White noise
Equal power at every frequency, generated a sample at a time. The brightest and most effective masker of the three noise colours, and the one most people find harshest — which is why the other two are here as well.
What white noise actually is
White noise has the same power at every frequency, which is what makes it the reference the other colours are described against. The name is borrowed from white light, which was once thought to contain all visible wavelengths in equal measure. The definition is a statement about the power spectrum and nothing else — any signal with a flat spectrum is white noise, however it was produced.
This is also why it sounds so bright. Human pitch perception is roughly logarithmic: the octave from 10 kHz to 20 kHz is one octave, exactly like the one from 100 Hz to 200 Hz, but it is a hundred times wider in hertz and so contains a hundred times the power in a flat spectrum. Half of all the energy in white noise sits in the top octave alone. That is the hiss, and it is why white noise is fatiguing at volumes where brown noise is not.
For masking, that brightness is a genuine advantage. Speech carries most of its intelligibility between about 500 Hz and 4 kHz, and consonants — the part that makes speech decodable rather than merely audible — live at the top of that range. White noise puts energy exactly there. If the noise you are trying to cover is a conversation through a wall, white beats brown at the same perceived loudness.
White noise and sleep: what is actually supported
The evidence is thinner than the number of products would suggest. Systematic reviews of continuous noise for sleep consistently reach the same conclusion: the studies are small, short, methodologically mixed, and their results point in several directions at once. Some find modest improvements in time taken to fall asleep, particularly in noisy environments. Some find no effect. A few find sleep quality slightly worse, which is not a surprise — sound during sleep does not stop being sound just because you chose it.
What is on firmer ground is the mechanical part. Waking is triggered far more reliably by a *change* in sound level than by a constant one, and raising the noise floor makes every change smaller relative to it. If you live on a bus route, that is a real effect and you do not need a study to notice it. If your room is already quiet, adding noise is adding sound to a quiet room, and it is entirely reasonable for that to make things worse rather than better.
Two practical cautions, both about level rather than about noise itself. Sleeping every night with sound at high volume through headphones is prolonged exposure, and prolonged exposure is how hearing is damaged; keep it at the level of a quiet conversation or below. And infant sound machines have been measured producing output well above safe limits at maximum volume, which is a good reason to put any such device across the room rather than in the cot, and to run it well below full.
The three noise colours, side by side
The slope is the whole difference between them: how fast the power falls as frequency rises. Everything else — how bright it sounds, what it masks well — follows from that one number.
| Colour | Slope | What it is |
|---|---|---|
| White noise | 0 dB/oct | Equal power at every frequency. The reference the other two are shaped from, and the harshest of the three because human hearing is most sensitive where it has the most energy. |
| Pink noise | -3 dB/oct | Equal power per octave, which is how most natural broadband sound is distributed — rain, a waterfall, wind in trees. Reads as softer than white at the same loudness. |
| Brown noise | -6 dB/oct | The integral of white noise — a random walk. Almost all the energy is below 500 Hz, which is why it is the colour people reach for to cover traffic and voices through a wall. |
All three sliders are in the mixer above, so you can compare them directly rather than from a description.
Common questions
What is the difference between white, pink and brown noise?
The slope of the spectrum. White is flat, pink falls by 3 dB per octave, brown falls by 6. In practice white hisses, pink sounds like a shower, and brown sounds like a distant engine. All three are here on their own sliders.
Is it safe to run white noise all night?
At a low level, through a speaker across the room, it is ordinary background sound. The things worth avoiding are high volumes, headphones worn for eight hours at a stretch, and any device close to an infant, where measured output can be far higher than expected.
Is this real white noise or a recording of it?
It is generated by a pseudorandom sequence in your browser, which is white by construction. A recording would be band-limited by whatever captured it and would repeat; this does neither.