Why Does White Noise Help You Sleep? The Masking Mechanism, Explained

The short answer

Sound interrupts you because it changes what you can hear. In a quiet bedroom a floorboard or a slammed door lands as a large jump in level against the silence around it, and change is what the auditory system is built to notice. A steady bed of noise fills in that silence, so the same event starts from higher and the jump is smaller. That is masking: a claim about acoustics, not about your brain.

Everything else said about white noise and sleep falls into two buckets. The defensible one is the paragraph above — fewer disruptions reaching you. The unestablished one is that the sound itself acts directly on sleep, making you settle faster or go deeper. Most disappointment with a machine comes from expecting the second and receiving only the first.

Colour, level and distance are yours to tune, and none is a matter of efficacy. The right colour is whichever you can leave running at a low level without noticing it.

• Defensible: steady sound raises the audible floor, so intermittent noise carries less contrast.
• Not established: any direct effect of the sound itself on how fast, long or deeply you sleep.
• Also covered: ADHD, babies, habituation, deeper sleep, and what to try when it does nothing.

The three colours, defined by their spectrum

“Colour” in noise means the shape of the spectrum — how total energy is spread across the frequencies you can hear. The names come from coloured light and refer to nothing audible in the sound. The machine linked at the bottom of this page synthesises exactly three, so these are the only definitions you need.

White noise: roughly equal energy per frequency

White noise carries about the same energy at every frequency in its range — flat on a graph, and the version people mean when they say white noise.

The catch is how flatness behaves in octaves. Each octave spans twice the frequency range of the one below it — 1,000 Hz to 2,000 Hz is twice as wide a slice as 500 Hz to 1,000 Hz — so a spectrum that is flat per frequency puts roughly twice the energy into every higher octave, a rise of about 3 dB per octave. That is why flat white noise sounds bright and hissy: the top end does most of the shouting, and it is the hardest part to tolerate for hours.

Pink noise: energy falling with frequency

Pink noise rolls off as you go up, losing roughly 3 dB per octave. That decline cancels the doubling above, so each octave contributes about the same energy as any other — pink is equal energy per octave. It sounds softer and more even, with the harsh sheen off the top, and people compare it to heavy rain or wind. Preferring it is taste, not performance.

Brown noise: a steeper fall-off

Brown noise falls away faster than pink — commonly described as about 6 dB per octave — so most of its energy sits at the bottom of the range. The same spectrum is often called red noise; “brown” is the name that stuck. It reads as a low rumble rather than a hiss, and it is the one some people can stand when the other two are unbearable. Others turn it up because it seems quiet, which defeats the point: a rumble you have to raise is no longer quietly masking anything.

Why people rate them differently

Tolerance of the high end is not universal. It shifts with age and noise exposure, it depends on how far the source sits from you, and the room changes the answer, since bedding and a closed door absorb treble before the sound arrives. So arguments about which colour wins are arguments about tolerance rather than masking power: all three fill the same floor across the range they cover, and no ranking of them by effectiveness has been established. If white is unbearable, try pink; if pink is still too bright, try brown, or use the tone control below.

How the masking actually works

Three parts make this work, and separating them is useful, because most advice about noise machines merges all three into “sound relaxes you”.

Contrast is the trigger, not absolute level

The auditory system is a change detector far more than a level meter. A hum you stopped noticing an hour ago can be louder than the creak that pulls you back, and the creak is the one that gets through. What matters is the step between the background and the event.

A noise machine moves the background. With a steady bed of sound already there, the door and the floorboard start higher, so the same event is a smaller step above it. The intrusive sound did not change; its contrast did. This is also why turning it up to beat a particular noise backfires: if you have to raise the level until the truck outside disappears, you have not buried the truck, you have added a loud sound and left it sitting on top.

Your ears keep working while you sleep

Sleep does not switch peripheral hearing off. Sound still moves the eardrum and still reaches the inner ear, and the sleeping brain keeps sorting some of it, which is why a crash or a name can pull you out of a sleep that a background television will not. What changes with sleep depth is the threshold and what gets selected, not the plumbing. So the useful framing is not “noise keeps you awake” but “noise is still being monitored for events”, and a steady spectrum produces no events.

Why predictability matters

Novelty is what breaks through, and a sound with no pattern to learn stops asking for attention. A sleep playlist of birds, waves and gusts fits this job worse than a plain hiss, because it is made of transitions, and transitions are what you were trying to stop reacting to.

The machine linked here leans into that by construction: each colour is synthesised from random numbers into a four-second buffer that repeats, so there is no structure to track. One honest consequence is that the loop has no crossfade, so it restarts abruptly every four seconds. With white noise the seam is inaudible, because that signal jumps sample to sample anyway; with the bass-heavy colours you may notice a faint pulse at the loop point.

Two claims, and which one is which

The first claim — a steady sound reduces how much intermittent noise stands out, and therefore how often it intrudes — follows from acoustics and from change detection. That is the defensible one, and it is what this guide has been arguing for.

The second, that the sound itself has a direct physiological effect on sleep, changing how quickly you settle or how deep you go, is not established. Studies have looked at this area, including work on whether particular patterns of sound change measured sleep depth and work on whether steady sound changes reported sleep quality. That research is where the “deeper sleep” and “sleep better” phrasing comes from. It has not produced a settled result you can order for yourself, and anywhere promising a specific sleep stage from a specific colour is selling the unproven half.

The distinction changes what you judge success by. Expect the room to stop throwing surprises at you and the test is concrete: the hallway conversation, the early lorry, the partner coming home late. The tool's own page holds the same position — personal preference, not a medical treatment, and no health claims.

• “It masks disruptions” — defensible, and the mechanism you are actually getting.
• “It has a direct physiological sleep effect” — not established; treat promises of it as marketing.
• Nothing here or in the linked tool is a health claim, a diagnosis or a treatment.

Volume, distance and placement

The machine is useful when it is in the room and out of your attention, which is a question of level and position. Treat this as practical placement advice rather than a medical instruction: this guide states no numeric level limits and no age rules.

On level, keep it low. The volume slider runs from 0 to 100 on a squared curve, which matters more than it sounds — a setting of 50 delivers one quarter of what you get at 100, and 25 is a sixteenth of it. Start near the bottom and raise it only when a specific sound is still winning; if you are straining to hear the neighbours under the hiss, the level is wrong.

On distance, put the source out of the bed: across the room, on a surface, not aimed at the pillow. A speaker at arm's length is a hiss you have to actively ignore, and distance softens the top end before the sound arrives. In-ear buds suit this job badly, since they put the source at the eardrum and seal away the ambient sound instead of raising the floor.

• Tone before volume: if it is harsh, dull the top end with the Softness control or change colour.
• For an infant: place the machine away from the crib rather than beside it, keep the level low, and route cords out of reach. What is quiet enough, and what belongs in a baby's sleep space at what age, are hearing and safe-sleep questions for your paediatrician or public health guidance rather than for a tool page — be suspicious of anywhere that hands you numbers instead.

Does white noise help with ADHD?

What people report is consistent, and it deserves respect as description rather than as a finding. Many people with ADHD say silence is worse than noise — that a quiet room makes every distant creak enormous, and that the unpredictability is the intolerable part. A steady bed of sound takes the unpredictability away, so the environment stops generating events, and some add that a constant background gives the restless part of attention something to hold. Those reports are mostly about waking focus.

The evidence situation, in general terms: research has looked at whether added background sound changes attention and task performance, including work framed around stimulation levels rather than the sound itself, and the results do not line up into a recommendation you could rely on. Much of it concerns daytime concentration, so it does not reach the sleep question at all. Nothing here is a treatment claim, and it does not replace anything a clinician has prescribed.

So treat it as a cheap, reversible experiment. If your household noise is intermittent and your nights are chopped up by it, masking is aimed at a real problem you have, ADHD or not. If a fair trial changes nothing, the honest move is to stop rather than to turn it up.

• Reported: less intolerance of silence, fewer surprises from household noise, an easier time settling.
• Not established: any effect on ADHD itself, and no therapeutic claim is being made here.

White noise for a baby

The baby version asks the same mechanism to do the same job, with more hope attached. What households report is this: the baby stirs or wakes fully when the house makes a sound, and a constant bed of noise means those events stop producing a full startle.

Be careful about the size of the claim. An infant's settling involves feeding, timing, comfort, development and routine, and a sound machine is one variable among many. If you read that it makes babies sleep longer or settles colic, that is the unproven category, and no page, this one included, can promise an outcome for a particular child.

Placement is where the advice genuinely differs. Put the machine away from the crib rather than on or near it, at a low level, with cords unreachable. Choose a tone quiet enough that you can still hear the room, because that is not negotiable, and a machine loud enough to mask a household is loud enough to mask a child. The numbers — what level, what age, what belongs in a sleep space — belong to safe-sleep guidance and your paediatrician rather than to a tool page.

On this machine, the sleep timer offers only 15, 30, 45, 60 or 90 minutes and runs inside the browser tab, so for naps the tab has to stay open and a backgrounded tab may delay the fade.

Habituation, dependence, and how to test where you are

After a few weeks the machine can start to feel mandatory. Two different things sit behind that, and telling them apart is worth doing before you decide what to do about it.

The first is that the machine is doing what it does, and removing it removes the effect. If you live somewhere with intermittent noise and you take the mask away, the events come back at full contrast and interrupt you again. That is not dependence on a sound; that is a noisy room. The tell is that the room is genuinely not quiet.

The second is conditioning, where the sound has become the cue for settling and silence now reads as wrong. That is what people call dependence. It is not automatically bad, since a lot of sleep runs on routine cues, but the two have different exits: a noisy room is dealt with at the room, with masking or with earplugs, while a conditioned cue is only discovered by removing the cue.

How to test which one you are in: three or four nights without it during a period when you know the household is quiet. If nothing changes in that genuinely quiet room, the machine was masking a problem that is not currently present. If something does, that still proves nothing about the sound itself, so try alternating nights, one on and one off, which separates “I need this sound” from “I need this room quiet”.

• Signs worth noticing: you cannot sleep away from home because the room is too quiet, you notice the instant it stops, and you keep raising the level week after week for the same cover.
• Needing a hiss is not a moral failing and not a symptom. It is a trade, and this guide will not call either side of it correct for you.

What to try when it does not help

Plenty of people try a machine, get nothing, and write the idea off. Usually the arrangement failed rather than the principle, so work through these in order.

Too loud. A hiss you have to tune out is a stimulus, and you will spend the night doing that work. Drop the level substantially, remembering the squared slider.

Wrong end of the spectrum. Flat white is bright, and a machine near the pillow is brighter. Switch to pink or brown, or pull the Softness cutoff down. If it sounds like a broken television, that control is the fix rather than a new colour.

Off when the noise is on. If the run ends at midnight and the lorry comes at five, you get the largest contrast available: a freshly silent room and a loud event. Cover the whole noisy window, or cover none of it.

The noise is too big for masking. A shouting street, a loud snoring partner, thin walls at party hours — masking shaves the edges off a creak and will not erase a sound louder than the level you are willing to sleep under. Fix it at the source first.

Sound may not be the cause. Light, temperature, caffeine timing and a phone in bed all produce the same “I keep surfacing” experience with no noise involved. And if what is in the room is snoring or gasping, that is a conversation for a clinician rather than a volume setting.

Or a different strategy: earplugs cut everything at the ear instead of raising the floor; a fan in another room is the same masking principle arriving pre-filtered by walls; and moving bedtime earlier so the noisy part of the night is already finished is underrated. If none of it helps, that is a legitimate result — masking may simply not be the lever your situation has.

Trying it: what this machine actually makes

The linked tool is deliberately small, and knowing its construction tells you what it can and cannot do.

What it generates

Three colours and nothing else: white, pink and brown. No recordings and no scenes — there is no rain, ocean or fan to browse. Each colour is synthesised from random numbers inside the page: white straight from the browser's random source, pink through a small recursive filter of Paul Kellet's design, brown through a leaky integrator that accumulates the random input with a slow leak. Each is peak-normalised into a four-second buffer and looped.

Because the sound is generated where you are reading it, no audio files are downloaded and no streaming is involved; the page states that once the page is loaded the machine runs fully offline. Nothing is recorded and nothing is sent out.

The controls

Start and Stop; three colour buttons; a volume slider from 0 to 100 applied on a squared curve; a Softness control that sets a low-pass cutoff anywhere between roughly 120 Hz and about 26 kHz; and a sleep timer of Off, 15, 30, 45, 60 or 90 minutes.

Softness does not change which colour you picked; it changes how much of the top survives, so it doubles as a fade toward nothing at the bottom of its range. Audio cannot begin until you press Start — a browser rule about autoplay rather than a quirk of this page, since the audio engine is created inside the click.

What it will not do

The timer fades, but it lives in the tab. The page warns that background tabs may be throttled by the browser so the fade can land up to a minute late — a 15-minute setting could let sound run to about 16 minutes — and if the tab closes, the timer goes with it. The loop has no crossfade, so it restarts abruptly every four seconds. There is no seamless all-night playlist, no named scenes, no saved presets and no claim of any sleep benefit.

Frequently asked questions