The 3AM Overthinker

Brown noise vs white noise vs pink noise for sleep

Published November 10, 2026 · By The 3AM Overthinker, from the research behind the 3AM Coach

On this page
  1. Quick comparison
  2. Music: the one with real evidence
  3. White noise: very low-quality evidence
  4. Pink noise: where the deep-sleep reputation comes from
  5. Brown noise: popular, untested
  6. Nature sounds: good for winding down, not tested for sleep
  7. Binaural beats: some relaxation, no good support for brainwave claims
  8. A note on volume and babies
  9. How the 3AM Coach uses this

None of the three noise colours has good evidence behind it as a sleep aid. A review of 38 studies rated the evidence for continuous noise as very low quality, brown noise hasn't been specifically tested in sleep studies, and a small 2026 lab trial found pink noise cut REM sleep. The sound with the best support is slow, soft instrumental music, played quietly for 30 to 45 minutes at bedtime.

Quick comparison

SoundBest claim the research supportsEvidence strength
Slow instrumental musicBetter self-rated sleep quality in adults with insomniaModerate
White noiseMaybe faster sleep onset in noisy roomsVery low
Pink noise (continuous)Masks other noise; one small trial found less REM sleepLow, with a signal of harm
Brown noiseNothing yetNo sleep studies
Nature soundsLower stress while awakeLow to moderate, not sleep research
Binaural beatsMedium pooled effect on memory, attention, anxiety and pain in mixed-quality studies; brainwave claims unsupportedLow

Music: the one with real evidence

Of all the sounds people play at night, slow instrumental music has the most behind it. A Cochrane review of 13 randomised trials with 1,007 adults with insomnia found that listening to prerecorded music for 25 to 60 minutes a day improved self-rated sleep quality. Across the 10 trials (708 people) that used the Pittsburgh Sleep Quality Index, scores improved by close to 3 points on its 21-point scale compared with no treatment or usual care, a meaningful difference on self-rated sleep (Jespersen et al., 2022). Cochrane rated the certainty of that result as moderate. People in these trials knew they were getting music, so expectation may account for some of it. A 2018 network meta-analysis of 20 trials, by Feng and colleagues, ranked music listening above exercise, acupuncture and placebo music for the same outcome.

The music in these trials had a lot in common: 60 to 80 beats per minute, soft and simple, no lyrics, no accented drums or syncopation, played for 30 to 45 minutes at bedtime at a comfortable, low volume (one trial set it at 50 to 60 dB). A 2025 review that pulled these features together only included trials with positive results, so it describes what the successful trials used and can't show those features matter.

The benefit shows up mostly in how people rate their sleep, and effects on measured sleep time are smaller and less certain. Most trials also compared music with doing nothing, so some of the gain may come from having a calm bedtime routine at all.

If you use it: pick something slow and dull enough that you won't follow the tune. Keep the volume below normal conversation, and set a 30 to 45 minute timer rather than letting it run all night.

White noise: very low-quality evidence

White noise is widely sold as a sleep aid, but the evidence is very weak. Riedy and colleagues at the University of Pennsylvania reviewed 38 studies of continuous broadband noise and concluded that the evidence it improves sleep is "of very low quality" (Riedy et al., 2021). There were hints of faster sleep onset, and at least one study suggested the noise disrupted sleep instead.

If you use it: the plausible use is masking a noisy street or a snoring partner. Keep it quiet (well under 50 dBA at the ear), put the speaker away from your head, and use a timer. Foam earplugs are worth trying first; see the pink noise section for why.

Pink noise: where the deep-sleep reputation comes from

Continuous pink noise has low-quality evidence, and the most carefully controlled recent trial found a downside. In a 2026 randomised crossover study, 25 healthy adults slept seven nights in a sleep lab. Pink noise at 50 dBA cut REM sleep by almost 19 minutes. Combined with intermittent aircraft-type noise, it added about 15 minutes of wake. Foam earplugs prevented most of the deep sleep lost to that noise, though they started to fail at the loudest level tested (Basner et al., 2026). It was one small trial in young, healthy adults, so the result needs repeating before anyone treats it as settled.

A smaller 2026 pilot of 12 adults found quieter pink noise (45 dB) reduced the brief arousals caused by traffic noise, so masking may help a little with fragmentation, but no study has shown it improves sleep overall.

The "deep sleep" reputation comes from closed-loop lab studies, where short bursts of sound are timed to the brain's slow waves using live EEG. An app playing a steady hiss all night is doing something quite different. Even in the lab setting, the memory benefit has faded as more groups have tried to repeat it: a 2023 meta-analysis found the pooled effect was small and not statistically reliable, and later studies found less than earlier ones, with the newest finding no benefit by 2021.

If you use it: the same rules as white noise. Low, away from the head, on a timer.

Brown noise passed 95 million TikTok views in 2022, according to news coverage at the time, and has no sleep studies of its own. Two sleep researchers writing that year noted it hasn't been specifically tested (Paech & Micic, 2022). It's deeper and less hissy than white noise, which many people find more pleasant, though finding a sound pleasant doesn't tell you much about whether it helps you sleep. Acoustically it is still continuous broadband noise, so the concerns above likely apply.

If you use it: treat it like white noise. Quiet, timed, and not a substitute for earplugs in a loud room.

Nature sounds: good for winding down, not tested for sleep

Natural sounds reduced stress and annoyance in a meta-analysis of 18 studies published in PNAS. Water sounds had the biggest effect on health outcomes and birdsong the biggest effect on stress (Buxton et al., 2021). These were mostly short listening sessions with people awake. The widely quoted "184% improvement" is a relative figure pooled from varied small studies, so it says less than it sounds. No trials were found testing nature sounds for falling asleep.

If you use it: a few minutes of birdsong or water while you wind down is reasonable. An all-night rain track is, acoustically, broadband noise again.

Binaural beats: some relaxation, no good support for brainwave claims

Binaural beats play slightly different tones in each ear. A 2019 meta-analysis of 22 studies found a medium overall effect across memory, attention, anxiety and pain, with mixed study quality. The idea behind "delta waves for deep sleep" is that the beat drags brainwaves to match it. A 2023 systematic review of 14 EEG studies in awake people, only 3 of them at delta frequencies, found that 8 contradicted this, 5 supported it and 1 was mixed (Ingendoh et al., 2023). A 2025 review said they "show promise but require further validation" for sleep.

If you use it: it may relax you about as much as any calm audio. It needs headphones, which are uncomfortable in bed and make it easy for the volume to creep up, so keep it quieter than normal conversation.

A note on volume and babies

Keep it quiet, below normal conversation. That goes for anything you fall asleep to. You may see the WHO and ITU safe listening figure of 80 dB for no more than 40 hours a week quoted online. That is a hearing-safety ceiling across all your listening in a week, and it is not a bedtime level.

Researchers testing 14 infant sleep machines at maximum volume found all of them exceeded 50 dBA at 30 cm, the limit recommended for hospital nurseries, and three exceeded 85 dBA (Hugh et al., 2014). The 2026 pink noise team also raised concern for infants and children, who spend more of the night in REM sleep. If you use a machine in a nursery, keep it low, far from the cot, and off once the baby is asleep.

How the 3AM Coach uses this

The coach's sounds follow the music research above: five slow instrumental presets between 60 and 72 bpm, played low on a 30 to 45 minute timer, with no white, pink or brown noise option. If your room is noisy, it suggests foam earplugs instead.

Try the coach

The numbers, for the curious

Sources

More guides