Back to Blog
Guide2026-10-2710 min read

Aliasing and Oversampling: The Hidden Cost of Every Saturator

Distortion creates harmonics, and the ones above Nyquist fold back as inharmonic garbage. How aliasing works, how to hear it with a sine sweep, and when oversampling is worth its CPU and latency.

#mixing#dsp#sound-quality#cpu

The Problem Starts With a Simple Fact

A digital system can only represent frequencies below half its sample rate - the Nyquist frequency. At 44.1 kHz that's 22.05 kHz, at 48 kHz it's 24 kHz. Anything a plugin generates above that line doesn't disappear. It folds back into the audible range at a mirrored frequency. That fold-back is aliasing, and every non-linear plugin you own can produce it.

Why Saturation Creates Frequencies at All

A linear process - a volume change, a standard EQ - can only change the level and phase of frequencies that are already there. A non-linear process bends the waveform itself, and a bent waveform contains new frequencies: harmonics at whole-number multiples of the input. Saturators, clippers, tape and tube emulations, amp simulators and waveshapers are non-linear by design. Compressors and limiters with fast attack and release are too, because changing the gain quickly inside a waveform cycle also creates new frequency components.

Doing the Math Once

Take a 5 kHz sine at 44.1 kHz and push it through a saturator that produces odd harmonics:

  • 3rd harmonic: 15 kHz - below Nyquist, fine.
  • 5th harmonic: 25 kHz - above Nyquist, folds to 44.1 − 25 = 19.1 kHz.
  • 7th harmonic: 35 kHz - folds to 44.1 − 35 = 9.1 kHz.
  • 9th harmonic: 45 kHz - folds to 45 − 44.1 = 0.9 kHz.

Real harmonics are musically related to the note. Aliased ones are not: 9.1 kHz and 0.9 kHz have no harmonic relationship to 5 kHz. Played as music, with many notes and moving pitch, this becomes a grainy, metallic haze and a loss of depth that engineers often describe as "digital" or "harsh" without being able to point at the cause.

How to Hear and See It

The clearest test takes two minutes:

  1. Generate a sine sweep from about 1 kHz up to 20 kHz over ten seconds, at moderate level.
  2. Put the plugin on it with settings you would actually use - drive up.
  3. Watch a spectrogram (most analyzers have one) as the sweep plays.

Real harmonics rise with the sweep as parallel lines above it. Aliasing shows up as lines that move the wrong way - falling while the sweep rises, or bouncing off the top of the display. Compare the plugin at 1x and with oversampling on: the difference is usually obvious.

What Oversampling Does

Oversampling runs the non-linear part of the plugin at a higher internal rate:

  1. Upsample the signal by 2x, 4x, 8x or more.
  2. Apply the saturation. Harmonics now have room up to the new, higher Nyquist frequency.
  3. Low-pass filter everything above the original Nyquist frequency.
  4. Downsample back to the session rate.

The harmonics that would have folded back are now removed by the filter before they can alias. Not perfectly - very strong distortion creates harmonics beyond even 8x - but enough to push the remaining aliasing far down in level.

The Costs

CPU

The processing inside the oversampled section runs N times more often, plus the cost of the filters. A 4x saturator can easily take four to six times the CPU of the same plugin at 1x. Across sixty tracks that's the difference between a session that plays and one that doesn't.

Latency

The up- and down-sampling filters need look-ahead, especially linear-phase ones. Latency from a few samples to a few milliseconds is common. Delay compensation hides it on playback, but it matters while tracking through the plugin.

Filter character

Linear-phase oversampling filters keep phase intact but can pre-ring on sharp transients. Minimum-phase filters have less latency but shift phase near the top of the spectrum. Good plugins choose carefully; some let you pick.

Doesn't Running the Session at 96 kHz Solve This?

It helps, because Nyquist moves to 48 kHz and every plugin gets more headroom for harmonics. But it doubles the CPU of every plugin, not just the non-linear ones, doubles file sizes and disk throughput, and heavy distortion still produces harmonics far above 48 kHz. Targeted oversampling on the handful of plugins that need it is usually the better deal.

A Practical Strategy

  • Track and write at 1x or 2x. Keep latency and CPU low while you're making decisions.
  • Render with high quality. Many plugins offer a separate oversampling setting for offline rendering, or a high-quality mode. Turn it on before the final bounce.
  • Prioritize by content. Bright, sustained material - cymbals, synth leads, distorted vocals, the master bus - aliases audibly. A saturator on a sub bass barely needs oversampling, because its harmonics stay low.
  • Watch your gain staging. More drive means more harmonics means more aliasing. Hitting a saturator 6 dB hotter than intended is a common reason a mix turns brittle.
  • Test new plugins once. Thirty seconds with a sweep tells you how a plugin behaves at 1x, so you know when its oversampling setting matters.

Aliasing isn't a reason to avoid saturation. It's a reason to know which of your plugins handle it well and where to spend the CPU.

Key Takeaways

  • •A well-organized plugin ecosystem saves hours per week and prevents session-killing issues.
  • •Version tracking and systematic backup are the foundations of a reliable studio setup.
  • •ProducerGrid automates plugin scanning, version tracking, and organization so you can focus on making music.

Ready to organize your plugin ecosystem?

ProducerGrid scans, organizes, and tracks all your plugins automatically. Free for personal use.

Download ProducerGrid