to see anything. Some dynamic plug-ins have a built-in knee which means that as long as the ratio is above 1:1 high levels will trigger compression even when the threshold is set to 0. So giving those plug-ins a peak level of -6 dBFS or lower will be better from a practical standpoint.

It's often quite simple to test what happens by changing the levels while matching other factors, and then doing a null test. For instance you can boost the input level +24 dB (and overload massively) into a Logic Pro Channel EQ, then proceed to do a series of extreme +24 dB boosts and -24 dB cuts and compare that to the same thing while coming in at only -12 dBFS. The two results will null, i.e. the plug-in isn't expecting any particular level and will provide identical results in both circumstances.

[Ashermusic]

REALLY well done sir!

Glad to see that you make the same overall point as I do in my books that even in 32 bit float, good mix practices such as paying attention to your gain structure matters, especially while using 3rd party FX.

[Eloheim]

Lagerfeldt:

I just wanna thank you for making that excellent summary of DAW gainstaging. I would recommend everyone that mixes in a 32-bit floating point DAW to read it. I finally feel confident for the first time in my gain staging and levels and all!! The main question I kept running into was about whether or not using gain plugs degraded your sound in any way. I kept meaning to ask people here to help me sort out the conflicting messages I've received about it but just kept putting it off. I was afraid I might be loosing headroom or bringing up the noise floor any time I boosted and then attenuated a signal. Now that I know it's something I don't have to worry about, it's like the world inside my DAW has opened up anew to me. I had also heard all those "optimal levels" arguments, and as a result tried to keep my levels SUPER low at all mixing times, but I would always run into problems with not having enough gain to drive my dynamics plugins

For users of Ableton Live (like myself), here's an idea for an awsome rack you can create very easily to drive plugs with little trouble:

Create an Audio Effect Rack. Put one Ableton "Utility" as the first plugin in the rack's chain, and then another "Utility" as the last one. Map BOTH of the Utility's Gain knobs to ONE of the rack's macro. Then go into the "Macro Mapping" mode, and switch the second Utility's range to ([+35db]-->[-35db]), and leave the first one's as the opposite ([+35db]-->[-35db]). Now put the plugin that needs to be driven (or even all of that tracks plugs) in the rack between the two Utilites. You can instantly drive the plugin with one turn of the Macro dial, without any increase (or decrease) in overall gain! And don't forget that if you have any midi controllers with knobs, just clicking on the rack should automatically map the Macro knobs to your control surface.

[theblue1]

Originally Posted by Lagerfeldt

Some of the most repeated questions on this and other forums are questions about levels in digital audio.

I've written this practical guide to address most of those questions and [...]

Good for you, L!


I glanced over it and it looks great; looks like lots of clear cut info in an easy to grasp format.

One minor glitch I did happen to notice: the use of the word bitrate [in the dither section] where digital word length or at least bit depth would be appropriate. A bitrate, as I know we all know, is a measure of transmission bandwidth, typically used for comparing data density in lossy format compressed audio like MP3s, AACs, WMAs and such. A minor glitch in what appears to be otherwise fine work.

[nharmonic]

[Lagerfeldt]

Originally Posted by theblue1

One minor glitch I did happen to notice: the use of the word bitrate [in the dither section] where digital word length or at least bit depth

Thanks, that's a blunder, it should be bit depth. Corrected.

[theblue1]

Originally Posted by Lagerfeldt

Thanks, that's a blunder, it should be bit depth. Corrected.

Man... that was faster than a Reaper patch!

[peacenik]

Clear & practical info as usual from Mr. L! Bravo!!!!

Hmmm, if you have the time and interest I have a few requests below that 'might' be considered related to this guide:

+ a short & practical summary regarding the different metering systems (like K, BBC, dbfs, etc.)...
how to use them & when to consider using one over another.

+ Also I'm still trying to properly grasp what the proper finished Peak & RMS levels should be for different clients..
i.e. TV, Film, Music etc...

Thanks again!

[oudplayer]

Well-written and clear... Thanks for creating and sharing this!

[steveschizoid]

L, thanks for the generosity with the info and your time!

My question actually applies to tracking as well as mixing.

I have been mixing with a good number of hardware inserts (up to 14 channels), and I generally try keep the roundtrip level of these signals averaging -16 dBfs, which is line level on my converter (a Lynx Aurora 16). On relatively steady state signals this is fairly straightforward, but I have always wondered if I have been gainstaging the more dynamic signals (drum overheads for example) optimally. It seems to me like a bit of a compromise - you can either set your peaks (snare hits e.g.) or the lower amplitude portion to approximate line level - or I suppose you can shoot for midway between. In general I have tended towards the former - letting the snare hits peak at - 12 to -7 or so while the rest stays at more or less line level - what would you consider the best way to approach this?

[cinealta]

Great, thank you!

[robertshaw]

good stuff........ thanks

[ToneCre8]

Thanks...thumbsupthumbsupthumbsup

[riverrun]

Thanks!

[nas]

As usual... very helpful and informative. Thanks

[beatflux]

Nice guide!

[Macky]

thumbsup Thankyou

[ajbianco]

OK this article has tied in a lot of fragmented knowledge i had. it all comes together and makes sense now..

THANK YOU!!!!!

if i was in denmark i buy you a bear... or 10....

[Lagerfeldt]

That's great, you could always FedEx me a case.

[Firechild]

Originally Posted by Ethan Winer

Excellent!

--Ethan

Winer, 9.995 posts not bad ! Soon 10.000 !