AubiomixVisit Website →
← Back to blog

Aliasing Distortion: Hear the Metallic Artifacts, Then Fix Them

October 15, 2026
Aliasing Distortion: Hear the Metallic Artifacts, Then Fix Them

Aliasing distortion is what happens when frequencies above the Nyquist limit get mathematically folded back into the audible range, creating harsh, metallic, inharmonic artefacts that weren't in the original sound. It's usually caused by nonlinear processing like saturation, clipping or waveshaping generating harmonics your system can't represent cleanly. The fastest way to check if you've got a problem: flip on your plugin's oversampling setting and A/B listen, or run a sine sweep through the chain and watch a spectrogram for anything that doesn't belong.

TL;DR:

  • Test a suspect nonlinear stage with a 20 Hz to 20 kHz sine sweep and spectrogram; downward bending energy after Nyquist reveals foldback.
  • Toggle oversampling at 2x and 4x, then compare short renders; disappearing or shrinking foldback confirms the plugin as the source.
  • Use 2x for gentle saturation and 4x or higher for aggressive distortion on bright, complex material; 8x is usually reserved for extreme waveshaping.
  • Enable oversampling only on confirmed problem stages, since higher factors raise CPU use and may add latency; a short isolated render aids comparison.
  • Once foldback is baked into a render, EQ and mastering cannot separate it from legitimate audio, so prevent aliasing at the nonlinear stage.
Aubiomix
Catch Mixing Issues Before Release
Upload your song to Aubiomix for detailed feedback and actionable steps on its mixing and mastering.
Get feedback on your mix

Table of Contents

What is aliasing: Nyquist, sampling theorem and the foldover effect

Every digital audio system has a speed limit. It's called the Nyquist frequency, and it's exactly half your sample rate. At the standard 44.1kHz, that ceiling sits at 22.05kHz, just above the edge of human hearing. Anything above that line can't be captured accurately: the system doesn't throw it away, it misrepresents it as something else entirely.

That "something else" is aliasing. A frequency above Nyquist doesn't disappear, it folds back down into the audible band at a lower, unrelated frequency, and that new frequency has nothing to do with the harmonic series of your original sound. It's mathematically unrelated, which is exactly why it sounds wrong to our ears.

Here's the arithmetic, worked through simply:

  • At a 44.1kHz sample rate, the Nyquist limit is 22.05kHz.
  • A component above that limit reflects back across the Nyquist point, landing at a lower frequency that has no harmonic relationship to the sound that created it.

The fold happens because aliases repeat at intervals of the sample rate itself, and the apparent frequency you actually hear is always the true frequency folded back into the range between zero and the Nyquist point. Push a tone high enough above the ceiling, and it reappears somewhere you never asked it to be, often right in the middle of your mix where it clashes with everything else going on.

This is why aliasing feels so disorienting the first time you notice it. A bright pluck synth or a saturated bass can suddenly throw out tones that don't track with the pitch you played, because those tones were never part of the intended harmonic content. They're reflections, and reflections don't follow the rules of the thing they're reflecting.

How aliasing sounds and where to listen for it

Aliasing has a signature, and once you've heard it, you'll start catching it everywhere. It tends to sound metallic, brittle and slightly inharmonic, like a bell that's been struck slightly out of tune with itself. Some engineers describe it as "digital grit" or a faint ringing that sits on top of the sound rather than inside it.

It shows up most readily in a handful of predictable places:

  • Cymbals and hi-hats pushed through heavy saturation or bus compression.
  • Bright synth leads run through distortion, bitcrushers or aggressive waveshaping.
  • Bowed strings and other instruments rich in high-frequency harmonic content.
  • Distorted guitars, particularly amp sims with high-gain settings.
  • Mastering chains where multiple nonlinear stages compound on top of one another.

A 5kHz tone pushed through a hard clipper can generate harmonics at 10kHz, 15kHz, 20kHz and beyond, and once those harmonics cross the Nyquist ceiling, aliasing is the result. That's a useful mental model: the more aggressive the nonlinearity and the brighter the source material, the more harmonic content gets pushed into the danger zone.

Visually, aliasing leaves fingerprints you can spot on a spectrogram or FFT analyser. Watch for energy appearing at frequencies that don't move proportionally with pitch, or faint diagonal lines that seem to "bounce" off the top of the display rather than continuing upward. A sine sweep is the cleanest way to demonstrate this: sweep a pure tone from low to high while watching the display, and any aliasing will show up as spurious energy reflecting downward once the sweep crosses Nyquist, visible as a mirrored line heading the wrong way.

How to test, locate and measure aliasing in your session

Finding aliasing doesn't require specialist gear, just a methodical approach and a spectrogram you trust. Here's a repeatable process:

  1. Run a sine sweep through the suspect stage. Feed a sweep from roughly 20Hz to 20kHz into the plugin or chain you're worried about, with a spectrum analyser or spectrogram open on the output.
  2. Watch for foldback. Legitimate harmonic content moves upward in a straight line as the sweep rises. Aliasing shows up as energy that bends back downward once the sweep crosses Nyquist, a visual mirror effect that's hard to miss once you know what to look for.
  3. Toggle oversampling and compare. Switch the plugin's oversampling between off, 2x and 4x while the sweep plays. Oversampling raises the internal Nyquist ceiling temporarily, so if the foldback disappears or shrinks when oversampling is enabled, you've confirmed the source.
  4. Try a short render at a higher project sample rate. Bouncing a test section at 88.2kHz or 96kHz and listening back can reveal whether the harshness softens, which points squarely at aliasing rather than an EQ or dynamics issue.
  5. Isolate stage by stage. In a chain with multiple plugins, mute and unmute inserts one at a time while watching the analyser, so you can pin down exactly which stage is responsible rather than treating the whole chain as guilty.
  6. Check aux sends and parallel chains separately. It's easy to assume the main signal path is the culprit while a parallel saturation return quietly introduces its own aliasing untouched.

Pro Tip: Render a short, isolated section (just a few bars through the suspect plugin) rather than analysing a full mix bounce, because isolating the stage makes foldback patterns far easier to spot on a spectrogram.

This kind of diagnostic work is exactly what separates a guess from a fix. Once you know which stage is generating the problem, the remedy is usually straightforward, and you won't waste time applying EQ to a problem that was never a frequency balance issue in the first place.

Practical prevention and mitigation strategies

Once you've located the source, the fix nearly always comes down to giving the nonlinear process more headroom above Nyquist before it does its damage, and that's what oversampling does.

Oversampling temporarily raises a plugin's internal sample rate, pushing the Nyquist ceiling higher so that the harmonics generated by saturation or clipping have room to exist without folding back.

  • 2x is often enough for gentle saturation or mild tape emulation.
  • 4x suits more aggressive distortion, particularly on material with dense high-frequency content.
  • 8x tends to be reserved for specialist cases such as extreme waveshaping or heavily driven distortion units.

Practitioner reports suggest aggressive distortion on complex, bright material typically needs 4x oversampling or higher, while simpler saturation often sounds clean at 2x. That's a useful rule of thumb when you're deciding where to spend your CPU budget.

There's a trade-off, though: higher oversampling factors cost CPU and can introduce latency, which matters during tracking or live monitoring. Raising your entire project's sample rate to 88.2kHz or 96kHz is an alternative that avoids per-plugin latency differences, though it demands more processing power across the whole session rather than just the offending stage. Our guide to oversampling in mixing explains why targeted, per-plugin oversampling can be more efficient than applying it across an entire session.

Placing a gentle low-pass filter immediately after an aggressive nonlinear stage, before the signal hits the next plugin in the chain, also helps contain harmonic content that's crept close to the Nyquist edge.

On the research side, newer anti-aliasing methods such as ADAA and its refinements use continuous-time approximations to substantially reduce aliasing from nonlinear waveshaping without needing to run at a higher sample rate at all. These techniques are promising for memoryless nonlinearities, though they come with their own computational trade-offs and aren't yet universal across every plugin you'll encounter.

Practical prevention and mitigation strategies — overview diagram

Mixing workflow guidance and a troubleshooting checklist

A sensible approach to oversampling isn't "turn it on everywhere." That burns CPU for no audible benefit on stages that were never going to alias in the first place. A tighter workflow looks like this:

  • Only enable oversampling on the specific plugin where you've actually heard or measured a problem, not across the whole session by default.
  • Confirm the fix with a short render and an A/B comparison before committing, rather than trusting your ears alone mid-session.
  • Place a minimal low-pass filter directly after any aggressive nonlinear stage to protect whatever plugin comes next in the chain.
  • Avoid switching on oversampling for aux sends and parallel return chains as a blanket habit: a filtered send or a single oversampled stage earlier in the signal path usually does the job with less overhead.
  • Revisit your session's sample rate choice if you're consistently fighting aliasing across multiple tracks rather than one isolated plugin.

This kind of targeted approach keeps your session responsive while still catching the handful of stages that genuinely need the extra headroom.

Why prevention matters more than correction

Aliasing is one of the few mix problems that's genuinely permanent once it's there. Once those fold-back frequencies are baked into your rendered audio, they're mathematically indistinguishable from legitimate content: no EQ move, no corrective plugin and no amount of mastering can separate an alias from the real signal it's tangled up with. Prevention at the nonlinear stage is the only reliable cure, which is exactly why we check for it early rather than leaving it for a mastering pass to catch.

Our reports provide measured, actionable feedback on mixing and mastering, helping producers identify areas to work on with tools like Spotify Audio Feature Analysis for Developers that illuminate how high-frequency content affects audio delivery.

— AubioMix

Get a mix report that catches aliasing before release

Upload a mix to AubioMix and we'll provide detailed feedback and actionable steps on your mix.

Aubiomix

  • Reports provide actionable guidance on mixing and mastering.
  • A 1 Mix Review costs £4.99 one-off, or subscribe to Professional at £19.99 per month for ongoing feedback across your sessions.

Check pricing and plans to get a report on your next mix before it goes out the door.

FAQ

How does aliasing sound?

Aliasing tends to sound metallic, brittle or slightly inharmonic, like a faint ringing layered on top of the sound rather than part of it. It's most noticeable on bright cymbals, distorted synths or heavily saturated material, where the artefact doesn't track with the pitch of the note being played.

What is FXAA aliasing?

FXAA stands for Fast Approximate Anti-Aliasing, a technique used in computer graphics and game rendering to smooth jagged edges on screen, not an audio concept. It belongs to the visual aliasing side of signal processing rather than the audio foldover described throughout this piece.

What is aliasing in an image?

In images, aliasing appears as jagged or stair-stepped edges where a smooth diagonal or curved line should be, caused by sampling a continuous image at too low a resolution. Antialiasing techniques like prefiltering and supersampling reduce this by softening transitions before the image is sampled.

What is an example of aliasing in real life?

A classic visual example is a car wheel or helicopter rotor that appears to spin backwards or stand still on camera, because the camera's frame rate doesn't sample the rotation fast enough to capture its true speed. In audio, the equivalent is a bright synth or distorted bass producing a metallic tone that has no relation to the note being played, caused by harmonics folding back from above the Nyquist limit.

Do all distortion plugins cause aliasing?

Not every distortion plugin causes audible aliasing, since some apply internal oversampling or anti-aliasing filtering automatically before processing. Whether it's a problem in practice depends on how aggressively the plugin is driven and how much high-frequency content is already present in the source material.

Sources

For readers who want to go deeper, Columbia's Music and Computers primer covers Nyquist and foldover in detail, while Sound on Sound's oversampling guide is a solid practitioner reference. For the research side, the DAFx paper on continuous-time convolution and CMU's chapter on aliasing explain the maths behind foldback in more depth.