Measured
Brown Noise, Pink Noise and White Noise, Measured
What each noise color actually is, shown as spectra measured from real audio, with a sample of each and what the research says about focus, ADHD and sleep. Every research figure was checked against its primary source.
Last verified · 7 sources · 7 audio measurements
Key facts
- Brown noise loses about 6 dB per octave (measured −6.10), twice as fast as pink noise.
- Pink noise loses about 3 dB per octave (measured −3.05).
- White noise is flat: equal power at every frequency.
- Green noise has no agreed definition.
- Underlay.fm slopes like brown noise up to 1 kHz, then drops far faster (42 to 83 dB down at 4 kHz vs 32 for brown).
- No controlled studies of brown noise for ADHD were found in the 2024 meta-analysis.
- Evidence that continuous noise improves sleep is rated "very low" quality.
- WHO-ITU: 80 dB for 40 hours a week is the safe listening limit for adults.
Brown noise
What brown noise is
Brown noise loses about 6 dB of power per octave, twice as fast as pink noise. Measured: −6.10 dB per octave, against a theoretical −6.02. At 4 kHz it is 32 dB below its level at 100 Hz (Underlay.fm measurement, 2026).
Brown noise is random noise weighted heavily toward low frequencies. Its power falls by a quarter every time the frequency doubles, which works out to about 6 dB per octave. Most of what you hear is rumble: a fast river, wind against a window, distant thunder.
The name has nothing to do with the color. It comes from Brownian motion, the random drifting of particles first described by the botanist Robert Brown, because brown noise is what you get when you add up a long run of random steps. That is exactly how our generator makes it: it sums white noise sample by sample. Some sources call it red noise.
Measured spectrum, relative to 100 Hz
- White noise
- Pink noise
- Brown noise
- Underlay.fm (range across sessions)
Pink noise
What pink noise is
Pink noise loses about 3 dB of power per octave, so each doubling in pitch carries half the power of the octave below. Measured: −3.05 dB per octave, against a theoretical −3.01 (Underlay.fm measurement, 2026).
Pink noise sits between white and brown. Its power halves every time the frequency doubles, about 3 dB per octave, which means every octave carries the same total energy. Because human hearing is also organized roughly by octaves, pink noise tends to sound balanced, like steady rain or wind in leaves.
Our generator makes pink noise with the Voss-McCartney algorithm, which adds together random sources that update at halving rates. It lands within 0.04 dB per octave of the theoretical slope between 100 Hz and 1 kHz.
Measured spectrum, relative to 100 Hz
- White noise
- Pink noise
- Brown noise
- Underlay.fm (range across sessions)
White noise
What white noise is
White noise has equal power at every frequency. Measured from the generator Underlay.fm's sounds are built from, its spectrum slopes −0.05 dB per octave between 100 Hz and 1 kHz, flat within measurement error (Underlay.fm measurement, 2026).
White noise has the same power at every frequency, the way white light mixes every color. Each octave spans twice as many frequencies as the one below it, so white noise piles most of its energy into the upper range and sounds bright and hissy: TV static, a fan on high, air escaping a tire.
That brightness is also what makes it good at masking sudden sounds like conversation or a door, which carry much of their detail in the upper range. It is the noise most studies of noise and attention have used.
Measured spectrum, relative to 100 Hz
- White noise
- Pink noise
- Brown noise
- Underlay.fm (range across sessions)
Green noise
What green noise is (and isn't)
Green noise has no agreed definition. White, pink and brown noise are each defined by a spectral slope, but we found no acoustics standard or paper that defines green noise that way; the name is used loosely for mid-frequency or nature-like noise (Underlay.fm review, 2026).
Green noise is marketed alongside the others, but it isn't the same kind of thing. White, pink and brown noise each have a precise definition, a slope you can measure. We found no acoustics standard or research paper that defines green noise the same way.
In practice the name is used for at least two different sounds: noise concentrated in the middle of the hearing range, often said to be centered around 500 Hz, and recordings of natural environments. We could not trace the 500 Hz figure to a primary source, so we have not measured or plotted a green noise curve.
Our own sound
Where Underlay.fm sits
Across 4 recorded Underlay.fm sessions, the output falls −4.4 to −6.2 dB per octave between 100 Hz and 1 kHz, close to brown noise (−6.1). Above 1 kHz it drops much faster: at 4 kHz it sits 42 to 83 dB below its 100 Hz level, against 32 dB for brown noise (Underlay.fm measurement, 2026).
Underlay.fm's sounds are synthesized from white, pink and brown noise layered with tuned tones, then filtered. So we measured the finished output too: the sample on our homepage and 3 full sessions recorded straight from the audio engine, across different moods and tasks.
Up to 1 kHz the output slopes like brown noise, −4.4 to −6.2 dB per octave. At 1 kHz it sits 19 to 28 dB below its 100 Hz level, about where brown noise sits (20 dB). Above that it falls much faster, because the final mix passes through a low-pass filter set at 3 or 5 kHz depending on the session, and most of the layers are low-passed too. At 4 kHz it is 42 to 83 dB down, where brown noise is 32 dB down.
In plain terms: Underlay.fm sounds about as deep as brown noise, with far less hiss. That is a design choice, not a finding about focus. The research below doesn't show that any noise color is better for concentration than another.
Measured spectrum, relative to 100 Hz
- White noise
- Pink noise
- Brown noise
- Underlay.fm (range across sessions)
Noise and ADHD
Brown noise and ADHD: what the research shows
A 2024 meta-analysis of noise and ADHD searched for brown noise studies and found none. White and pink noise gave a small benefit to young people with ADHD (g = 0.249, 13 studies) and a small harm to people without ADHD (g = −0.212) (Nigg et al., 2024).
Brown noise for ADHD is one of the most shared focus tips online, but the controlled research is on white and pink noise. A 2024 meta-analysis pooled randomized studies of children and young adults with ADHD or high ADHD symptoms, and searched for brown noise studies along the way. It found none that met its criteria.
For white and pink noise, it found a small benefit on attention tasks for people with ADHD and a small cost for people without it. Averages hide a split, too: in a 2024 study of 43 children with clinical ADHD, about one third performed worse under noise, mostly those with more hyperactive and impulsive traits.
None of this shows brown noise doesn't help. It shows nobody has tested it properly yet.
Effect of white or pink noise on attention tasks
Sources: Nigg et al. (2024), Journal of the American Academy of Child and Adolescent Psychiatry; Söderlund et al. (2024), Scandinavian Journal of Child and Adolescent Psychiatry and Psychology
Sleep
Pink and white noise for sleep
A systematic review of 38 studies rated the evidence that continuous noise improves sleep as "very low" quality (Riedy et al., 2021). The pink noise study most often cited for deep sleep used short pulses timed to brain waves in 13 older adults, not a continuous track (Papalambros et al., 2017).
A 2021 systematic review gathered 38 studies of continuous white or broadband noise and sleep. Results ran from noise improving sleep to noise disrupting it, and the reviewers rated the overall evidence "very low" using the standard GRADE scale. They warn against promoting noise as a sleep aid until better studies exist.
The pink noise result people usually cite is a different kind of experiment. In a 2017 study, 13 adults aged 60 to 84 slept wearing EEG sensors, and a computer played short pulses of pink noise timed to the peak of each slow brain wave. After that night, they recalled 9.2 more word pairs than the evening before, against 3.1 after a sham night.
Two details rarely make it into the headlines: slow-wave activity across the whole night was about the same with and without stimulation, and participants rated their sleep no better. A pink noise track playing all night is not what was tested.
Overnight gain in word pairs recalled
Sources: Riedy et al. (2021), Sleep Medicine Reviews; Papalambros et al. (2017), Frontiers in Human Neuroscience
Hearing
How loud is too loud
All 14 infant sleep machines tested exceeded the 50 dBA hospital nursery limit at maximum volume, measured 30 cm away, and 3 exceeded 85 dBA (Hugh et al., 2014). The WHO-ITU safe listening standard allows 80 dB for 40 hours a week for adults and 75 dB for children (WHO & ITU, 2019).
No noise color is harmful in itself; volume and hours are. The WHO-ITU safe listening standard budgets sound by the week: 80 dB for 40 hours for adults, 75 dB for children. Every 5 dB louder cuts the safe time sharply.
Devices sold for sleep can exceed these limits. When researchers measured 14 infant sleep machines at full volume, every one was louder than the 50 dBA limit used in hospital nurseries at 30 cm, and 3 passed 85 dBA, a level that breaks adult workplace limits if played for more than 8 hours.
Safe weekly listening time by volume
Sources: Hugh et al. (2014), Pediatrics; WHO & ITU (2019), World Health Organization; WHO (2026), World Health Organization
Method
How these were measured and checked
The noise curves come from the same generator functions Underlay.fm's stems are synthesized with: 20 seconds of each color at 44.1 kHz from a fixed random seed. Each signal's spectrum was estimated with Welch's method (16,384-point segments). Slopes are least-squares fits of power in dB against the logarithm of frequency between 100 Hz and 1 kHz. Levels are the power in a one-third-octave band, relative to the band around 100 Hz.
The Underlay.fm measurements are recordings of the audio engine's final output, taken from the browser before it reaches the speakers: the homepage sample, plus two-minute sessions with the first 20 seconds dropped so the fade-in doesn't skew them. We report the range across all of them rather than an average, since each session draws different layers.
The listening samples are 10 seconds of the same noise, matched to the same loudness (−24 LUFS), with inaudible content below 24 Hz removed so the brown noise sample can't overdrive small earbuds. Research figures were each checked against the paper's full text or publisher abstract. Claims we left out on purpose:
- "Brown noise is proven to help ADHD." No controlled studies were found. Social media view counts are not evidence.
- "Pink noise gives you deeper sleep." The study behind it used EEG-timed pulses, not a continuous track, and whole-night slow-wave activity did not change.
- "Green noise is centered at 500 Hz." We could not trace this to any primary source.
- "Brown noise masks low-frequency sounds best." Plausible, but we found no study that tested it.
Show the measured data
| Signal | dB/octave, 100 Hz to 1 kHz | 1 kHz | 2 kHz | 4 kHz |
|---|---|---|---|---|
| White noise | −0.05 | −0.2 | −0.2 | −0.2 |
| Pink noise | −3.05 | −9.8 | −12.9 | −16.1 |
| Brown noise | −6.10 | −19.6 | −25.6 | −31.6 |
| Underlay.fm: homepage sample | −6.16 | −22.1 | −51.9 | −83.1 |
| Underlay.fm: Deep Focus, Scattered session | −4.94 | −27.7 | −52.1 | −83.2 |
| Underlay.fm: Creative, Tired session | −4.41 | −24.4 | −21.1 | −42.3 |
| Underlay.fm: Light Work, Wired session | −5.57 | −18.7 | −35.0 | −48.7 |
Levels in dB relative to 100 Hz.
Takeaways
Choosing a noise color
The colors differ in one measurable way, how fast they lose power as pitch rises, and that is what you hear as bright or deep. The research doesn't crown a winner for focus: the evidence is for white and pink noise, it is small, and it helps some people while slightly hurting others. So pick by ear, keep the volume low, and give each a full session rather than a few seconds.
If you like the depth of brown noise but find flat noise tiring over hours, that is the sound Underlay.fm was built for: brown-noise weight in the low end, with the hiss filtered out and no melody or beat to follow.
Common questions
What is brown noise?+
Brown noise is random noise whose power falls by about 6 dB for every doubling in frequency, so most of its energy sits in the low end and it sounds like a deep rumble, closer to a river or distant thunder than to static. We measured −6.10 dB per octave from our generator. The name comes from Robert Brown and Brownian motion, the random movement it mathematically resembles, not from the color.
What's the difference between white, pink and brown noise?+
How fast power falls as pitch rises. White noise is flat (0 dB per octave), pink noise falls about 3 dB per octave, and brown noise about 6. By 4 kHz, pink noise is 16 dB below its level at 100 Hz and brown noise is 32 dB below, which is why brown sounds so much deeper.
Is brown noise good for ADHD?+
Nobody has tested it properly yet. A 2024 meta-analysis searched for controlled studies of brown noise and ADHD and found none. It did find a small benefit from white and pink noise for young people with ADHD (g = 0.249) and a small harm for people without ADHD. In a separate 2024 study of 43 children with ADHD, about one third performed worse with noise.
Is pink noise good for sleep?+
The evidence for playing continuous noise at night is weak: a review of 38 studies rated it "very low" quality. The well-known pink noise sleep study used brief pulses timed by an EEG to each slow wave, in 13 adults aged 60 to 84. A pink noise track on a speaker does not do that.
Is it safe to listen to brown noise all day?+
The noise color matters less than the volume. The WHO-ITU standard sets 80 dB for 40 hours a week as the safe limit for adults, and WHO guidance cuts that to 4 hours a week at 90 dB. Background noise for focus should sit well below that; if you can't hear someone speaking at a normal volume, it is too loud.
What is green noise?+
A loose label, not a defined signal. It is usually described as noise concentrated in the middle of the hearing range, or as the sound of natural environments. Unlike white, pink and brown noise, it has no spectral slope that defines it, so two "green noise" tracks can sound quite different.
Does Underlay.fm play brown noise?+
Not as a single noise color. Its sounds are synthesized from white, pink and brown noise layered with tuned tones, then filtered. Measured across 4 sessions, the result slopes like brown noise up to 1 kHz and then falls away much faster, so it is darker than brown noise in the upper range.
Sources
- Nigg, J. T., Bruton, A., Kozlowski, M. B., Johnstone, J. M., & Karalunas, S. L. (2024). Systematic Review and Meta-Analysis: Do White Noise or Pink Noise Help With Task Performance in Youth With Attention-Deficit/Hyperactivity Disorder or With Elevated Attention Problems?. Journal of the American Academy of Child and Adolescent Psychiatry, 63(8), 778 to 788. doi:10.1016/j.jaac.2023.12.014
- Söderlund, G. B. W., Hadjikhani, N., Thorsson, M., et al. (2024). Sensory white noise in clinical ADHD: Who benefits from noise, and who performs worse?. Scandinavian Journal of Child and Adolescent Psychiatry and Psychology, 12(1), 92 to 99. doi:10.2478/sjcapp-2024-0010
- Riedy, S. M., Smith, M. G., Rocha, S., & Basner, M. (2021). Noise as a sleep aid: A systematic review. Sleep Medicine Reviews, 55, 101385. doi:10.1016/j.smrv.2020.101385
- Papalambros, N. A., Santostasi, G., Malkani, R. G., Braun, R., Weintraub, S., Paller, K. A., & Zee, P. C. (2017). Acoustic Enhancement of Sleep Slow Oscillations and Concomitant Memory Improvement in Older Adults. Frontiers in Human Neuroscience, 11, 109. doi:10.3389/fnhum.2017.00109
- Hugh, S. C., Wolter, N. E., Propst, E. J., Gordon, K. A., Cushing, S. L., & Papsin, B. C. (2014). Infant sleep machines and hazardous sound pressure levels. Pediatrics, 133(4), 677 to 681. doi:10.1542/peds.2013-3617
- World Health Organization & International Telecommunication Union (2019). Safe listening devices and systems: a WHO-ITU standard. World Health Organization, Geneva. https://www.who.int/publications/i/item/9789241515276
- World Health Organization (2026). Deafness and hearing loss: Safe listening. World Health Organization, questions and answers, March 6, 2026. https://www.who.int/news-room/questions-and-answers/item/deafness-and-hearing-loss-safe-listening
Want the numbers on music, lyrics and attention? See the statistics →
Looking for the ADHD-specific case for non-music audio? Read that guide →
Try it free. One session, no account required.
Start a free sessionThen a free session every day · Pro from $4/mo