Tuning: why the tuner says it's fine and the mix still sounds out of tune

A tuner measures one note on its own

A tuner gives you a binary answer — green or red — and that answer misleads for two separate reasons. The first is scale. The needle moves in cents, hundredths of a semitone: 1200 to the octave. A decent tuner resolves one cent because it can, not because one cent matters.

The second is the one that actually counts, and no better device fixes it: a tuner measures the fundamental of one note on its own, and what you call being out of tune is almost never the fundamental. It's what happens when two things sound at once and their harmonics — exact multiples of each fundamental — cross. The mistuning in cents is identical across every harmonic, because it's a ratio, but the collision isn't: the fifth harmonic clashes five times faster than the fundamental. That's how two tracks can both read green and still sound harsh together.

Beating: mistuning is counted, not judged

When two nearly identical notes sound together you don't hear two pitches: you hear one that rises and falls in level. That swell is called beating, and it pulses at exactly the difference between the two frequencies. An A at 440 Hz against another at 442 beats twice per second. Neither wave changes level at any point; the swell comes from adding them.

Here's the useful part, and it's the part almost nobody mentions: beating is measured in hertz and mistuning in cents. Because a cent is a ratio rather than a fixed amount, the same number of cents produces completely different beat rates depending on how high up it happens.

Two consequences come out of that table, and they govern half a mix. First: the same mistuning that reads as width up top reads as the level moving on its own down below. Second: when a chord sounds harsh, the culprit isn't in the fundamentals — which is where you look — but up top, where the harmonics beat fast.

Equal temperament spreads the error around

The intervals your ear recognises as pure are ratios of small whole numbers: the octave is 2:1, the fifth 3:2, the major third 5:4. At those ratios, some harmonics of the two notes land on exactly the same frequency and there is no beating at all. That is what reads as “clean”.

Equal temperament — the tuning of every keyboard, every synth and every fret — doesn't use those ratios. It splits the octave into twelve identical steps so you can play in any key without retuning anything, and that arrangement forces every interval except the octave slightly out of true.

The fifth is bent by two cents and nobody has ever noticed. The major third is bent by almost fourteen, and that's the whole story. Play a low A and its major third on a keyboard: the fifth harmonic of the A lands at 550 Hz and the fourth harmonic of the tempered third lands at 554.4. They beat four times a second. That tremble is the entire distance between a chord that “sounds good” and one that “sounds like a keyboard”.

The practical consequence is concrete: when a sustained chord sounds dirty and everything is in tune, suspect the third. Taking it out of the pad and leaving it to the voice or to an instrument with vibrato — which doesn't hold a fixed pitch and therefore doesn't beat — fixes it more often than EQ does. And it's the reason, not the only one but the main one, that distorted guitars are played in fifths: a fifth barely beats, while saturation multiplies harmonics and so multiplies whatever clashes there were between them.

Detuning on purpose: width up top, mud below

The detuned unison — two or more copies of the same sound a few cents apart — is modern production's most-used device and its worst-judged one. It works because of beating: in the mids and highs it produces movement fast enough that the ear reads it as width and density rather than as error.

Down low it does the opposite, for the same reason. Two basses 5 cents apart at 45 Hz beat once every eight seconds: for four seconds they add and for another four they partly cancel. That isn't width. It's a level that moves on its own, a limiter chasing it, and low end that sounds different on every listen.

• Detune whatever lives above roughly 300 Hz and keep the low register on a single source. It's the same rule that says centre the bass rather than widen it, and it comes from the same physics.
• If the synth has global detune, don't switch it off: high-pass the detuned voices and let just one oscillator through whole. You keep the width up top and get a stable low end back.
• A vocal double recorded twice detunes itself, and that's what makes it wide. Snapping it to the same grid as the lead is the quickest way to turn two takes into one thick voice.
• Detune and chorus aren't the same thing. A chorus modulates a delay, and varying the delay varies pitch: the resemblance isn't a coincidence, but it also duplicates and shifts in time, which brings phase relationships a pure detune doesn't have.

On that last point, the phase and polarity guide covers what happens when two copies of the same thing arrive offset; and on the first, the stereo image guide has the reasoning behind keeping low end in the centre.

When it's the reference that moves

A440 is an agreement, not a physical constant. It's standardised, but plenty of orchestras play at 442 or 443, recordings from before the fifties sit wherever they sit, and any tape or record played back at a slightly different speed arrives transposed by a few cents. A sample lifted from there isn't tuned to anything in particular.

In a session it always shows up the same way: a loop that never quite sits and that doesn't improve with EQ. Before you touch the EQ, check the pitch. A loop 8 cents below everything else doesn't sound “out of tune” — it sounds wrong without your knowing why — because there are too many notes at once to locate any one beat.

Percussion has pitch too

A kick with any decay isn't noise: it has a fundamental, nearly always between 40 and 60 Hz, and that fundamental lands on some note. If it sits a semitone from the bass, it beats against it about three times a second every time it sounds. The symptom is the usual one — “the low end never quite cleans up” — with the addition that EQ doesn't fix it, because both elements are on exactly the same frequency.

You check it by soloing the kick and the bass and listening to the tail: if it wobbles, they're close without arriving. The fix is to transpose the kick a few semitones onto the root, the fifth or the octave, not to EQ it. The kick and bass guide covers the rest of dividing up the low register, but this step comes before all of them: if the two notes are fighting, no sidechain is going to settle it.

How to train this

The skill you need isn't naming notes. That's perfect pitch, it's rare, and it's useless for everything above. What you train is noticing beating and telling fast from slow, and that takes minutes to learn: it's counting, not judging.

The shortest exercise there is: on the online piano, hold a fifth and then the same root with its major third. The fifth sits still; the third trembles. That tremble is equal temperament's fourteen cents, and once you've heard it on purpose you can't stop hearing it in a mix.

To work with numbers, the tuner reads out deviation in cents, the unit this whole guide runs on: sing a sustained note against an instrument and watch how many cents separate you from the beat you're hearing. And for the harmonic side — which interval you're using and what it's going to do — there's Chord Cosmos, where you build chords and hear them on different instruments.