If a sound jumps between loud and quiet more than you want, reach for a compressor. If it's the right level but sounds thin or disappears on small speakers, reach for saturation.
Saturation does a kind of compression as a side effect: it lowers the peaks against the average, which is why a saturated drum loop feels denser. It barely touches the difference between loud and quiet passages, though, and that's the job a compressor does. A compressor can distort as a side effect too, but only when its attack and release are set very fast.
At a glance
| Property | Saturation | Compression |
|---|---|---|
| What it changes | The shape of the waveform | The level over time |
| How fast | Instantly, sample by sample | Over milliseconds, set by attack and release |
| Harmonics | Adds them, more as you push it | Almost none at normal settings |
| Peaks against the average | Always lowers them | Depends on the attack; a slow attack raises them |
| Loud against quiet passages | Barely changes the gap | Closes the gap, as much as you set |
| Control over transients | Shaves the tips, can't let them through | Can catch them or let them through |
| Main controls | Drive, plus tone and mix on most plug-ins | Threshold, ratio, attack, release, makeup |
| Pushed too far | Fuzz and harshness | Pumping and flat, lifeless hits |
What's the difference between saturation and compression?
It comes down to time. A compressor measures how loud the sound is, then turns it down or back up over anything from a few milliseconds to a second. Saturation doesn't measure anything. It runs every single sample through a fixed curve that squashes big values more than small ones, so it reshapes each cycle of the waveform as it passes.
That one difference explains the rest of the table. Turning the volume up and down, however quickly, keeps the shape of the wave, so a compressor at normal settings adds next to nothing to the tone. Bending the wave itself creates new frequencies called harmonics, so saturation always changes the tone, even when you can't see any change in level.
You can watch both in the lab. Pick the 110 Hz sine, select Saturation and look at the 20 ms close-up: the round tops of the wave flatten out. Switch to the compressor and the same close-up shows a line lying on top of the original, while the amber lane on the big screen shows the compressor holding the level down by about 20 dB.
A compressor changes the level
A compressor turns the sound down whenever it goes over the threshold. The ratio sets how much: at 4:1, a hit that goes 8 dB over comes out 2 dB over. Attack and release set how quickly it clamps down and lets go, and makeup gain turns the whole result back up so you can compare it fairly.
Because it works over time, a compressor can do things saturation can't. A slow attack lets the first few milliseconds of a snare through before turning the rest down, which makes the hit sound sharper. A long release holds a vocal down through a whole loud phrase, so it doesn't bounce back up between words. The full explanation of each control is in the compression lesson.
Saturation changes the shape
Saturation is gentle distortion. The name comes from analogue gear: push enough level into tape, a valve or a transformer and it can't keep up, so the loudest parts of the wave get rounded off. Saturation plug-ins copy that with a curve that's almost straight for quiet signals and bends more and more as the level rises.
The rounding adds harmonics, which are new frequencies at whole-number multiples of each note. Mixed in low, they make a sound feel thicker and closer, and engineers call that warmth. Pushed harder, the same harmonics turn into audible grit and then fuzz. The saturation lesson covers the different flavours, and saturation vs distortion covers where one turns into the other.
The saturation in this lab is a soft-clip curve. It treats the top and bottom of the wave the same way, so it only adds odd harmonics, much like tape. A single valve stage bends one side sooner than the other, which adds even harmonics as well. The tape vs valve lesson compares the two.
Does saturation compress the sound?
Partly. A saturation curve squashes loud samples more than quiet ones, so the peaks of the wave come down against its average level. That's why engineers talk about tape compression, and why a saturated drum loop can feel denser without a compressor anywhere near it.
Take the curve in the lab at 12 dB of drive. A sample at 0.9, close to full scale, comes out at 0.72, which is 2 dB down. A quieter sample at 0.3 comes out at 0.29, only 0.3 dB down. The loud sample lost 1.7 dB more than the quiet one, so once the lab matches the loudness, everything under the peaks ends up a little louder relative to them.
The lab measures two different kinds of evening out, and they show where the two processors part company. The crest factor is the gap between the peaks and the average level. The loud-to-quiet gap cuts the loop into 30 ms slices and compares its loudest and quietest moments, which is closer to what people mean by dynamic range. On the bass loop:
| Setting | Crest factor | Loud-to-quiet gap |
|---|---|---|
| Original | 17.4 dB | 20.5 dB |
| Saturation, 18 dB of drive | 12.5 dB | 19.7 dB |
| Compressor, 4:1 at −30 dB, 10 ms attack | 20.2 dB | 11.6 dB |
Saturation took almost 5 dB off the crest factor and left the loud-to-quiet gap nearly where it was. The compressor did the reverse: it closed the gap by 9 dB, and its 10 ms attack let the front of each note through, so the crest factor went up.
Saturation leaves that gap alone because the curve has no memory. It reacts to how loud each sample is and ignores how loud the last half-second was, so it can't hold a vocal level from verse to chorus: a quiet verse sits on the straight part of the curve where nothing happens. It also can't let a transient through untouched, because the tallest part of a hit always gets bent the most.
Does compression add distortion?
At normal settings, hardly at all; at very fast settings, yes. The reason is the same timing difference. If the gain changes slowly compared with the waveform, each cycle comes out as a smaller copy of itself. If the gain changes as fast as the waveform, the compressor starts reshaping the cycles, which is what a saturator does.
The lab measures this on a 110 Hz sine. With the threshold at −30 dB, a 4:1 ratio, a 10 ms attack and a 100 ms release, the compressor adds about 0.5% of new harmonics, against about 5.6% from saturation at 12 dB of drive. Drop the attack to 0.1 ms and the release to 10 ms, and the compressor's figure climbs to about 5.7%, level with the saturation. Stretch the release to 700 ms and it falls to about 0.1%.
Bass shows this first. A 55 Hz note takes about 18 ms per cycle, so a release of 20 ms or so can rise and fall inside every cycle. If a compressor makes a bass sound fuzzy, slow the release before you touch anything else.
Hearing the harmonics
The harmonic screen at the bottom of the lab is the clearest way to tell the two apart. It runs a pure sine through the current settings and measures what comes out. A sine has one frequency, the fundamental, so every bar to its right is something the processor added.
Saturation at 18 dB of drive puts the 3rd harmonic, at 330 Hz, about 16 dB below the note and the 5th, at 550 Hz, about 30 dB below, with more trailing off above. A compressor with a 30 ms attack and a 100 ms release puts its loudest added harmonic about 49 dB below the note, which is very hard to hear on a real instrument.
Those harmonics are why saturation helps a bass on small speakers. Phone and laptop speakers produce very little deep bass, so a 55 Hz note on its own is close to silent on them. Its harmonics at 165 Hz, 275 Hz and up are much easier for a small speaker to play, and your ear fills in the missing note from them. A compressor can make that bass more even, but it can't make it more audible on a phone.
Transients and punch
Every drum hit starts with a transient, a sharp spike a few milliseconds long. Both processors can bring that spike down, and they go about it very differently.
Saturation flattens the tip of the spike instantly. A little of it makes the drums sound denser and a touch louder at the same peak level; a lot makes them lose their snap and pick up crunch. You can't ask it to leave the transient alone, because the transient is the loudest part and the loudest part is what the curve bends most. Pushed hard, a saturator behaves like a clipper.
A compressor gives you the choice. With an attack under 1 ms it catches the spike and the drum sounds softer and further away. With an attack of 20 to 30 ms it lets the spike through and turns down the body of the hit behind it, which makes the drum sound sharper and closer. In the lab, the one-hit screen shows both, so watch the front of the hit as you move the attack.
When to use saturation or compression
| The problem | Saturation | Compression |
|---|---|---|
| A vocal is too loud on some words | No, it distorts the loud words | Yes |
| A bass disappears on a phone | Yes | No |
| Drums feel thin and too clean | Yes | Can help, with a slow attack |
| Drum peaks are much louder than the rest | Yes, lightly | A clipper or limiter does it better |
| Snare needs more crack | Adds brightness, but flattens the spike | Yes, with a 20 to 30 ms attack |
| A synth sounds dull and flat | Yes | No |
| A mix sounds loose | A little, across the bus | Yes, 1 to 2 dB of gain reduction |
Level problems belong to a compressor, because it's the only one of the two that can respond to how loud the last few hundred milliseconds were. Tone problems belong to saturation, since a compressor at sensible settings adds almost no harmonics.
Using saturation and compression together
You'll often want both on the same track. The order changes what you get, because saturation reacts to the level going into it. Loud notes get more distortion than quiet ones, so a performance that jumps around in level also jumps around in colour.
- On vocals, bass and synths, compress firstA compressor in front evens out the level, so every note reaches the saturation at a similar level and picks up a similar amount of colour. Aim for 3 to 6 dB of gain reduction on the loudest notes.
- Then add saturation to tasteRaise the drive until you hear the sound thicken, then back off a little. If the plug-in has a mix control, 20 to 40% keeps the original underneath, which is called parallel saturation.
- On drums, try saturation firstA saturator in front shaves the sharpest tips of each hit, so the compressor after it reacts to the body of the drums and pumps less. Keep the drive low enough that the snare doesn't crunch.
- Match the level before you judgeBoth processors change loudness. Turn them off and on with the output level matched, the way the A/B button in the lab does, and keep only what still sounds better.
The wider question of chain order is in the effect order lesson.
Common mistakes
Using saturation to control a vocal's level
The loudest words get the most distortion and the quiet ones get none, so the vocal ends up with grit on its peaks and is barely more even than before. Use a compressor for the level and add saturation after it.
Judging either one by loudness
Saturation raises the average level against the peaks as it rounds them off, and makeup gain on a compressor can push the level up too. Louder sounds better on a quick listen, so compare at matched loudness every time.
A fast release on bass
If a compressor makes a bass sound fuzzy and you didn't want that, the release is following the waveform. Set it to 100 ms or longer.
Saturating sub-bass hard
Every harmonic a saturator adds to a 40 Hz note lands at a multiple of it, at 80, 120 or 160 Hz and up, which is where the kick and the bass already overlap. A little helps the bass be heard; a lot turns the low end muddy. If you want grit on a bass, saturate a copy and cut its lows with a high-pass filter.
Training your ears to tell them apart
The quiz under the lab plays the current sound through saturation or a compressor at matched loudness, with the screens covered, and you say which. Added grit or brightness on the notes points to saturation; loud moments sinking while quiet ones rise points to the compressor. Chords and bass are the easiest to start with.
Questions people ask
Is saturation a type of compression?
No, though it has a compressing side effect. Saturation bends the loudest parts of each wave more than the quiet parts, so peaks come down against the average level. It does this instantly and adds harmonics while it does it, and it barely changes the difference between loud and quiet passages, which is what a compressor is for.
Should saturation go before or after compression?
For an even amount of colour on vocals and bass, compress first, so the saturation sees an even level. For drums, saturating first shaves the sharpest peaks so the compressor reacts to the body of the hits. Try both orders and compare at matched loudness.
Can saturation replace a compressor?
Only for light peak control on sounds that can take the distortion, like drums or a bass you want gritty anyway. It can't make a quiet verse and a loud chorus sit at the same level, because it has no memory of how loud the last second was.
Why does my compressor distort the bass?
The attack and release are fast enough to follow the bass waveform itself. A 55 Hz note takes about 18 ms per cycle, so a release in that range lets the gain rise and fall inside every cycle. Slow the release to 100 ms or more and most of the distortion goes.
What does tape compression mean?
It describes how tape rounds off peaks as you record louder onto it. The peaks come down gradually as the tape saturates, which sounds like gentle, fast compression, with added harmonics and a little loss of top end.
Do I need both on every track?
No. Use a compressor when the level jumps around more than you want, and saturation when a sound needs more harmonics to be heard or to feel thicker. Plenty of sounds need neither.