White, Pink and Brown Noise in Sound Design: When to Use Each

What is the difference between white, pink and brown noise?

The color of a noise describes how its energy is spread across the frequency range: white is flat, pink falls 3 dB per octave, and brown falls 6 dB per octave. All three are random signals with no pitch; only the tilt of the spectrum changes.

VCV Rack's Noise module, which is free and generates seven colors, documents the slopes plainly: pink has more bass than white and drops 3 dB per octave, red has more bass than pink and drops 6 dB per octave, and blue and violet rise by 3 and 6 dB per octave^[1]. Brown and red are the same thing. The name brown comes from Brownian motion, described by the botanist Robert Brown, not from a color^[2].

One clarification before going further: this article is about generating noise on purpose. If you are trying to remove unwanted hiss or room tone from a recording, that is a different job, covered in how to clean up room noise from vocals.

Why does white noise sound so much brighter than pink noise?

White noise sounds bright because each higher octave contains twice as many hertz as the one below it, and white noise puts the same energy into every hertz. Hearing works in octaves, so the top of the spectrum dominates.

Do the math on the audible range. The octave from 10 kHz to 20 kHz spans 10,000 Hz, while the octave from 20 Hz to 40 Hz spans only 20 Hz. With equal energy per hertz, that top octave carries about half of all the energy in white noise. Pink noise corrects for this by giving every octave the same energy, which is why it sounds balanced and why engineers use it for testing rooms and systems^[2]. Brown noise tilts further still, so most of what you hear is below a few hundred hertz.

Making any color from white noise

Most synths only offer white noise, but you can tilt it yourself. A tilt EQ set to about -3 dB per octave gets you close to pink. A 6 dB-per-octave (1-pole) low-pass filter with a low cutoff gets you close to brown above that cutoff. A high-pass filter at 5-8 kHz turns white noise into the thin, blue-ish fizz used for air. Use your ears and an analyzer; the goal is the sound, not a lab-perfect slope.

Which noise color works best for risers and sweeps?

White noise is the default for risers because its dense top end makes a filter sweep obvious, while pink noise gives a fuller, less harsh build and brown noise works better for falling downlifters. Many producers layer two: white for the rising edge, pink or brown for weight.

The core move is a filter sweep: run noise through a low-pass or band-pass filter and automate the cutoff upward over 8 or 16 bars, while raising the volume and the reverb send. Band-pass with moderate resonance sounds more like a whoosh; low-pass sounds more like air opening up. For full riser, downlifter and impact recipes, see how to make risers and downlifters.

How do you use noise to build or layer a snare?

A synthesized snare is usually a short tuned body plus a burst of filtered noise for the rattle, and a noise layer is the fastest way to make a weak sampled snare snap. The Roland TR-808 works exactly this way: two decaying oscillators for the drum body, plus white noise through an envelope-controlled VCA and a high-pass filter for the 'snappy' part^[3].

1. Start with white noise
   Use a noise oscillator on its own, with no pitched waveform. Set the amp envelope to 0 ms attack, zero sustain and 100-200 ms decay.
2. High-pass it
   Cut below about 1-2 kHz so the layer adds rattle without muddying the snare body. Keep resonance at zero; resonance makes the tail whistle.
3. Shape the tail
   Shorter decay (60-100 ms) for tight trap and drill snares, longer (200-350 ms) for pop and house claps.
4. Blend under the sample
   Bring the noise up until the snare sounds brighter and longer, then back it off 2-3 dB. Group both and compress them together so they hit as one sound.
5. Add space last
   A short room or plate on the noise layer only makes the snare feel bigger without washing out the body.

Can you make hi-hats and cymbals from noise?

Yes. High-passed white noise with a very short envelope makes a usable closed hat, and a longer envelope makes an open hat or shaker. It will sound clean and slightly generic, which is often what a busy trap or house mix needs.

Classic metallic hats are not pure noise, though. The TR-808's cymbal starts from six square-wave oscillators tuned to non-harmonic frequencies and then filters and envelopes them^[4]. That clang comes from inharmonic tones, not randomness. For more metal, blend a noise hat with a few detuned square waves or an FM layer.

• Closed hat White noise, high-pass at 6-9 kHz, 20-50 ms decay, zero sustain. Map velocity to decay so accents ring slightly longer.
• Open hat Same source, 200-500 ms decay. Put it in a choke group with the closed hat so it cuts off when the closed hat plays.
• Shaker Band-pass white noise around 4-6 kHz with 5-10 ms attack. The slower attack is what separates a shaker from a hat.
• Mix check Noise hats pile up fast in the top octave. The hi-hat mixing guide covers the EQ and level side.

How do you use noise for texture and atmosphere?

Low-level noise glues sparse arrangements together and adds a sense of air or age. The trick is keeping it quiet enough to feel rather than hear: start 30-40 dB below your loudest element and raise it only until the track feels less empty.

A few reliable moves: a pink or brown noise bed, low-passed around 2-4 kHz and sidechained to the kick, makes a lo-fi beat breathe. Band-passed noise with slow cutoff modulation from an LFO sounds like wind and works under ambient intros. Stereo noise with different random signals in each channel, which Logic's Test Oscillator calls decorrelated noise, sounds wide without any widening plugin^[5]. If you would rather start from recordings, free FX and foley packs include vinyl crackle and room tone you can layer the same way.

How do you mix with a pink-noise reference?

Play pink noise at a fixed level, solo each track against it, and set each track's gain so it is just barely audible above the noise. Remove the noise and you have a quick, surprisingly balanced rough mix. Sound On Sound described the method step by step in a 2014 walkthrough^[6].

Pink noise works for this because its equal energy per octave is closer to the tonal balance of a finished mix than white noise, so no single frequency range gets favored. Logic Pro users can generate it with the built-in Test Oscillator, whose noise waveforms Apple notes are useful for tuning rooms and speakers^[5]. In other DAWs, use any test-tone plugin or a bounced pink-noise file.

1. Set the noise level
   Put pink noise on the master or its own track at a comfortable mixing level. In the SOS example, noise at -11 dB read about -9 dB RMS on a K-14 meter and left around 11 dB of peak headroom^[6].
2. Keep faders at unity
   Balance with clip gain or a gain plugin at the top of each channel so faders stay free for automation later.
3. Solo one source at a time
   Raise its gain until you can just hear it poking through the noise, then move on. Treat groups such as backing vocals as one source.
4. Watch the extremes
   Sub kicks and hi-hats can seem masked by the noise; resist boosting them once they are audible. For vocals, listen for the whole line, not just the consonants^[6].
5. Remove the noise and compare
   Bypass the noise, listen, and nudge obvious outliers. Then compare against a commercial track using a reference track in your DAW.