Distortion and saturation

Rock and roll got its snarl by accident. In the late 1950s and early 1960s, guitarists discovered that a damaged amp or a torn speaker made their guitar sound fat, fuzzy and dangerous. Link Wray is widely reported to have poked holes in his amp's speaker for "Rumble", an instrumental some radio stations refused to play. The Kinks' Dave Davies slashed his speaker cone with a razor blade and got the riff of "You Really Got Me". By 1965, Keith Richards was playing the riff of "(I Can't Get No) Satisfaction" through a fuzz pedal, and distortion was on every radio.

Why does it sound like rebellion? Partly because it is the sound of equipment pushed past what it was built for. But there's more to it, and you can hear it below. Turn Amount up slowly. Before you do, predict: will the guitar get louder, or something else?

It doesn't mostly get louder. It gets thicker. New frequencies pour into the spectrum, notes sustain longer, and the sound fills out. Distortion doesn't turn up what's there. It makes new sound.

Drive bends the wave

A clean amplifier copies the wave, only bigger. A distortion copies it unevenly: quiet parts pass almost unchanged, while the loud peaks get squashed, rounded or chopped flat.

A transfer curve shows that rule: what goes in on one axis, what comes out on the other. A straight diagonal line is clean. The further the curve bends away from the line, the more the wave's shape changes.

Clipping flattens the tops of the wave. Hard clipping chops them off with a sharp corner; saturation bends them gently. Gentle bending sounds warm and thick, the sound of tape pushed hot or a tube amp working hard. Hard clipping sounds buzzy and aggressive, the fuzz pedal sound.

Because the curve flattens loud parts more than quiet ones, distortion also compresses. The decaying tail of a note stays closer in level to its attack, so it seems to sustain forever. That's the endless, singing sustain of a lead guitar solo.

And because the curve depends on level, how hard you hit it matters as much as the Drive knob. The same setting is nearly clean on a soft note and filthy on a loud one, which is exactly why guitarists can clean up their sound by turning down the guitar's volume knob. A real amp is more than one fixed curve: its tone, its power supply and its speaker also react to how hard and how long you play, which is why capturing one is hard. Neural amps and captured gear shows what a model learns from an amp's recordings and what it misses.

A changed shape means new frequencies

A pure sine wave has one frequency. Change its shape and it needs extra frequencies to describe it: harmonics, at whole-number multiples of the note. A 100 Hz sine, clipped, grows components at 300, 500, 700 Hz (and with uneven clipping, at 200, 400 Hz too).

That's what happened to the guitar. Harmonics are why a distorted bass is audible on a phone speaker that can't reproduce its fundamental: the speaker plays the upper harmonics and your brain fills in the note. They're why a driven guitar cuts through a mix that buries a clean one.

Now feed it two notes at once. Switch the lab to the two-tone source: 300 Hz and 500 Hz together. Distortion doesn't only add harmonics of each note; it creates new frequencies from combinations of the two too, like 100 Hz (twice 300, minus 500) and 700 Hz (twice 500, minus 300). That's intermodulation, and it's why power chords (root and fifth only) sound so solid with distortion while a full major chord can turn to mud. A root and fifth throw off combination tones that sit neatly in one harmonic series, an octave below the root among them. Add a third and the crowd of new products no longer lines up, so the chord gets rough and murky.

Why the high notes fizz

Here's where digital distortion gets strange. Digital audio at a given sample rate can only hold frequencies up to half that rate, the Nyquist frequency. At 48 kHz, that's 24 kHz.

Distortion keeps generating harmonics higher and higher. On a low note they fade out long before 24 kHz. On a high note they don't. A 6 kHz tone's fifth harmonic is at 30 kHz, past the limit. It doesn't vanish. It folds back into the audible range as an alias, landing at 18 kHz (48 − 30). And here's the giveaway: as the note goes up, the folded frequency goes down. Those tones don't belong to the note. That's the fizz, the "broken radio" sound on high notes.

You can hear folding clearly with an effect that's built to do it. Crush holds every sample at a fixed rate of 12,000 per second, which means anything above 6 kHz folds back. Play a sine sweeping slowly from low to high and watch the spectrum:

Crush does this on purpose; it's the gritty sound of early samplers and video games. A good drive tries hard not to. Maru's Drive works internally at four times the sample rate (oversampling), which gives those high harmonics room to exist before they're filtered away. It doesn't make aliasing impossible, but it pushes it far down.

Compare at the same level, with the same input

Distortion is very easy to fool yourself with, because more drive often means more level, and louder sounds better. Compare at matched level (the lab's bypass does that), and keep the input the same. If you turn the input down to match levels, you've also changed how hard it hits the curve, and the distortion with it.

Then listen to three things separately: the body of the note, the attack, and the top-end texture. A setting that thickens the body may smear the pick attack. A bright edge that sounds great solo may fight with the vocal in the mix.

Mix blends the clean and driven guitar. A bit of clean signal under heavy drive keeps the attack and note definition while the distortion adds weight, a trick bass players use all the time.

Where a favourite preset fails

A drive that sounds great on the low E string can turn fizzy on a solo at the 15th fret, as the harmonics climb toward the top of the spectrum. Check the highest note you'll play before you save.

And harshness isn't always too much gain. It can be aliasing on high notes, intermodulation in a dense chord, or a recording that was already harsh. Lower the gain and try a simpler chord voicing, or cut some lows before the drive. An EQ before the drive changes what gets distorted; an EQ after it shapes the result. They're different tools.

Same idea, different place: Oscillators and waveforms: a square wave is close to a sine clipped hard, and a naive digital saw aliases on high notes for exactly the reason you heard here.

Make it yours

Take a bass or guitar part with both a low note and a high fill. Give the drive a job: a bass that's audible on a phone, a rough attack for a riff, or a thick layer under a chorus. Keep the input level fixed and find an Amount that works across the whole part, not only the note you started on.

Then open Drive from the lab and save the setting with the range it suits: "Bass grit, fine up to the 12th fret". A preset that knows its limits is worth more than one called "warm".

Quick check

1. Both an EQ boost and a drive make a bass brighter. Are they doing the same thing? No. The EQ turns up frequencies already there. The drive creates new ones (harmonics) by reshaping the wave.
2. You raise a distorted note and hear a whine that falls in pitch as the note rises. What is it? Aliasing: a harmonic above half the sample rate has folded back down. Its pitch moves against the note's, which gives it away.
3. You turn the source down to match two drive settings, and one suddenly sounds cleaner. Why? Turning the input down meant it hit the curve less hard, so you changed the distortion. Keep the input fixed and match the level at the output.

Next steps

• Go deeper: Delay and echo. Put drive inside a feedback loop and every echo gets distorted again. What happens when an effect feeds itself?
• Use it: Signal flow and effect order. Choose what reaches the distortion and what shapes its result.