Why Your Plugins Might Be Lying to You

You know that harsh, brittle sound some mixes have? That "something's wrong but I can't EQ it out" feeling? There's a technical reason for it, and once you understand it, you'll make better decisions about which plugins to use and how hard to push them.

This paper explains aliasing—a form of digital distortion that's invisible on most meters, impossible to remove once it's there, and hiding in a surprising number of professional tools.

The Myth of the Staircase

You've probably seen diagrams showing digital audio as tiny stair steps—little blocks stacked up to approximate a smooth wave. This is completely wrong. It's like explaining cars by saying "the wheels are square, but if you have enough of them, it feels round."

Here's what's actually happening:

Digital audio is more like a connect-the-dots puzzle, except the "dots" aren't just dots—each one contains instructions for drawing smooth curves that blend perfectly with the curves from neighboring dots. When you play back the audio, these instructions recreate the original smooth wave exactly. Not approximately. Exactly.

The Speed Limit

Digital audio has a speed limit called the Nyquist frequency—half your sample rate. At 44.1 kHz (CD quality), that's 22,050 Hz. At 48 kHz, it's 24,000 Hz.

Think of it like a camera taking photos of a spinning wheel. If the wheel spins too fast relative to your camera's shutter speed, weird things happen—the wheel looks like it's spinning backward, or barely moving, or jumping around randomly. You've seen this in movies with car wheels or helicopter blades.

That visual glitch? That's aliasing. And it happens in audio too.

The Problem with Distortion

Here's where it gets interesting. Clean audio—just playing back a recording, adjusting volume, adding reverb or EQ—doesn't create aliasing. These are "linear" processes. What goes in comes out, just modified in predictable ways.

Distortion is different. Saturation, compression, limiting, clipping—these add new frequencies that weren't in the original signal. That's literally what "harmonics" are: new stuff created by the process.

When you run a 10 kHz tone through a saturator, it doesn't just get louder or change shape. It sprouts new frequencies at 20 kHz, 30 kHz, 40 kHz, and beyond. Those are the harmonics that give analog gear its character.

But wait—at 44.1 kHz, your speed limit is 22,050 Hz. Where does that 30 kHz harmonic go?

It bounces back. Like a racquetball hitting a wall, it reflects off the Nyquist frequency and comes screaming back down into your audible range. A harmonic that should have been at 30 kHz lands at around 14 kHz instead. And it's not harmonically related to your original note anymore—it's just... wrong. Dissonant. Harsh.

The Multiplication Problem

Different types of processing create different amounts of new frequencies:

Here's the brutal part: these multiply together.

Imagine your signal goes through a subtle tape emulation (3×), then a compressor (3×), then hits a limiter (10×). Your bandwidth just tried to expand by 3 × 3 × 10 = 90 times. At 44.1 kHz, that means a 5 kHz guitar note is trying to create harmonics up to 450 kHz. Everything above 22 kHz bounces back as garbage.

This is why some plugin chains sound increasingly harsh and "processed" while others stay clean and open. It's not just about taste—it's about math.

Aliasing Is Forever

This is different from other problems. Too much high end? EQ it down. Muddy low-mids? Scoop them out. Room resonance? Notch it. But aliasing? It's baked in permanently.

What It Sounds Like

Aliasing doesn't sound like "distortion" in the obvious sense. It sounds like:

• Harshness you can't EQ away
• A "brittle" or "digital" quality
• Mixes that feel small and constrained
• High frequencies that are "present" but unpleasant
• A general sense that something's wrong, even if you can't identify what

Producers in the 1990s and early 2000s learned to work around these problems intuitively. They'd back off the processing, avoid certain combinations, stick to what "worked." They didn't know why some moves sounded terrible—they just learned to avoid them. Today we understand the cause: their tools were breaking the rules of digital audio.

Your Compressor Might Be Lying About Its Attack Time

This is where it gets really interesting. Aliasing doesn't just affect the sound coming out of a plugin—it affects how the plugin makes decisions.

A compressor has a "sidechain"—an internal circuit that listens to the audio and decides when to turn down the volume. This sidechain is doing heavy math: detecting levels, comparing to thresholds, calculating how fast to react.

All that math? It's nonlinear processing. Which means it creates new frequencies. Which means it can alias.

• Gates close too early or too late
• Attack times become erratic instead of smooth
• The compressor "sees" peaks that aren't there, or misses peaks that are
• Release times get jumpy and nervous
• Limiters miss the true peaks of your waveform entirely

This is why some limiters sound "pumpy" or "crunchy" on fast settings while others stay transparent. It's why some compressors feel "analog" and musical while others feel "digital" and harsh—even with identical settings.

The plugin's meters might say "2ms attack," but if its sidechain is aliasing, it's not actually responding in 2ms. It's responding erratically, sometimes faster, sometimes slower, always wrong.

Why Most Limiters Ignore True Peaks

Here's a dirty secret: many limiters don't actually limit to the level you set. They let peaks through that exceed your ceiling.

Why? Because properly tracking true peaks requires a sidechain that doesn't alias. Building that is hard. It's easier to just... not do it, and hope nobody notices.

When you see "True Peak" limiting, that's the plugin claiming it actually solved this problem. Some did. Many didn't.

The Oversampling Approach

The most common solution is oversampling: temporarily increasing the sample rate inside the plugin so there's more headroom above 20 kHz for harmonics to exist without bouncing back.

If a plugin runs at 4× oversampling in a 44.1 kHz session, it's internally working at 176.4 kHz. The speed limit is now 88.2 kHz instead of 22 kHz. Those harmonics at 30 kHz and 40 kHz? They fit now. Before the audio leaves the plugin, it filters out everything above 22 kHz (where you can't hear anyway) and drops back to 44.1 kHz.

The result: The harmonics exist while the processing happens, so the math is correct. Then they're cleanly removed. No bouncing, no garbage, no aliasing.

Smart vs. Dumb Oversampling

Here's where plugin design gets clever or lazy.

Dumb approach: Oversample the entire input and output of the plugin. This works, but it's expensive (CPU) and can introduce other issues.

Smart approach: Only oversample the parts that actually create new frequencies. If you're just adjusting EQ or gain, no oversampling needed—those can't alias. But the moment the signal hits saturation or compression, that specific stage runs oversampled.

The smartest plugins take it further: they don't even oversample the audio path if they don't have to. They oversample only the nonlinear "change" being applied to the signal. The original audio passes through bit-perfect, and only the distortion/compression artifacts get the anti-aliasing treatment.

Choosing Plugins

Not all plugins with the same name on the knobs perform equally. A "tape saturation" plugin from Company A might handle high frequencies beautifully while Company B's version falls apart above 7 kHz.

You can't always tell by listening to presets or simple tests. Aliasing often reveals itself in complex material, dense mixes, or when several processors are chained. It accumulates.

Questions to ask about your plugins:

• Does the manufacturer mention anti-aliasing or oversampling?
• Does the plugin have an oversampling option? (Good sign they're thinking about it)
• How does it sound when you push it hard on bright, complex material?
• Does it maintain clarity when stacked with other processors?

The Chain Matters

Remember that bandwidth multiplier table? Even if individual plugins are well-designed, you can overwhelm a poorly designed one downstream by feeding it harmonically rich material from upstream processors.

A 5 kHz synth note through three saturators might be generating harmonics up to 50 kHz by the time it hits your limiter. If that limiter isn't prepared for that, those inaudible harmonics bounce back as audible garbage in its output.

Signal Bandwidth Matters

This whole problem scales with the frequency content of your input.

• A bass guitar with nothing above 4 kHz? Even a mediocre saturator will probably sound fine.
• A full mix with content up to 20 kHz? You're pushing the limits from the start.
• Bright cymbals and synthesizers? Maximum stress test.

This is why you can "slam drums through distortion and it sounds great" but the same move on a full mix sounds terrible. Drums are mostly noise and low-frequency content. Full mixes push the entire bandwidth.

What Changed (And What Didn't)

Early digital audio tools often couldn't deliver what their controls promised. A "saturation" knob might add harmonics in theory, but in practice half those harmonics were aliased garbage. Producers learned to work around this without understanding why.

Modern well-designed plugins actually deliver on their promises. That saturation knob adds real harmonics that behave like analog harmonics. The compressor attack time is actually the attack time.

But not all modern plugins are well-designed. Plenty of current products still ignore these problems. Some very expensive, heavily marketed plugins from major companies fail basic anti-aliasing tests. The name on the box doesn't guarantee quality.

The Point

You don't need to understand the math to benefit from this knowledge. Just understand these principles:

1. Distortion creates frequencies that might not fit in your digital system. The more distortion, and the higher the input frequencies, the worse the problem.
2. When frequencies don't fit, they bounce back as permanent garbage. This sounds harsh, brittle, and "digital" in the bad way.
3. This affects both the sound AND how your dynamics plugins make decisions. Sidechain aliasing makes compressors and limiters behave erratically.
4. Good plugins solve this problem. Cheap or lazy ones don't. Price isn't always a reliable indicator.
5. The problem accumulates through your chain. Each processor can make it worse for the next one.
6. Once it's there, it's there forever. No EQ or processing will remove aliasing.

Knowing this, you can make smarter choices: about which plugins to trust, how hard to push them, and why some combinations sound great while others fall apart.

Aliasing

When frequencies that are too high for your system bounce back down as wrong, dissonant frequencies that can't be removed.

Nyquist Frequency

The speed limit—half your sample rate. At 44.1 kHz, it's about 22 kHz. At 48 kHz, it's 24 kHz.

Sample Rate

How many snapshots of audio your system takes per second. 44,100 (CD quality), 48,000 (video standard), 96,000, and 192,000 are common.

Oversampling

Temporarily running at a higher sample rate inside a plugin so there's more room for harmonics to exist without bouncing. Usually 2×, 4×, 8×, or higher.

Linear Processing

Operations that don't create new frequencies—EQ, reverb, delay, volume. These can't alias.

Nonlinear Processing

Operations that create new frequencies—saturation, compression, limiting, clipping. These can alias if not designed carefully.

Sidechain

The part of a compressor/limiter that listens to the audio and makes decisions about when and how much to compress. If this aliases, the plugin makes wrong decisions.

True Peak

The actual loudest point of the continuous audio waveform, not just the loudest sample. Properly detecting this requires anti-aliased sidechain processing.

Harmonics

New frequencies created by distortion, typically at multiples of the original frequency (2×, 3×, 4×, etc.). These are what give analog gear its character—when they don't alias.