Blue Noise

The mirror image of pink noise: power rising at +3.08 dB per octave. Genuinely useful — mostly in image processing rather than in bedrooms.

What Blue Noise Actually Sounds Like

Blue noise is thin, airy and unmistakably bright. The closest everyday comparisons are a steam valve venting, an aerosol can held down, the spray of a shower with the drumming on the tray removed, or the long tail of a cymbal after the strike has gone. It is the sound of the very top of the spectrum with the body taken out from underneath it.

If white noise is described as sharp, blue noise is what sharp sounds like when you commit to it. It has almost no weight at all: nothing to sit under the sound, nothing to fill the room. On headphones it appears to hover somewhere above and slightly outside your head, which is a genuinely distinctive spatial impression and part of why some people find it interesting to listen to for a minute or two.

Two minutes is roughly the honest recommendation. This is not a sound most people want running for eight hours, and the rest of this page explains why — and what it is actually for.

The Spectrum: Rising at +3.08 dB Per Octave

Blue noise is pink noise turned upside down. Where pink falls at 3 decibels per octave, blue rises at 3 decibels per octave — its power spectral density is proportional to frequency rather than inversely proportional to it. Doubling the frequency doubles the power per hertz.

Measured over 120 seconds of this player's output at 48 kHz, using Welch's method and a least-squares fit from 100 Hz to 10 kHz, the slope came out at +3.08 dB per octave, with a deviation from a straight line of ±0.43 dB. It is generated by the same filter topology as pink noise with the poles and zeros exchanged, which is why the two are such precise mirrors of each other.

The consequence is that blue noise piles almost all of its audible energy into the 2–8 kHz region — which is, unhelpfully, exactly where human hearing is most sensitive, thanks to the resonance of the ear canal. Two signals with identical RMS levels, one blue and one brown, are not remotely equally loud: the blue one is punishing while the brown one is barely there. This is precisely why every colour on this site is normalised by perceived loudness — −20 LUFS under ITU-R BS.1770-4 K-weighting, with 0.6 dB of spread across all five — rather than by signal level. Without that step, switching to blue would be an unpleasant surprise every time.

Where Blue Noise Genuinely Matters: Not Audio

Blue noise is a real and important term of art, but its home is signal and image processing rather than sound therapy. It is worth knowing this, because it explains why the phrase turns up so often in technical writing while almost never appearing in sleep research.

The canonical reference is Robert Ulichney's Dithering with blue noise (Proceedings of the IEEE, 1988; 76(1):56–79), which established blue-noise dither patterns as the standard for digital halftoning. The reasoning is elegant: when you have to introduce error — quantising a photograph down to black and white dots, for instance — you want that error concentrated at high spatial frequencies, because that is where the human visual system is least able to resolve structure. Push the error where the eye cannot follow it and the picture looks clean.

The same logic recurs across computer graphics and rendering: blue-noise sample distributions for Monte Carlo integration, stippling, and dithered transparency all exist because high-frequency-weighted error is the least visible kind of error. In audio, blue-weighted dither is used for the same reason at the quantisation stage — put the noise where the ear is least attentive.

Notice that the argument for blue noise in every one of those cases is that the recipient will not notice it. That is a very different proposition from deliberately playing it into your ears at a level where you certainly will.

What Blue Noise Is Actually Good For as a Listening Sound

There is a narrow band of genuine use, and it is worth being specific rather than padding it out.

Covering high-pitched intrusions. Masking works best when the masker sits in the same frequency region as the thing you are trying to lose. If the sound that is bothering you is high — a coil whine from a laptop, a squealing fan bearing, a mosquito-pitched electronic whistle, the click of a mechanical keyboard across the desk — none of the darker colours will touch it, and blue noise will. This is the one job at which blue genuinely beats the other four.

Short, task-bounded sessions. A quarter of an hour of blue noise while you deal with a specific noisy neighbour of a machine is a reasonable use. It is not a bedtime sound.

Comparison and calibration of your own ears. Listening to blue noise immediately after brown noise is the fastest way to understand what "spectral tilt" actually means, and it makes the other four colours much easier to tell apart afterwards.

Blue noise is available in the Android app as well as in this web player, so if it does turn out to be the one that suits your particular problem you can use it offline and on a timer like any of the others.

What Blue Noise Is Bad At

Sleeping. Every property that makes blue noise useful in dithering makes it a poor overnight companion. It is bright, it is fatiguing, it has no low-frequency weight to settle behind, and there is no research base of any kind supporting it as a sleep sound. If a page recommends blue noise for sleep, it is almost certainly listing colours for completeness rather than telling you anything.

Low-frequency noise of any kind. Traffic, HVAC, footsteps overhead, a neighbour's bass — blue noise does nothing about any of them, and turning it up will make your room louder without making theirs quieter. Use brown noise instead.

Hours of exposure. Because blue noise concentrates its energy in the region where the ear is most sensitive, and because 3–6 kHz is the band most associated with noise-induced hearing damage, it is the colour where volume discipline matters most. The risk from any sound is a function of level and duration, not of colour — but blue noise is the one that gets you to a given perceived loudness with the most energy in the sensitive band. Keep it quiet, keep it short, and do not fall asleep to it on headphones.

Tinnitus, in most cases. Tinnitus percepts are frequently high-pitched, and a bright masker can sit directly on top of them in a way that some people find helpful and others find makes things markedly worse. Responses vary enormously between individuals and this site cannot predict yours. Most people who use noise for tinnitus masking end up somewhere darker; there is more on that on the brown noise for tinnitus page. None of this is medical advice, and tinnitus is worth discussing with an audiologist.

How It Compares With the Other Colours

Arrange the five by spectral tilt and blue sits at the far bright end: brown at −5.98 dB per octave, pink at −3.07, white at 0.00, and blue at +3.08. Green noise does not belong on that line at all, because it is a mid-frequency emphasis rather than a tilt.

For almost every reason people come to a noise site — sleeping, concentrating, covering a road or a conversation — one of the other four is the better answer. Blue noise earns its place here because it is a real signal with a real definition, because it is the correct tool for a specific and uncommon problem, and because leaving it out would mean pretending the spectrum stops at white.

Frequently Asked Questions

Related pages

• Brown Noise for Sleep — The free web player, plus the full guide to brown noise for sleep, focus, and tinnitus.
• White Noise — The flat reference colour, measured at 0.00 dB per octave — the best of the five at making speech unintelligible.
• Pink Noise — Equal energy in every octave, measured at −3.07 dB per octave — the natural-sounding default and the audio industry's calibration signal.
• Green Noise — The mid-frequency colour — a pink base with a broad 500 Hz emphasis, and the only one of the five with no standard definition.
• Brown Noise — The deepest colour, measured at −5.98 dB per octave — best for traffic rumble and long overnight sessions.
• Blue vs White Noise — Rising at +3.08 dB per octave against flat — the one high-pitched job blue does better, and why white noise already covers most of it.
• Brown Noise for Tinnitus — Masking versus habituation, why the pitch of your tinnitus decides the colour, and where sound therapy actually sits in the 2014 clinical guideline.
• Brown Noise App — Which brown noise app to install, whether it is really free, and the five things an installed app does that a browser tab is not allowed to.