Pink Noise

Steady rainfall rather than static. Equal energy in every octave, measured here at −3.07 dB per octave — and the subject of the most widely misquoted study in this field.

What Pink Noise Actually Sounds Like

Pink noise is the colour people describe without reaching for machinery. Steady rain on a roof, wind moving through a tree in full leaf, surf a few hundred metres away, a shower running behind a closed door. Where white noise sounds electrical and brown noise sounds geological, pink noise sounds like weather.

That is not a coincidence. A great many natural processes produce spectra close to 1/f — the mathematical description of pink noise — and the ear appears to treat that slope as unremarkable in a way it does not treat a flat spectrum. Whatever the reason, in blind comparisons pink noise is the colour listeners most often call natural, and it is the one that provokes the fewest strong opinions in either direction.

It is also the most balanced of the five in a literal sense: you can hear detail at the top and weight at the bottom simultaneously. Brown noise gives you only the bottom; white noise gives you a great deal of the top; pink noise is the one where nothing is obviously missing.

Equal Energy Per Octave — and the Measured Slope

Pink noise falls by 3 decibels per octave. The useful way to understand that number is not as a slope but as a consequence: at −3 dB per octave, every octave band contains the same total power. The octave from 100 to 200 Hz, the octave from 1 to 2 kHz, and the octave from 8 to 16 kHz all carry an equal share. White noise cannot say that — its top octave alone carries half of everything.

Measured over 120 seconds of this player's output at 48 kHz with Welch's method and a least-squares fit from 100 Hz to 10 kHz, the slope came out at −3.07 dB per octave. More usefully, the response is straight to within ±0.12 dB from 50 Hz all the way to 6 kHz — the flattest and most accurate of the five colours generated here, by a comfortable margin.

Equal energy per octave is also why pink noise, and not white noise, is the standard test signal in audio engineering. Feed pink noise into a PA system and a third-octave or octave-band analyser reads a flat line, so any bump or dip you see is the room or the rig rather than the signal. Room correction, loudspeaker alignment and live-sound ringing-out are all done with pink noise for exactly this reason. It is the only colour on this site with a formal engineering role in everyday professional use.

As with the other four, the output is normalised to −20 LUFS using ITU-R BS.1770-4 K-weighting so switching colours does not change perceived loudness, and the two channels are generated from independent random sequences rather than duplicated.

What Pink Noise Is Genuinely Good At

Being the default. If you do not know which colour you want, this is the one to start with. It has enough low-frequency weight to take the edge off traffic, enough high-frequency content to blunt speech, and not enough of either to be actively unpleasant. It is the least likely of the five to be wrong.

Long sessions without fatigue. Pink noise is dark enough to tolerate for hours but not so dark that it disappears on ordinary speakers. In practice it is the colour that most often survives the transition from "this is nice" to "I have been using this every night for a month".

Standing in for rain. Recorded rain has a spectral tilt very close to pink, minus the individual droplet transients. If you like rain sounds but find that the recorded loops start to repeat on you, pink noise gives you a large part of the same character with no loop point at all.

Rooms and speakers you do not control. Because its power is spread evenly across octaves, pink noise degrades gracefully. A phone speaker loses the bottom octave or two and what remains still sounds like pink noise; brown noise under the same conditions barely survives.

The Slow-Wave Sleep Study, Read Properly

Almost every article about pink noise cites one paper, and almost none of them describe it accurately. It is worth setting out what it actually did, because the gap between the study and the way it is used to sell things is unusually wide.

The paper is Papalambros and colleagues, Acoustic Enhancement of Sleep Slow Oscillations and Concomitant Memory Improvement in Older Adults (Frontiers in Human Neuroscience, 2017). The reported result is real and interesting: participants showed enhanced slow wave activity, increased sleep spindle density and amplitude, and improved overnight word recall — about 27% improvement with stimulation against about 6% with sham.

Now the details that matter. The stimulus was 50-millisecond pulses of pink noise, delivered in blocks of five, with roughly 1.2 seconds between pulses and about six seconds of silence between blocks. The timing was not arbitrary: pulses were phase-locked to the up-state of the participant's own slow oscillations, which requires live EEG driving a real-time algorithm. There were thirteen participants, all cognitively healthy older adults with a mean age of 75, studied in a sleep laboratory.

So the intervention was closed-loop acoustic stimulation timed to a brain rhythm — a laboratory technique — and pink noise happened to be the click it used. It was not "play pink noise while you sleep". Playing pink noise continuously overnight is a different intervention that this study did not test, and there is no basis for transferring the memory result to it. A sample of thirteen also means the effect size should be treated as provisional regardless.

This is not a reason to avoid pink noise. It is a reason to be sceptical of any product page — including any app listing — that cites this study as evidence that its continuous pink noise track improves memory. That claim has not been tested.

What Pink Noise Is Bad At

Heavy low-frequency masking. Pink noise has about 3 dB per octave more bass tilt than white noise and about 3 dB per octave less than brown. Against a lorry idling outside or a neighbour's subwoofer, that middle position is the wrong place to be: you will end up raising the volume to get low-frequency coverage and taking a lot of unnecessary midrange with it. Brown noise solves this at a fraction of the level.

Being decisive. The same balance that makes pink noise a good default makes it the colour least likely to be the best answer to a specific problem. If you have identified exactly what is bothering you — rumble, voices, a high whine — one of the other four will usually beat it.

Carrying the research. Beyond the closed-loop stimulation work described above, the evidence for continuous pink noise is thin. The 2021 systematic review in Sleep Medicine Reviews covering continuous broadband noise as a sleep aid rated the overall evidence as very low quality. The 2024 meta-analysis in the Journal of the American Academy of Child and Adolescent Psychiatry that did include pink noise found a small benefit on laboratory attention tasks for people with ADHD or high ADHD symptoms, and none for anyone else. Both are worth knowing; neither supports strong claims.

Nothing on this page is medical advice, and pink noise is not a treatment for anything.

Using It Well

Start lower than feels right. Pink noise is easy to over-set because it is pleasant — there is no unpleasant edge to warn you that you have gone too far. Set it to the point where the sound you are covering stops being distinct, and leave it there.

Use it as a baseline for comparison. Because it sits in the middle, pink noise is the sensible reference when you are working out which colour you want. Listen to it, then step one way to brown and the other way to white, and the direction you prefer will be obvious within a minute.

Web player and app. The player above runs 30-minute free sessions in the browser with no install. The Android app generates the same signal with no session limit, plus a sleep timer, offline playback, background play, and the option to layer rain or thunder on top.

How It Compares With the Other Colours

Pink noise is the midpoint of the set, which makes it the easiest one to describe by reference to its neighbours. White noise is 3 dB per octave brighter and much better at making speech unintelligible. Brown noise is 3 dB per octave darker and much better against rumble. Green noise is built on a pink foundation with a broad boost around 500 Hz, so it is the closest relative of the five. Blue noise is pink's mirror image, tilting upward at +3.08 dB per octave instead of down.

If you only ever use one colour, this is the reasonable one to pick.

Frequently Asked Questions

Related pages

• Brown Noise for Sleep — The free web player, plus the full guide to brown noise for sleep, focus, and tinnitus.
• Brown Noise — The deepest colour, measured at −5.98 dB per octave — best for traffic rumble and long overnight sessions.
• White Noise — The flat reference colour, measured at 0.00 dB per octave — the best of the five at making speech unintelligible.
• Green Noise — The mid-frequency colour — a pink base with a broad 500 Hz emphasis, and the only one of the five with no standard definition.
• Blue Noise — The bright end of the spectrum at +3.08 dB per octave — an engineering signal that is genuinely useful against high-pitched noise, and a poor choice for sleep.
• White vs Pink Noise — Flat per hertz against flat per octave — two kinds of even, why the ear prefers the second, and why no trial has ever ranked the two.
• Brown vs Pink Noise — 2.91 dB per octave apart — indistinguishable in the bass, nineteen decibels apart at the top, and one of them collapses on a phone speaker.
• Brown Noise App — Which brown noise app to install, whether it is really free, and the five things an installed app does that a browser tab is not allowed to.