Skip to content

Noise can make synths breathe

Acoustic instruments carry noise that moves with each note. Give a synth's noise the same envelope and it fuses into the sound instead of sitting beside it.

Updated · 6 min read Lesson 5 of 10 in Sound Design

In short

  1. Acoustic instruments carry noise that starts, changes and stops with each note, and a clean oscillator has none of it.
  2. The ear joins sounds that move together, so noise on the note's envelope fuses with it while a steady hiss track stays separate.
  3. Generate noise inside the voice, filter it with the oscillator and band-pass it to where you want breath, bite or air.
In this lesson (8)

The sterile digital synth

You load a wavetable synth, pick a clean saw and program a chord progression. The tuning is perfect and the filter sweep is smooth, yet it sounds cold, as if it lives inside the computer. You add chorus and a stereo delay. It gets wider. It still sounds sterile.

Acoustic instruments are never pure pitched tone. A violin has the scrape of the bow, a flute the rush of breath, a piano the thump of the hammer. Each of those noises starts with the note, changes with it and stops with it. A clean has none of that, and the ear notices what is missing.

Why it matters

Noise that moves with the note joins it

Adding noise is easy. Making it part of the instrument is the hard part. The ear groups sound by how it behaves over time: components that start together and rise and fall together are heard as one source, and components that do their own thing split off into separate streams (Bregman, 1990).

A vinyl crackle or hiss track running under the synth at a fixed level fails both tests. It starts before the note and keeps going after it, so the ear files it as a second sound beside the synth. A steady background also fades from attention through , so after a while it is only clutter.

The same noise inside the synth voice, shaped by the note's envelope and passed through the same filter, passes both tests. It starts with the attack, falls with the decay and darkens as the filter closes, so the ear hears one instrument with breath in it.

Figure 1. Two notes and their noise, sketched as levels over time. Above, the noise runs at one level through notes and gaps, so it is heard as a second sound. Below, it bursts on each attack, then falls and stops with the note, so it is heard as part of it.

The science

Sines plus noise

A standard way to describe instrument sounds splits them into two parts: a set of sinusoids that carry the pitch, and a noise part that carries everything else, with its own changing level and spectrum. Serra and Smith (1990) built an analysis and synthesis method on exactly this split. A simple version for one synth voice:

y(t)=s(t)+e(t) n(t)y(t) = s(t) + e(t) \, n(t)

Here s(t)s(t) is the oscillator, n(t)n(t) is a noise source, usually filtered, and e(t)e(t) is the envelope that sets how loud the noise is at each moment. With e(t)e(t) held constant you have a hiss track. With e(t)e(t) following the note, or a short envelope of its own on the attack, the noise becomes part of the note.

The two parts also sit differently in the spectrum. The harmonic part is a set of lines at multiples of the note. Noise has no pitch: it is a smooth spread of energy that fills the gaps between those lines, so it adds texture without clashing with the key. Band-passing it decides where the texture lives: low for thump, upper mids for breath and bite, the top for air.

Figure 2. One note split the way a sines-plus-noise model sees it. The oscillator is a set of lines at multiples of 110 Hz. The noise is a smooth band with no pitch, here band-passed to the upper mids, where it reads as breath or bite.

Try it

The in-voice noise test

Hear the difference between noise beside a synth and noise inside it.

0 of 7 steps done

The constant noise track sounds like hiss next to the synth. The in-voice noise sounds like breath or bow inside the note, and the short noise envelope gives every attack a little bite.

Common mistake

The static background loop

A vinyl crackle or rain sample on its own track for the whole song is a common way to add warmth. It is a valid lo-fi colour, but it does not make the synth itself sound played. It is one more layer, it adds steady energy in the upper mids and highs where it can blur , and the ear soon stops attending to it.

The opposite mistake is too much. Noise that is obvious in solo is usually too loud in the mix. It should be something you miss when you mute it, not something you hear first.

Takeaway

Put the noise inside the voice

Treat noise as part of the instrument. Generate it in the synth voice, run it through the same filter, give it the note's envelope or a short one of its own, and band-pass it to the range you want: low for thump, upper mids for breath, the top for air. On a bass, a short burst of filtered noise on each attack adds snap without adding any low end.

3 questions

Check yourself

  1. A hiss track runs at a constant level under a synth. Why does it sound like a separate sound?
  2. In y(t) = s(t) + e(t) n(t), what does making e(t) follow the note do?
  3. Why can noise add texture without clashing with the key of the song?

Sources

Show all 2
  • Bregman, A. S. (1990). Auditory Scene Analysis: The Perceptual Organization of Sound. MIT Press.
  • Serra, X., & Smith, J. O. (1990). Spectral modeling synthesis: A sound analysis/synthesis system based on a deterministic plus stochastic decomposition. Computer Music Journal, 14(4), 12-24.

Next lesson · 6 of 10

Distortion adds size without more volumeSaturation adds harmonics and lifts the average level against the peak, so a part sounds bigger at the same loudness. Level-match to hear it.

Previous: Attack time changes how a part feels

Portrait of Virzy Guns

Written by Virzy Guns

Songwriter and producer, founder of Virzy Guns Production. Now building HealingWave.