The research

White, pink and brown noise, and why the colours have names

22 July 2026

Noise colours sound like marketing. They are not. Each name describes a specific, measurable property: the slope of the sound's power spectral density, which is a precise way of saying how the energy is distributed across frequencies.

White noise is the reference point. Equal power at every frequency, the audio equivalent of white light containing every wavelength.

ColourWhite SlopeFlat MeaningEqual power at every frequency

That sounds like it should be neutral. It is not, because your ears are not neutral. Human hearing groups frequencies into octaves, and each successive octave spans twice as many hertz as the one below it. So a spectrum with equal power per hertz has far more energy in the top octave than the bottom, and white noise is perceived as hissy and bright rather than balanced.

Pink is the one that sounds even

Pink noise drops by 3 dB per octave, which exactly cancels the doubling. Equal power per octave rather than per hertz.

ColourPink Slope-3 dB / octave MeaningEqual power per octave

The result is what most people mean when they picture "noise": steady rainfall, a waterfall at a distance, wind through a large space. It is the distribution that natural broadband sound tends toward, which is probably why it reads as unremarkable rather than as a machine.

Brown is heavier still

Brown noise drops twice as fast, 6 dB per octave. The name has nothing to do with the colour brown. It refers to Brownian motion, because a signal generated by a random walk has exactly this spectrum.

ColourBrown Slope-6 dB / octave MeaningWeighted to the low end

The effect is a deep rumble with the hiss largely gone: heavy surf, a distant engine room, the inside of an aircraft. It is the colour most people reach for when they want something to disappear behind, and it is the one that masks low-frequency intrusions like traffic most effectively.

Which one for sleep

Here the evidence is genuinely better than it is for binaural beats, and it points somewhere slightly disappointing: what helps is masking, not anything specific to a colour.

Ding 2025 pooled twelve randomised controlled trials with 1,301 participants in Sleep Medicine and found sleep quality improved, though the size of the benefit varied by age group and setting. Khandelwal 2025, looking at seven trials with 496 hospitalised adults in Sleep Medicine: X, found sleep efficiency improved, in small trials that were nonetheless well tolerated.

Randomised controlled trials, pooled, in both cases. That is a stronger footing than anything in the binaural literature.

The caveat we would rather state than bury

Almost all of that research was done in hospital wards and intensive care, or with infants. Very little of it tested healthy adults in their own bedrooms.

That is not a small gap. A hospital ward at night is loud, unpredictable and full of sounds you cannot anticipate, which is close to the ideal case for masking. Your bedroom is probably quieter and more familiar. The finding transfers plausibly, but "plausibly" is the honest word, and an app telling you that twelve RCTs prove noise will fix your sleep is quietly swapping settings on you.

And no colour is proven better than another

The mechanism is masking, and any broadband sound masks. What varies is which intrusions it covers best: if what wakes you is low and rumbling, brown covers it better; if it is voices, pink or white does more.

Notice what the studies above do not say. They tested noise against no noise. They did not test brown against pink in the same room, and a study finding an effect for one colour has usually not compared it with the others at all. Any site ranking the colours for sleep is ranking them on something other than evidence.

The practical version

Pick by what you can hear from your bed, not by a chart. Set it just loud enough that the intrusion stops standing out, and no louder: masking works at modest volumes, and sleeping every night under loud sound is a hearing risk that accumulates quietly over years.

And if none of the colours suit, a recording of rain does the same job. It is approximately pink noise with structure, and the structure is the part people tend to prefer.

All posts