This one keeps coming up very often: “what numbers should I be shooting at when it comes to the loudness of my mixes (and self-masters)? I hear my song must be XYZ LUFS loud..”
I may well end the blog post right here by playing the ’whatever loudness sounds good is good’-card.
And while the above mantra is ever so true, I’ve felt a slight itch to comment on why getting married to a certain read-out on a loudness meter is sometimes less-than-ideal — and why.
As I picture you, the average reader of the blog, to be a music producer who is more often than not very technically skilled — you no doubt have come across the LUFS loudness scale.
If not, or as a short refresher: the LUFS (Loudness Units [relative to] Full-Scale) came into existence as the most recent addition to the various ways we have at our disposal to measure digital audio.
The idea behind LUFS was to introduce a measurement that would better reflect how the human hearing works, and, by extension, better reflect the subjective perception of loudness: you know how we’re more sensitive to high-mid frequencies, less sensitive (at regular playback volumes) to sub-bass, all that Fletcher-Munson thang.
There are a few more tricks in the bag woven into the LUFS protocol, related to (put simply) ignoring quiet passages: because LUFS was also meant to tame the loudness wars in, for example, TV commercials, ignoring (gating) quiet passages when measuring LUFS means that you technically cannot ‘hoard’ extra loudness by introducing quiet sections into your audio, thereby lowering the total LUFS read-out that way..
Now, LUFS is in my opinion a very useful metric, and it does a somewhat decent attempt at reflecting how our ears hear music. (It's especially useful for loudness consistency in long-form broadcast material.) For many, it's now become the de facto indusrtry standard of measuring loudness (if you were ever asked to deliver audio for TV, you no doubt had to go down the LUFS rabbithole).
Now, a little technical specs go a long way here, and will hammer home why I am sometimes concerned about people’s over-reliance on LUFS.
The LUFS measurement process begins with applying a filter to the audio, which basically performs a 4 dB high-shelf boost above about 2 kHz, and a 12dB/oct high-pass filter at 100 Hz. Because our ears are more sensitive to high frequencies, pre-emphasising the high-end makes sense. And because we’re hearing the sub-bass less (at regular playback volume - this bit is important in my rant), high-passing all the sub also does make sense. (Mind you, this filtering is only done for the purposes of measuring the RMS of the audio, it doesn’t actually change the output.)
But if you, like a lot of the producers I work with, are making club-music which heavily relies on the presence of sub-bass, and is played back over loud club PAs, discarding the sub-bass contents before measuring the loudness can be very problematic. I am sometimes getting asked by labels to make my master a certain LUFS. And while this makes all the sense for a lot of music, especially radio-type of stuff, and will probably reflect nicely how loud a song will sound played over a smart phone.. for orthodox club material, relying on LUFS can be quite misleading. Imagine a case of a very stripped-down ’sub-with-a-few-hats’-type of arrangement. If most of the song lives under 100Hz, it may become very difficult (and basically a little silly and quixotic) to meet those LUFS requirements — without introducing a pathological amount of distortion, anyway.
This is why, for club music, I am still partial to the good old RMS, as I feel I am being more honest to the context the music will be played in.
LUFS is a great tool for many types of audio programme, but not all, as I hope you can see now.