Discover more from Jamie's list
You need an offline loudness-measuring utility
{ Heads up: this post ended up being slightly on the longer side! You may need a minute or two longer than normal to read it. }
I was working with a mentee the other day, and they kept sending me a song file that was wildly off the mark in terms of loudness delivery. I don’t mean like it was a subtle thing ... I’m talking about a dense, full-production song that got sent over with a peak level of -7.41 dB. Like I said: wildly off. It should have had a peak of about -0.1 dB.
So I pointed this out, and they went into the session and figured out what had gone awry, and they corrected it and sent me a revised bounce. And the revised bounce came in at … -7.41 dB!
So, two things.
Thing one: the reason that I know this data about this audio file is that I have a utility app on my music computer that I can drag audio files into and it will show me all the stats on said audio files. Like this:
I know that a screen full of numbers can seem overwhelming, so I’ve highlighted the three data points that I look at most often. They represent, in order:
the peak level in the file,
the song’s overall loudness (LUFS Integrated), and
the loudness range — i.e., the difference between the short-term average LUFS in the quietest moment of the song and the short-term average LUFS in the loudest moment of the song.
But, that said, when you get a minute, just take a moment to take a no-pressure look at the rest of the stats in the pic. If you look at them line by line you will probably start to get a sense of kind of what’s going on — and you can always look up terms you’re unfamiliar with if you’re curious. There’s so much good, helpful data in there.
But anyway: how great is this info? It’s everything you need to know about a piece of audio, metrically speaking. I drag every file I get from someone into this app before I even listen to it, so I can see its numbers — because these numbers tell me a lot about what I should be listening for.
For instance, if I get sent a song like this one, with a true peak of -7.41 dB, and a loudness range of 4.4 LU (LU = loudness units = basically decibels. Like European decibels), and a program loudness of -17.5 LUFS, here’s what I know about it, without listening, just from seeing those numbers:
A 4.4 LU loudness range is not a very big range; there’s just 4.4 dB between the quietest part and the loudest part. This means that this is a compact, dense, compressed-and-limited piece of audio. Probably pop or rock or rap or dance or something loud and modern like that; probably not acoustic or classical or jazz or ambient or some other quieter genre.
Given this, the peak level of -7.41 dB is clearly not as it should be — i.e., there’s a technical error happening. There’s pretty definitely a limiter on the track — we can infer that from the -4.4 LU loudness range — so this would indicate either that the limiter output is set to -7.41 dB by mistake or that there’s a plugin after the limiter that’s set weirdly and causing the output to be attenuated. I always counsel my mentees to put a gain plugin after the limiter to automate up and down from -∞ at the beginning and end of the song, so maybe that’s where the issue is!
If you adjust upward to account for the fact that this file was accidentally output -7.41 dB too quiet, the integrated loudness is -10.09 LUFS. A little on the quiet side for a modern master in the genre area we’re assuming it’s in! We should be shooting for more like -8 LUFS. So surely one of the things I’ll be helping my mentee with is getting more loudness out of their master — once they figure out what’s up with their weird output level.
Isn’t it amazing that you can know all this about a song just from looking at three pieces of data about it?
Thing two: my mentee clearly does not have a reliable way to measure loudness data in the files they’re sending out. And they very much need it — they’re sending song files out to people and they literally have no idea what’s going on with those files! They’re basically just crossing their fingers and sending them out. (And it’s okay that they did this, by the way; they’re learning, and you’re supposed to make mistakes when you’re learning. It’s how you learn.)
So, for them and for you, here is a free offline loudness-measurement app:
https://www.orban.com/freeorbanloudnessmeter
The GUI is a little 2007, but it works great. All you have to do is launch it, click the Analysis tab, and drag your audio file(s) in there. And it will analyze them and tell you their basic data! It looks like this:
You’ll note that these measurements are ever so slightly different from the measurements I took in Myriad; this is to be expected, as there is a range of acceptable ways to measure loudness (and indeed you can control the integration time in this app’s preferences!). But, as you can see, they’re basically the same — no one’s going to ding you for a couple tenths of a dB here or there.
The bottom line is that, without a utility to measure audio files, you’re flying blind. How things sound is of course the most important — but a great-sounding mix that’s sent out into the world at a totally wrong loudness is somewhere between frustrating and problematic, depending on where the file ends up and who end up hearing it. As engineers, it’s our responsibility to take care of not just the sound of files but also their underlying data structure. God forbid someone who was thinking about working with you encounters a messed-up file that you unwittingly put out into the world, and thinks “Maybe it would be best to wait a bit on working with this person.”
Making music with a slide rule — jamie