the loudness your meter cannot see

you bounce your master and a reference track to the exact same integrated LUFS. you line them up, level-matched, and the reference still sounds bigger. fuller, more forward, just plain louder, at a number your meter swears is identical. you nudge yours up half a dB to check you’re not imagining it. you’re not. and the meter isn’t quite lying to you. it measures something other than the thing you hear.

the frame

a loudness meter measures energy. your ear measures something else: how that energy is spread across the bands of your hearing. saturation does almost nothing to the first and a great deal to the second, which is why a touch of it reads louder while the number barely moves.

once you see that gap, a lot of “why is mine quieter at the same LUFS” stops being a mystery and starts being a lever.

what the meter actually measures

LUFS comes from a standard, ITU-R BS.1770. under the hood it’s close to a single number: it applies a fixed filter that roughly matches the ear’s overall frequency sensitivity, squares the signal to get its energy, averages it over time, and gates out the quiet gaps so silence doesn’t drag the reading down. that’s the whole idea, and it’s a good one. it gave streaming a fair, repeatable way to turn everyone down to the same energy.

but notice what isn’t in there. there’s no model of your cochlea splitting sound into bands. there’s no accounting for how wide across the spectrum the energy is sitting. BS.1770 can’t tell a tight, narrow signal from a wide, busy one if they carry the same filtered energy. to the meter they are the same height. to your ear they aren’t remotely the same size.

what your ear actually measures

here’s the part most loudness talk skips. your hearing doesn’t sum energy in one bucket. it splits sound into critical bands, judges a loudness in each one, and adds those up. and the adding has a twist that was measured cleanly almost seventy years ago.

Zwicker, Flottorp and Stevens (1957) took tones and noise and slowly widened them while holding the level constant. inside a single critical band, spreading the energy out changed nothing: same loudness. but the moment the energy spread past the width of one critical band, loudness started to climb, with no change in level at all. same energy, more bands lit up, louder. they called it loudness summation, and it’s one of the most reliable facts in psychoacoustics.

there is a second reason this works, sitting underneath. loudness grows much more slowly than energy does: you have to roughly multiply the power to nudge the loudness. so concentrating energy in one place hits that ceiling fast and wastes most of it, while spreading the same energy into a new, empty band collects loudness the meter already counted somewhere else. you’re not making more energy. you’re spending it where the ear is still listening.

why saturation is the cleanest way to do it

now look at what a saturator does. you feed it a bass note, a vocal, a narrow synth, and it generates a harmonic series: copies of the signal climbing the spectrum at multiples of the fundamental. it takes energy that was crowded into a few bands and fans it up into bands that were sitting empty.

that’s loudness summation on purpose. the meter sees almost the same energy go in and come out, because the harmonics are small and the filter weights them modestly. your ear sees a signal that suddenly occupies twice the spectral real estate, and reads it as bigger. Durbridge, Hill and Taylor (2015) measured this directly: when they distorted tracks and asked listeners to match loudness, the distorted versions had to be turned down to sound equal. more harmonics, more perceived loudness, same level on the way in.

the 2026 trap

this matters more now than it did five years ago, because you’re mixing into a loudness-normalized world. you hit a LUFS target and assume you’ve hit a loudness target. you haven’t. you’ve matched energy. the master that reached that number with a little harmonic spread arrives sounding more present than the one that got there by pushing raw level, even though the platform turned them both to the same place.

the honest half: you’re using the gap between what the meter measures and what you hear, and the gap runs both ways. keep reaching for it and you’ll pile on perceived loudness the meter never warns you about, until the mix is harsh and fatiguing at a number that still looks polite. the meter is the floor of the conversation now. it was never the ceiling.

the same shape outside the studio

you already know a version of this. a speaker who uses their whole range, low and high, soft and sharp, reads as bigger and more present than one droning at a single pitch, even at the same volume. presence was never loudness. it was spread. the room hears the one who covers more of the spectrum, and turns the one who only got louder back down.

the tools

this is the actual engine inside WARM: it fans a signal’s energy into a controlled harmonic series, which is loudness summation you can dial. and because the meter can’t see that spread, WARM matches the energy for you with a K-weighted auto-compensation, so when you pull a character in, what’s left over is real perceived presence rather than a level bump in disguise. what you’re spending is bandwidth, and you can overspend it, which is why the gain is matched and the drive is gentle.

from the studio

tell me the track where a touch of saturation made it sit forward and the meter never moved, or the one where you chased a LUFS number and it arrived sounding small. i read every reply, and they turn into the next of these.

jonas