White, pink, brown. Which noise, and when.

The names are not marketing. They describe how sound energy is distributed across the frequency spectrum, and the differences are measurable rather than aesthetic.

White noise has equal energy at every frequency. Because human hearing has more sensitivity spread across the high end, that even distribution sounds bright and hissy — closest to untuned radio static.

Pink noise reduces energy by 3 dB per octave as frequency rises. That happens to match how our hearing weights the spectrum, which is why pink sounds “flat” and natural to us. Steady rain and wind through trees are close to pink.

Brown noise (Brownian, after the random walk, not the colour) drops 6 dB per octave. Deep, rumbling, most of its energy low down — a distant waterfall, or a heavy downpour heard from indoors.

Two more turn up often enough to be worth naming. Green noise concentrates its energy in the mid range, around 500 Hz, which is where our hearing is most sensitive and where rivers, surf and rain on leaves sit — so it reads as moving water rather than as static. Unlike the other three it has no agreed technical definition, so implementations differ. Violet (or purple) noise is the exact mirror of brown, rising 6 dB per octave with almost all its energy at the top. It is harsh for general listening and is used mainly in tinnitus masking, which is a clinical matter rather than a self-service one.

The practical consequence is simply which sounds each one masks best. Brown covers low rumble — traffic, machinery, a neighbour's bass. White covers high, sharp intrusions — a beeping device, keyboards, birdsong at dawn. Pink sits between and is the safest default when you do not know what you are masking. Green lands close to the human speech band, which makes it a reasonable pick when the thing breaking your concentration is conversation.

Sleep. Pink or brown. Both are gentler than white over hours, and the low end masks the traffic and building noise that most often wakes people. Small studies have looked at pink noise and slow-wave sleep, but the sample sizes are modest and it is fairer to say the evidence is promising than established.

Focus in an open office. White or pink. You are masking speech, and intelligible conversation is uniquely disruptive to concentration — far more than non-verbal noise at the same volume. Masking the speech band is the goal.

A partner who snores, or street noise. Brown. That energy is low, and you need low energy to cover it.

Tinnitus. Handle with care and read the tinnitus page first. Sound therapy is a genuine clinical approach, but the right sound is individual and should be set up with an audiologist rather than guessed at.

A baby sleeping. Follow paediatric guidance on volume and distance, not a website. Infant hearing is more vulnerable and the safe levels are lower than most people assume.

Every page below carries the same working generator, built with the Web Audio API, and each one plays all five colours — the noise is synthesised in your browser rather than streamed, so it never loops, never buffers, and works offline once loaded.

White noise — the bright, full-spectrum one. Best against sharp, high-frequency intrusions and speech.

Pink noise — the natural-sounding middle ground and the safest default. Rain, wind, waves.

Brown noise — deep and rumbling. Best against traffic, machinery and snoring; the most popular for sleep.

Green noise — mid-focused, around 500 Hz. Sounds like a river; good against speech, poor against rumble. The page also plays violet, for contrast.

They all run through speakers. Unlike binaural beats, noise does not depend on stereo separation, so headphones are optional — and for sleeping, a speaker is far more comfortable.

Use noise when the problem is outside your head: a noisy environment, an unpredictable sound landscape, anything where the goal is to stop being startled. Noise works through speakers, needs no attention, and is comfortable to run for eight hours.

Use binaural beats when the problem is inside your head: you want to shift toward a state rather than block something out. Beats require stereo headphones and are usually run for twenty to thirty minutes rather than all night.

Combining them is fine and common — noise underneath, beats on top — though it is worth noting that a 2025 randomised study found masking sound measurably affects the entrainment response. If you are trying to establish whether beats do anything for you, test them clean first.

Two deeper comparisons: beats versus white noise for the head-to-head, and beats versus the alternatives for isochronic tones and everything else.