White noise vs pink noise: which one actually helps

Short answer: white noise has equal power in every hertz of the spectrum, which puts just over half its total energy above 10 kHz and makes it sound bright and hissy. Pink noise has equal power in every octave, a slope of roughly -3 dB per octave, which is why it sounds like steady rain rather than an untuned radio. For sleeping in a normal home, pink is the better default, because the noise that reaches you through a closed door is mostly low and mid frequency and white noise spends most of its energy in a band where there is nothing to mask. Choose white only when the thing bothering you is itself bright and in the same room as you.

That is the whole decision. The rest of this is why, plus two things I only learned by building both and listening to them at 3am.

The one number that explains white noise vs pink noise

Take the audible band as roughly 20 Hz to 20 kHz. That is a bandwidth of about 20,000 Hz, and about ten octaves.

• White noise spreads power evenly across hertz. The octave from 10 kHz to 20 kHz is 10,000 Hz wide, so it holds about half of white noise's total power. Everything above 5 kHz holds about three quarters of it.
• Pink noise spreads power evenly across octaves. Each of those ten octaves carries about a tenth of the total. Everything below 500 Hz carries a little under half.

Human pitch perception is roughly logarithmic, so pink noise is the one that sounds spectrally balanced: each octave you hear as a "region" of the spectrum arrives with about the same weight. White noise, to the ear, is a treble signal with a rumour of bass underneath it.

Everything else people say about the two colours is downstream of those two sentences. If you remember nothing else: white is bright because of arithmetic, not because of a recording choice.

Why pink wins indoors: the wasted octave

Masking works locally in frequency. To cover a sound, you need energy near that sound's frequencies. Energy at 14 kHz does almost nothing to hide a bass line two rooms away.

Now think about what actually wakes people up in a flat or a house: footsteps on the floor above, a car pulling away, a boiler, a partner's snoring, the thud of a door, the low-mid mumble of a television through a wall. Building materials, soft furnishings and closed doors all attenuate high frequencies far more than low ones. That is why you hear the bass from a neighbour's party and not the cymbals. By the time a sound has passed through a wall, its top end has largely been removed.

So white noise arrives at the problem with half its power in a band that the problem no longer occupies. I call this the wasted octave.

The wasted octave: in white noise, the top octave holds about half the signal's power but does almost none of the masking work indoors, because the intruding sounds have already had their high frequencies stripped out by the walls and doors between you and them.

You pay for that octave twice. First, it does not help. Second, it is the band your ear treats as "sharp", so it is the part that makes noise feel fatiguing after an hour, and the part that makes you turn the volume down and lose the masking you actually wanted.

The case for white noise is narrower and real: when the intruder is bright and in the room with you. A hissing radiator, a whining fan, sibilant chatter at the next desk, keyboard clatter. Then the high-frequency energy is not wasted at all, it is exactly on target.

What I changed in Tysha after loudness-matching them

When we shipped white and pink as separate layers in Tysha, both are generated live rather than played from a file, so there is no loop point to hear. Pink is white passed through a cascade of one-pole filters that approximates -3 dB per octave. The approximation has some ripple against the ideal slope, small enough that nobody has ever noticed it in a bedroom, and it costs almost nothing to run, which matters when the whole thing has to work offline on an old phone.

The interesting part was not the filter. It was the fader.

We first matched the two colours by RMS level, which is the obvious engineering answer: same total power, same slider position, same loudness. It sounded wrong immediately. Pink at the same measured level sounds distinctly quieter than white, because loudness perception discounts low frequencies heavily. Nearly half of pink's power sits below 500 Hz, where the ear gives it little credit. So users switching from white to pink heard a drop, pushed the slider up to compensate, and ended up with more low-frequency energy in the room than they intended.

I now think of this as the equal-fader fallacy: the assumption that two different noise colours at the same slider position deliver the same perceived loudness. They do not, and the gap is large enough to change behaviour.

What we shipped instead is a per-sound gain trim set by ear at low listening levels, not by matched RMS, so that swapping colour mid-session changes the character of the sound without changing how loud it feels. Combined with a fade on the swap, you can go from white to pink without the room seeming to lurch. If you are building anything with multiple noise colours, do this before you do anything else. It is invisible when it is right and infuriating when it is wrong.

Your phone speaker may flatten the difference entirely

Here is the part most comparisons skip. A small phone speaker produces very little output below a few hundred hertz. Pink noise keeps a little under half of its power below 500 Hz. Strip that away and you have removed most of what distinguishes pink from white.

The practical consequence: on a phone speaker alone, the white versus pink question shrinks to almost nothing. Both come out as a mid-and-treble hiss, and pink will sound thinner than you expect from any waveform diagram. The choice starts to matter properly when the sound is coming out of something with real bass extension: a mains-powered speaker, a soundbar, a decent portable, headphones.

That is a falsifiable claim and easy to test. Play pink noise on your phone's own speaker, then the same pink noise on a proper speaker at the same perceived level. If the second one does not sound like a substantially different sound, I am wrong.

Focus and work: a different trade-off

For concentration the interference is usually speech, and speech intelligibility lives largely between roughly 500 Hz and 4 kHz. Both colours have energy there, so both can work. Two things push me towards pink anyway:

• Fatigue. Anything you keep on for six hours needs to be tolerable at hour five. White noise's top end is the first thing to become tiring, and the tiredness sneaks up rather than announcing itself.
• Headroom. Because pink feels gentler at the same measured level, you can usually run it a touch louder before it becomes intrusive, which means more actual masking where speech lives.

The counter-case: open-plan offices with lots of hard surfaces and bright, close-range noise. Chair casters, keyboards, a colleague's tinny earphones. There, white noise is genuinely a better tool.

How to choose in under a minute

1. Name the specific sound you want covered. Not "noise" — the actual thing. Traffic, upstairs footsteps, a television, a fan, chatter.
2. Is it low or muffled, or does it come through a wall, floor or closed door? Use pink.
3. Is it bright, hissy or sibilant, and in the same room as you? Use white.
4. Sit or lie exactly where you will be, at the volume you will actually use, and switch between them. Judge on how it feels after two minutes, not two seconds. First impressions favour whichever is brighter.
5. Check you can still hear what you need to hear. This one matters most for parents: white noise's energy sits in the same bright band as the beginning of a cry, a latch, a footstep on a stair. If you are listening out for something, the colour that hides it is the wrong colour.

Point five is the reason I keep coming back to pink as a default rather than a preference. Masking is not free. Whatever you cover, you cover for yourself too.

What I am not going to tell you

You will find plenty of pages claiming pink noise specifically improves deep sleep or memory consolidation. I have read some of that literature and I am not going to summarise it here, because I cannot evaluate it properly and because the versions of it that circulate online have usually lost every caveat that mattered. What I can tell you with confidence is what the two signals are, how they behave in a room, and what happens when you build them.

I am also not going to give you a volume number, for adults or for children. That is not mine to give.

Pick the colour whose energy lands where your problem actually is. Then leave the volume as low as does the job.