Harmonics, overtones and transients — what actually makes an instrument recognisable, and why it decides every EQ move you make.
Play A above middle C on a violin, a trumpet, a piano and a human voice. Every one of them is vibrating at 440 times per second. Every one of them is unmistakably itself, from the first fraction of a second, to anyone who has ever heard those instruments before.
Whatever is doing that identifying is not the note. The note is identical in all four cases. It is everything stacked on top of the note — and that stack is the single most useful thing to understand before you ever open an equaliser, because an EQ does not operate on instruments. It operates on that stack.
A pure sine wave — one single frequency, nothing else — barely exists in nature. You have to build one deliberately with an oscillator, and when you hear one in isolation it sounds strange and characterless, like a hearing test.
Real sounds are made of many frequencies at once. This is not an approximation or a simplification: it is provably exact. In the early nineteenth century the French mathematician Joseph Fourier showed that any repeating waveform, however complicated, can be broken down into a set of simple sine waves at specific frequencies, specific amplitudes and specific phase relationships — and, run backwards, that you can reconstruct any repeating waveform by adding the right sine waves together in the right proportions.
That result is the foundation of essentially all audio engineering. It is why a spectrum analyser can exist at all. It is why additive synthesis works. And it is why an equaliser is a meaningful tool rather than a vague tone control — because the thing it is adjusting genuinely is a collection of separable frequency components, not an indivisible blob of sound.
When a string, an air column or a drum head vibrates, it does not vibrate at just one rate. It vibrates at a lowest rate — the fundamental frequency — and simultaneously in a series of faster patterns along its length.
For strings and air columns, those faster patterns land at very tidy frequencies: whole-number multiples of the fundamental. That is what the word harmonic technically means — an integer multiple.
So a string with a fundamental of 110 Hz is also, at the same instant, producing:
Your brain does something remarkable with this. It does not hear six separate tones. It fuses the whole series into one note at 110 Hz with a particular character. The fundamental gives you the pitch; everything above it gives you the identity.
These get used interchangeably and they should not be, because the distinction explains a real category of instruments.
Strings and air columns produce overtones that are almost perfectly harmonic, which is why they sound so clearly pitched. But stretched membranes and struck metal do not. A drum head, a cymbal, a bell, a gong — these vibrate in complex two-dimensional patterns whose resonances fall at ratios that are nothing like whole numbers.
The consequence is exactly what you hear: those instruments have an indefinite pitch, or several competing pitches at once. A church bell famously seems to sound more than one note simultaneously, and that is not an illusion — there genuinely are several strong inharmonic partials, and your brain cannot settle on a single fundamental to organise them around. It is also why tuning a cymbal is not a coherent request, and why pitch-detection software falls apart on a tom fill.
One neat special case: a cylindrical tube closed at one end produces only the odd harmonics — the 1st, 3rd, 5th and so on, with the even ones absent. That missing-even-harmonic recipe is why a clarinet sounds hollow and woody compared with a flute of a similar pitch. The instrument's physical shape is directly audible in its harmonic content.
This is where synthesis and acoustics turn out to be the same subject. The classic synthesiser waveforms are not arbitrary shapes — each is a specific, predictable stack of harmonics, and knowing the recipe tells you what it will sound like before you hear it.
Subtractive synthesis, which is what most synthesisers do, follows directly from this. Start with a harmonically rich waveform — usually a sawtooth — and then use a filter to remove the harmonics you do not want. You are not creating tone; you are sculpting a surplus. That is why the filter, not the oscillator, is the heart of most synths.
Timbre — tone colour, the quality that makes an instrument identifiable — comes from three things together, and the third is the one people underrate.
The harmonic content of a real instrument is not fixed for the duration of a note. It shifts constantly, and it shifts most violently in the first few tens of milliseconds — the transient. A plucked string starts bright and harmonically crowded, then the upper harmonics decay much faster than the fundamental and the note mellows as it sustains. A bowed note builds the other way. A struck piano key contains hammer noise and inharmonic clatter that has nothing to do with the pitch at all.
There is a well-known demonstration of how much work that transient does: record a piano note, then edit off the first fraction of a second and play only the sustained part. It stops sounding like a piano. Most listeners guess an organ or an unremarkable synth pad. Nearly all of the identifying information was in the attack, and the steady part that lasts far longer carries surprisingly little of it.
This has direct consequences for you. It is why a compressor's attack setting changes the character of an instrument and not just its level — slow the attack and you let the identifying transient through untouched; speed it up and you flatten the very thing that makes the sound recognisable. It is why heavy limiting can leave a drum kit sounding synthetic. And it is why the first thing to check on a lifeless recording is not the EQ but what happened to the transients.
Here is where this becomes immediately practical. Every form of distortion, saturation, overdrive, fuzz and tape colour is doing exactly one thing: adding harmonics that were not in the original signal.
Feed a pure sine into a distortion and you get a sine plus a stack of new harmonics that the distortion invented. Which harmonics appear depends on how the circuit misbehaves, and that turns out to explain most of the arguments people have about gear:
So the perennial claim that valves sound warm and transistors sound harsh is not entirely folklore, and it is not about the components being magic. It is a statement about the harmonic content each circuit adds — and it is measurable. It also tells you something useful in a mix: a source that lacks presence can sometimes be helped more by gentle saturation than by EQ, because saturation creates high-frequency harmonic content that was never there, while EQ can only amplify what already exists. If there is nothing at 8 kHz, boosting 8 kHz raises the noise floor and nothing else.
One more layer, and it is the one that separates people who guess at EQ from people who do not.
An instrument is not just a vibrating string or column. It is a vibrating thing attached to a body — a guitar's soundbox, a violin's belly, the tube of a saxophone, the throat and mouth of a singer. That body has its own resonances, and here is the crucial part: those resonances stay at the same frequencies regardless of what note is being played.
These fixed resonant peaks are called formants, and they are the reason an instrument sounds like itself across its entire range. Play a low note and a high note on a cello and the fundamentals are far apart, but the same regions of the spectrum get emphasised both times, because the body has not changed shape.
So there are two entirely different things happening in any spectrum you look at:
A fixed EQ band interacts with these completely differently. Boosting a formant region reinforces the instrument's character consistently, whatever it plays. Boosting a fundamental region affects only the notes that happen to sit there, so it behaves unevenly across a part. That is why a low-mid cut can clean up one bass line and gut another, and why an experienced engineer asks what the part actually plays before deciding where to work. The music-theory specialization has a whole lesson on where these ranges sit for real instruments.
There is a distinction here that quietly confuses people for years, and it takes one sentence to clear up.
Harmonics climb by addition. Octaves climb by multiplication.
From a 110 Hz fundamental, the harmonics are 110, 220, 330, 440, 550 — steps of 110 each time, evenly spaced in hertz.
The octaves above 110 Hz are 220, 440, 880, 1760 — each one double the last. An octave is a 2:1 frequency ratio, which is why halving the length of a vibrating string raises it exactly one octave, and why two notes an octave apart sound like the same note in two places rather than like two different notes.
So the 2nd harmonic is an octave above the fundamental, and the 4th is two octaves above it, but the 3rd and 5th are not octaves at all — they are a fifth and a major third higher, respectively. The harmonic series and the musical scale overlap without matching, and every tuning system in history is an attempt to negotiate that gap.
Because pitch works multiplicatively, comparing small pitch differences in hertz is useless — 5 Hz is a huge error at 100 Hz and inaudible at 10 kHz. So we use a ratio-based unit, the cent:
That 5-cent figure is worth carrying. It tells you that detuning two oscillators by 3 or 4 cents will thicken a sound without anyone hearing it as out of tune, and that 15 or 20 cents will read as an error rather than as richness. It is also roughly the resolution you are working at when you tune a vocal, which is why over-correcting to perfect zero sounds unnatural — real singers live within a few cents of the target and move around inside it constantly.
If a musical tone is a well-organised stack of related frequencies, noise is the opposite: energy spread across many frequencies with no coherent relationship between them and no repeating pattern. There is no fundamental for your brain to organise around, so no pitch emerges.
This is a spectrum rather than a binary. A flute is nearly pure tone. A snare drum is mostly noise with a hint of pitch. A hi-hat is essentially all noise. A distorted guitar chord is a dense tone with substantial inharmonic content — which is exactly why it thickens a track and also why it fights everything else for the same space.
Start with a sine wave at 220 Hz. It sounds thin, hollow, electronic — clearly a note, clearly not an instrument.
Add the 2nd harmonic at 440 Hz, a little quieter. The tone immediately gets fuller and rounder. Nothing sounds brighter yet; it just sounds more solid, because you have reinforced the octave.
Add the 3rd at 660 Hz. Now something changes character rather than just weight — the tone gains a slightly reedy, hollow-woody quality, because 660 Hz is a fifth above the octave and starts to suggest an actual instrument body.
Add the 4th at 880 Hz and the 5th at 1,100 Hz, progressively quieter. The sound gains definition and begins to cut. This is roughly where it stops sounding like a test tone and starts sounding like something being played.
Now go back and remove only the 3rd harmonic at 660 Hz. Everything else stays exactly as it was. The sound hollows out — it loses its middle, and gains a slightly boxy, absent quality that is very hard to describe and very easy to hear.
Here is the point. That last move — pulling one narrow region at 660 Hz — is exactly what a narrow EQ cut does. You did not cut the note. The note is still at 220 Hz, unchanged and unmistakable. You cut one harmonic of it, and altered the instrument's identity while leaving its pitch alone.
Which is why the studio rule for this lesson matters more than it first appears. There is no such thing as EQing a vocal. There is only boosting or cutting particular harmonic regions of whatever note is being sung at that instant — and because the singer keeps changing note, the harmonics keep moving through your fixed EQ band. A 3 kHz boost lands on the 6th harmonic of a low note and the 3rd harmonic of a note an octave up, doing something quite different to each. That is why static EQ on a wide-ranging vocal never behaves consistently, and why dynamic EQ and automation exist.
Three things.
The fundamental gives you the pitch; the harmonics give you the identity. When someone asks for a brighter guitar they are asking you to raise its upper harmonics. When they say it is boxy they mean a low-mid harmonic region is over-represented. Translating vague requests into harmonic regions is most of the job.
Transients carry more identity than sustain. Protect them. Anything that flattens the first few milliseconds of a sound — heavy limiting, a fast compressor attack, aggressive tuning correction — is spending the very thing that makes the instrument sound real.
Inharmonic content is not a defect. Bells, cymbals, drums and distortion are all richer in inharmonic partials than in neat harmonics, and that is precisely what gives them their weight and their bite. It is also why they are so difficult to tune, to pitch-shift and to fit into a dense arrangement.
You never EQ a vocal. You EQ one harmonic region of it — and the note being sung decides where that region is.
Students pay for getting unstuck: ask a concept question, routing issue, DAW confusion, or mix decision tied to this lesson.
Free readers can learn from the public Q&A archive. Paid students can ask their own lesson-specific questions and get mentor replies.