Brown noise vs white noise: hear the difference

Brown noise falls about 6 dB per octave while white noise stays flat, so brown sounds like a low waterfall and white sounds like radio static: brown covers rumble, white covers hiss and voices. Slo plays both free in the browser at slonoise.com, one tap apart on this page, and 65.9% of its brown noise sits below 250 Hz against 1.0% of its white noise.

What neither one solves

Both add sound to a room and neither removes any, so a thin wall stays a thin wall and a snoring partner stays audible under either colour. Continuous noise for sleep also carries weaker research behind it than most articles imply. And a background sound that you end up listening to has made the room worse rather than better. If a noise wakes you every night, fixing the source beats covering it.

The spectrum, in one line each

White noise carries constant power spectral density. The ASA/ANSI acoustical terminology database defines it at entry 2.37 as “noise for which the spectrum density or spectrum level is independent of frequency over a specified range”. Equal power in every equal-width slice of frequency.

Brown noise has no entry in that database, which surprises people who assume the colours carry equal standing. The acoustics literature defines it by its slope instead: power proportional to 1/f², a fall of about 6 dB per octave. Voss and Clarke used exactly that 1/f² signal alongside white and 1/f noise in their 1978 study in the Journal of the Acoustical Society of America. The name points at Brownian motion and the botanist Robert Brown, not at the colour brown.

The slope decides what your ear registers. An octave stretching 100 to 200 Hz holds 100 one-hertz bins while an octave of 5,000 to 10,000 Hz holds 5,000, so flat-per-hertz white noise piles most of its total energy into the top octaves and sounds bright. Brown noise reverses the weighting.

Measured on the files this site plays

Taken on 4 September 2026 out of the generator the site ships. Each colour ran for 30 seconds at 48 kHz mono through slo-noise-worklet.js, and an averaged 8192-point Hann FFT gave the spectrum. Loudness gives the integrated EBU R128 value at the generator’s default level, so the two columns compare directly.

Measurement	Brown noise	White noise
Power below 250 Hz	65.9%	1.0%
Power 250 Hz to 4 kHz	31.7%	15.6%
Power above 4 kHz	2.3%	83.4%
Power-weighted mean frequency, Hz	537	12,026
Integrated loudness, LUFS	−15.5	−11.5
Full-file RMS, dBFS	−13.9	−14.7

The last two rows explain a common complaint. At the same slider position the two files land 0.8 dB apart in raw RMS, yet white noise measures 4.0 LU louder once the meter weights frequency the way hearing does. Switch to white without touching the volume and it will jump at you. Drop the level a few notches on the way across, or you will judge white noise on loudness rather than on character.

What each one actually covers

Masking works locally in frequency. Fletcher’s 1940 work on auditory patterns established that a sound loses out mainly to the energy sitting in a narrow band around its own frequency, not to loudness elsewhere. That single fact decides this comparison.

Brown noise covers low, slow sounds: traffic through a window, a heat pump, a lorry idling, footsteps on the floor above. Two thirds of its power lands under 250 Hz, where those sounds live. Against typing two desks away it does almost nothing, because only 2.3% of its power reaches that high.

White noise covers sharp, bright sounds: speech consonants, cutlery, a door latch. Its 83.4% high-band share puts energy exactly where those events have their edges, and edges make a sound identifiable. Against a bass rumble it plays the wrong role.

One consequence people miss: brown noise needs a speaker that can reproduce it. On a phone most of that 65.9% never arrives, so brown noise on a handset sounds thin and covers less than the numbers suggest. White noise survives small drivers intact.

Loudness, comfort and long sessions

Human hearing runs least sensitive at the bottom of the range. ISO 226:2023, the current edition of the normal equal-loudness-level contours, tabulates how much more sound pressure a low tone needs to match the loudness of a mid one. That explains why brown noise can run at a higher measured level while feeling gentler, and why it tends to win over an eight-hour night.

The reverse case holds too. Brown noise loud enough to cover something can carry more energy into the room than it seems to, which matters if you share a wall. And a small speaker asked to produce heavy low content often distorts before it reaches a useful level.

Which people tend to pick

Preference splits along the disturbance and the length of the session, as far as anyone can tell by how people describe it. Long overnight listeners lean brown, because high-frequency hiss turns tiring over hours. Open-plan office workers lean white, because speech bothers them. Neither pattern counts as a study result, and no evidence links a colour to a personality type.

Try each one for a full minute rather than a few seconds. First impressions reward whichever sounds louder, and after a minute you notice which one you have stopped hearing.

Decision table

Your situation	Start with	Switch if
Traffic, HVAC or a rumbling neighbour	Brown	Your speaker distorts or the bass feels uneven
Voices, typing, sharp room noise	White	The hiss turns tiring within the hour
All-night playback on a decent speaker	Brown	You wake to sounds it never covered
Phone speaker with no headphones	White	It sounds harsh at a usable level
Deep focus on difficult reading	Silence first	Silence leaves the room too exposed
Shared bedroom, two different tastes	Brown at a low level	Either of you starts listening to it

Pink noise sits between the two, falling 3 dB per octave, and our three-way comparison covers it alongside these two if you want the middle option.

What the sleep research supports

Less than the category implies. Riedy and colleagues reviewed the human studies of continuous noise for sleep in 2021 and rated the evidence very low quality with mixed results, so no page should promise faster sleep onset for either colour. The World Health Organization’s 2018 environmental noise guidelines run firmer in the other direction: environmental noise can disturb sleep, and reducing the source beats covering it wherever that works.

Laboratory work on brief pink-noise pulses timed to measured slow waves turns up in this argument sometimes. That protocol has nothing to do with a track playing all night, and it says nothing about brown or white noise.

Keep the level modest whichever you choose. The WHO advises adults to keep average personal-audio exposure below 80 dB and to reduce listening time as the level rises. Alarms and smoke detectors should stay audible over anything you play.

Questions people ask

Is white or brown noise better for you?

Neither wins in general. Brown noise carries 65.9% of its power below 250 Hz, so it suits low rumble. White noise puts 83.4% above 4 kHz, so it suits hiss, clicks and voices. Match the colour to the sound you want covered.

Is it OK to play brown noise all night?

People do, and the risk sits in the level rather than the colour. Keep it low enough to talk over, keep alarms audible, and set a timer if you only need it while you settle.

What is the best colour noise for sleep?

No colour wins on evidence. A 2021 systematic review of continuous noise for sleep rated the research very low quality with mixed results. Pick the colour that covers your actual disturbance at the lowest level that works.

Is brown or white noise better for studying?

Brown noise makes the usual first try, because it carries no words and less high-frequency edge over a long session. White noise covers typing and speech better. Test each one against silence on the same task before you decide.

Read next

• Green noise explained covers the mid-weighted colour sitting between these two.
• What is red noise explains why red and brown name the same 6 dB per octave slope.
• Brown vs white vs pink noise adds pink to the comparison.
• The science behind brown noise goes deeper on what the research does and does not support.
• How to use brown noise for sleep covers level, timers and placement.

Sources

• White noise, entry 2.37, ASA/ANSI Standard Acoustical and Bioacoustical Terminology database. Brown noise and red noise have no entry.
• Voss and Clarke (1978), 1/f noise in music: Music from 1/f noise, Journal of the Acoustical Society of America 63(1), page 258.
• Fletcher (1940), Auditory Patterns, Reviews of Modern Physics volume 12, page 47.
• ISO 226:2023, Acoustics, normal equal-loudness-level contours.
• Riedy and colleagues (2021), Noise as a sleep aid: A systematic review.
• World Health Organization (2018), Environmental Noise Guidelines for the European Region.
• First-party measurements in src/data/sound-measurements.ts, taken 4 September 2026 against the shipped generator.