Do you actually prefer brown noise?
Brown noise has a reputation, and a reputation is an expectation effect. In audio those are large — which is why blind listening is standard practice in audio research and almost never done by the people writing noise guides. Five comparisons below, labels hidden, levels matched. Two minutes.
You state what you think you prefer before you hear anything. That is the part that makes the result worth something: the interesting outcome is not which colour wins, it is whether it is the one you predicted.
Before you start
Two questions, and the second one is the whole point — stating your expectation first is what makes the result worth anything.
What would you use it for?
Which do you think you prefer?
Guess. The test compares this against what you actually pick.
Five comparisons, about two minutes. Headphones give a more meaningful result than a laptop speaker, because the colours differ most in the frequencies a small speaker cannot reproduce. Nothing is recorded or sent anywhere.
Runs entirely in your browser. Nothing recorded, nothing sent. Last updated 2026-10-07.
How the test is built
Six colours, five rounds, king of the hill. The winner of each comparison carries forward to meet the next challenger, which produces a definite result in 5 comparisons instead of the fifteen a full round-robin would need. The order is deliberate: the canonical brown-versus-white question first, then the middle ground, then the gentle options, and the bright outlier last — the people who still pick blue at the end are usually masking something high-frequency, and that is worth surfacing to them.
The levels are loudness-matched, not peak-matched. This is the part most informal comparisons get wrong, and it decides the outcome. Every colour on this site is normalised to the same peak, but equal peak is not equal loudness: the ear is far less sensitive below a few hundred hertz, and brown noise puts almost all of its energy there. Left alone, a blind test would mostly measure which option is louder, and louder reliably wins — the tool would systematically pick bright colours and call it preference.
| Round | Challenger | Loudness trim |
|---|---|---|
| starts as champion | Brown Noise | ×2.60 |
| round 1 | White Noise | ×1.00 |
| round 2 | Pink Noise | ×1.70 |
| round 3 | Green Noise | ×1.40 |
| round 4 | Grey Noise | ×1.00 |
| round 5 | Blue Noise | ×0.72 |
The trims are approximations derived from the shape of the equal-loudness contours, not per-listener measurements, and they cannot be exact — equal loudness depends on absolute level and on your own hearing. They are published here for the same reason the filter settings are: a comparison tool that hides its own weighting is not a comparison tool.
What the result does not tell you
Preference is not effectiveness. The test measures which colour you reach for, and that is a different question from which one helps you sleep or concentrate. The two come apart more often than people expect. The only way to settle it is two weeks with something countable — hours slept, or tasks finished — because subjective impressions of your own focus are unreliable, and especially so when you have already read that a colour is supposed to work.
Five comparisons is a preference, not a measurement. It is single-blind, on whatever hardware you happen to have, with no repetition and no control for order effects. That is enough to be interesting and not enough to be evidence.
About a third of people do worse with noise than without it. A 2024 study of 43 children with diagnosed ADHD found roughly that proportion performed worse with noise exposure, with the negative response tracking hyperactivity and impulsivity rather than inattention. If the winner of your test does nothing for you, that is a real result rather than a failure of technique.
And the reason this page exists. The 2024 meta-analysis of white and pink noise in youth with attention problems — 13 studies, 335 participants, a small pooled effect of about 0.25 — lists as one of its main limitations that participantscannot be blinded to which noise they are hearing, so expectation bias cannot be excluded from the literature. That is precisely the gap a blind comparison closes, and it is why hiding the labels is not a gimmick. The same analysis notes that no study in it examined brown noise at all.
Frequently asked questions
What does a blind noise test actually tell you?
Which colour you reach for when its reputation cannot influence you. Brown noise in particular carries a strong reputation from ADHD communities, and reputation is an expectation effect — in audio those are large and well documented, which is why blind listening is standard in audio research and almost never done by the people writing noise guides. The test does not tell you which colour helps you sleep or concentrate; preference and effectiveness come apart more often than people expect.
Why are the levels adjusted instead of being identical?
Because identical level is not identical loudness. The ear is far less sensitive below a few hundred hertz, so brown noise — which puts almost all its energy there — sounds much quieter than white noise at the same measured peak. An unadjusted test would mostly measure which option is louder, and louder reliably wins. Each colour therefore gets a loudness trim approximated from the equal-loudness contours. It is an approximation, not a per-listener measurement, and it cannot be exact because equal loudness depends on absolute level and on your own hearing.
Does it matter whether I use headphones or speakers?
Yes, and it can change the result outright. The colours differ most in exactly the frequencies a laptop or phone speaker cannot reproduce — brown noise over a phone speaker is a thin shadow of brown noise over headphones. If you take one result seriously, take the headphone one.
Is the test blind in both directions?
The options are labelled A and B only, nothing on screen is tinted with the colour's own palette while a round is running, and which side carries the previous winner is shuffled each round. It is single-blind: you cannot tell which is which, though the page obviously knows. A double-blind design would need a second person running it.
Are my answers collected?
No. The test runs entirely in your browser, nothing is transmitted, and there is no record of your result once you close the page. That also means this page cannot tell you how your answer compares to other people's — there is no data to compare against, and inventing a figure would be worse than not having one.
What should I do with the result?
Treat it as a starting point rather than a conclusion. Play the winner for two weeks and track something countable — hours slept, or tasks finished — because that is the only thing that separates a preference from an effect. Keep the volume at 45 to 60 dB measured at your pillow, which matters more than the colour.