Pink, brown, white noise: which actually helps sleep

Noise 'color' is defined by how power is distributed across frequency. White noise has flat power spectral density: equal energy per Hz, so it sounds bright and hissy because each octave up doubles the bandwidth and thus the energy. Pink noise follows a 1/f slope (power falls ~3 dB per octave), giving equal energy per octave, which the auditory system perceives as balanced; many natural sounds (rainfall, wind) approximate 1/f. Brown (Brownian/red) noise follows 1/f^2 (power falls ~6 dB per octave), concentrating energy in low frequencies, hence the deep rumble. The slope, not loudness, is the variable that separates the three.

Masking is the dominant real-world mechanism. A steady broadband floor raises the threshold at which a transient sound (a door, a snore, traffic) becomes audible enough to trigger a cortical arousal. Stanchina et al. 2005 (Sleep Med) showed that adding white noise in a simulated ICU raised the baseline sound level and reduced the number of arousals caused by peak-noise events, because what wakes a sleeper is the change in sound, not the absolute level. Masking works at any noise color; pink and brown are simply more comfortable for many listeners because less energy sits in the harsh high frequencies. This is a passive, all-night effect and does not require any EEG triggering.

A separate and far more specific mechanism: closed-loop auditory stimulation. An EEG system detects the slow oscillation of NREM sleep (~0.5-1 Hz) and fires short pink-noise bursts (~50 ms) timed to the up-state of each wave. This phase-locking amplifies the slow oscillation rather than merely covering external sound. It is an active intervention requiring real-time EEG and is the only configuration with direct evidence of enhancing slow-wave activity. No consumer all-night sound machine reproduces this, because none of them read your EEG or phase-lock to it. [VERIFY: consumer wearables claiming closed-loop stimulation have limited independent validation against lab EEG triggering.]

Three primary results anchor the field. Ngo et al. 2013 (Neuron) delivered phase-locked pink-noise clicks to healthy young adults and found increased slow-oscillation amplitude, more sleep spindles, and better overnight retention of word-pairs versus sham stimulation; bursts delivered out of phase did not help. Papalambros et al. 2017 (Front Hum Neurosci) replicated the slow-wave enhancement and concomitant memory gain in older adults, the group whose slow-wave activity declines with age. Zhou et al. 2012 (J Theor Biol) reported that steady pink noise reduced sleep-onset latency and increased stable sleep in a small sample, framed around 1/f matching the brain's own activity. Riedy et al. 2021 (Sleep Med Rev), a systematic review, concluded the passive-noise evidence base is mixed and methodologically weak, with effects on onset and continuity that are small and inconsistent across studies.

For passive masking, color is mostly a comfort choice, not an efficacy one. White (flat) covers the broadest frequency range, including the high frequencies that make speech intelligible, so it masks conversation well but can sound harsh. Pink (1/f) keeps masking power across the spectrum while sounding more balanced and is the spectrum used in the slow-wave trials. Brown (1/f^2) emphasizes low-frequency rumble and is gentlest on the ears, but its thin high-frequency content masks high-pitched disturbances (alarms, beeps) less effectively. The viral 'brown noise' popularity is about subjective calm, not a documented sleep-architecture advantage over pink or white. [VERIFY: the specific brown-noise focus and calming claims circulating on social platforms are largely anecdotal and lack controlled-trial support.]

Target ~45-60 dB(A) at the pillow, roughly the level of light rain or a quiet refrigerator, loud enough to raise the masking floor but well under the 70 dB(A) range where sustained exposure begins to carry cumulative hearing risk. Place the machine across the room rather than beside the head so the sound field is diffuse and the source is not close to the ear. Use a continuous track, not a looping clip with an audible seam that can itself trigger arousal. Expect some sleepers to habituate or find broadband noise activating; if onset or continuity does not improve within 1-2 weeks, the mechanism is masking and the effect may simply not apply to your sleep environment.