Who Owns
the Bottom

The mix is muddy, so you turn the bass up, and it gets worse.

Everyone meets this problem in the same order. The low end sounds simultaneously heavy and weak. There is plenty of energy down there, the meters agree, and yet nothing lands. You push the kick and lose the bass. You push the bass and the kick disappears. You put a limiter on the master and it pumps on every beat.

The usual advice is to carve one out of the other with EQ. That is part of the answer and it is the smaller part. The thing making this hard is that at these frequencies you are not really mixing two sounds. You are mixing two pressure waves that are slow enough to interfere with each other, in a region where the ear cannot tell them apart anyway.

So the real question is not what to boost. It is who owns the bottom, and whether the two of them ever occupy the same instant.

Why the bottom is a special case

Three things are true down there that are not true anywhere else in the spectrum.

The waves are long. A 50 Hz cycle takes twenty milliseconds and stretches nearly seven metres. In a normal room a single low note does not fit between the walls, which is why bass response depends so heavily on where you happen to be standing.

The ear cannot separate them. Your inner ear analyses sound in critical bands, and below about 500 Hz those bands are roughly a hundred hertz wide. A kick centred at 60 Hz and a bass note at 80 Hz are not two things to your hearing. They are one event in one band, which is the condition for maximum masking.^*

The energy is enormous. Most of the physical power in a record is in the bottom two octaves, so whatever happens there sets your peak level and therefore how loud the whole mix can be. Two low sources that reinforce each other cost you headroom everywhere else.

Twenty milliseconds

Here is the mechanism nobody mentions when they tell you to high-pass the bass.

Two sine waves at the same frequency do not simply add up. They add according to their phase relationship. Perfectly aligned, they sum to double the amplitude, which is six decibels. Half a cycle apart, they cancel completely and you get silence. Everything in between is available.^†

At 50 Hz, half a cycle is ten milliseconds. Ten milliseconds is nothing. It is a slightly late bass note, or a kick sample with a few milliseconds of silence at the front, or the difference between two microphones a few metres apart. At 5 kHz an error like that is inaudible smearing. At 50 Hz it is the difference between a mix with weight and a mix with a hole in it.

Drag the offset below. Nothing changes except when the second tone starts.

One of them owns the sub

The single most useful decision in the low end is not a setting. It is an assignment: below roughly sixty hertz, one instrument is in charge, and the other is not there.

Which one depends entirely on the music, and it is a stylistic choice rather than a technical one. It is also not negotiable once made, because the whole point is to stop two sources sharing a band where the ear cannot tell them apart and the physics lets them cancel.

Carving, not boosting

Once you have decided, the EQ move is straightforward and mostly subtractive. The non-owner gets a high-pass filter somewhere between about 60 and 110 Hz, placed by ear, and the owner is left alone.

The instinct to boost the owner instead is worth resisting. Boosting adds energy to a region that already dominates your headroom, and it does nothing about the interference, because both sources are still down there. Removing one of them solves the masking and the phase problem in a single move, and costs no level.

A useful refinement: rather than a matched pair of cuts and boosts at the same frequency, give each instrument one clear region of its own. The kick keeps its body and its beater click. The bass keeps its low mids and the string or filter movement that makes it legible on a laptop. They can overlap in the upper mids all they like. Up there the ear separates them easily.

Play the loop and put the split in and out. The overlap readout is a crude proxy, but the audible difference is not subtle.

Separating in time instead

Frequency is not the only axis available. If two sources cannot share a band, the other option is that they never occupy it at the same moment.

That is all sidechain ducking is. The bass is pulled down sharply on each kick and allowed to return over the next fraction of a beat. For the twenty or thirty milliseconds where the kick is doing its work, it has the bottom to itself, and the bass comes back to fill the space between hits.

In dance music this is so standard that the pumping became an aesthetic in its own right. But the underlying idea is general and it is worth separating from the genre: you are time-multiplexing a scarce resource. Two things need the same band, so they take turns, faster than anyone can consciously follow.

It also solves the phase problem for free, which rarely gets mentioned. Two waves cannot interfere if only one of them is playing.

Why the bottom goes mono

Almost every engineer sums the low end to mono below somewhere around 100 to 150 Hz, and the reasons are more interesting than the convention suggests.

You barely localise low frequencies anyway. Direction comes mostly from level and timing differences between your ears, and at wavelengths of several metres your head is too small an obstacle to produce much of either. A wide sub is largely a wide sub that you cannot hear as wide.^§

What you can hear is what happens when that width collapses. Anything played on a single speaker, in a club, or through a phone gets summed to mono somewhere, and if the two channels were out of phase down there they will partially cancel exactly as in Fig. 01. Making the bottom mono deliberately is how you find out now rather than later.

The speaker that cannot play it

A laptop speaker produces essentially nothing below 200 Hz. A phone is worse. And yet a well-made record still sounds like it has bass on a phone, which ought to be impossible.

The trick belongs to your hearing, not the speaker. Given a series of harmonics spaced fifty-five hertz apart, the auditory system infers a pitch of fifty-five hertz, whether or not any energy is actually present there. The phenomenon is called the missing fundamental, and it is why a bass line survives being played through something that physically cannot reproduce it.^‡

Which gives the practical move: put harmonics on the bass, usually with saturation, so that the note has an upper structure to be reconstructed from. A pure sine sub has nothing above the fundamental, so when the fundamental is gone there is nothing left. Take the bottom away from the figure below with harmonics on, then again with them off.

Louder without going lower

That gives the most useful single technique in the low end, and it is worth being precise about why it works.

Saturation folds the waveform. A folded sine is no longer a sine, and the difference shows up as harmonics at two, three, four times the fundamental. The important part is the direction: those harmonics appear above the note, never below it. You are adding energy at 110 and 165 and 220 Hz, and adding none whatsoever at 55.

Which is exactly the trade you want. The bass gets louder in the region your hearing is most sensitive to, and the sub region where the kick is trying to live is left completely alone. It also survives small speakers, for the reason in the previous figure.

Turn the drive up and watch the bar at 55 Hz. It does not move.

The rest of the toolkit

Saturation is the one everybody reaches for, but it is one of about eight, and several of the others are both cheaper and less destructive. What matters is that each has a different consequence for the kick, which is the half of the question usually left out of the advice.

Two of them cost nothing at all and are worth doing before you open a single plugin: getting the polarity and timing right, and summing the bottom to mono. Both are recovering level you already had and were throwing away.

An order of operations

Roughly what I would do, in this order, because each step makes the next one easier:

• Pick the sounds so they already fit. Far and away the highest-leverage step. A kick with a long sub tail and an 808 will never be fixed with a filter. Choose a shorter kick or a tighter bass and most of the work disappears.
• Decide who owns below sixty. Write it down if it helps. Do not revisit it every time something sounds off.
• Check polarity and timing before touching EQ. Flip the phase of one and see whether the low end gets bigger. Nudge a layered kick by a millisecond or two. Free level if it works.
• High-pass the non-owner. Subtract rather than boost. Move the corner until the owner speaks clearly.
• Duck if they still collide. Even a couple of decibels of short ducking buys a lot of clarity without sounding like an effect.
• Mono the bottom, then saturate the bass. One makes it survive the club, the other makes it survive the phone.
• Split the bass before you process it. Clean sub below about a hundred, everything interesting above. This is where most of the loudness comes from without any of the cost.
• Check on something small and something loud. Low end is the one part of a mix your room is actively lying to you about.

Decide, then execute

The reason low end feels harder than the rest of a mix is that it punishes the usual method. Everywhere else you can adjust by ear, incrementally, and converge. Down here, adjustments interact: raising the bass changes how the kick sums, which changes the peak level, which changes what the limiter does, which changes how loud the whole thing seems.

What breaks the loop is making the decision before you touch anything. One instrument owns the bottom octave. The other one is elsewhere, or it is somewhere else in time. Every filter after that is just carrying out a choice you have already made, which is a much easier thing to do by ear.

(Masking and critical bands come up again from the other direction in The Window at Three Kilohertz, which is about a voice rather than a kick drum.)