White Noise Generator

How the White, Pink and Brown Noise works

Noise is a random signal, and its "colour" describes how its energy is spread across the frequencies. The three here are the ones people actually reach for.

How the white noise generator makes each colour

White noise is a fresh random value for every sample, so every frequency gets the same energy per hertz. Because there are far more hertz in the treble than the bass — the octave from 10 to 20 kHz is ten thousand hertz wide, the octave from 50 to 100 Hz only fifty — white noise sounds bright and hissy, like an untuned radio.

Pink noise falls by 3 dB per octave, which gives every octave the same energy. That is how the ear groups pitch, so pink noise sounds even and balanced, like steady rain. It is the standard signal for setting speaker levels and the one most people find easiest to sleep to. It is made here by passing white noise through a bank of simple filters (Paul Kellet's method), accurate to within a fraction of a decibel across the audible range.

Brown noise (named after Robert Brown and Brownian motion, not the colour) falls by 6 dB per octave. Each sample is the previous one plus a small random step, so the signal wanders rather than jumps. The result is a deep rumble like surf or a distant waterfall, with very little hiss.

Stereo and looping. Playback uses eight seconds of independent noise on each channel, so the sound is wide rather than centred, and eight seconds of random data is far too long for the ear to notice the repeat. The export generates the full length you ask for, with a one-second fade at each end so it does not click when it starts or stops.

Volume. The slider follows a squared curve, which tracks loudness more naturally than a straight line. For sleep, keep it quiet — noise masks other sounds well below the level at which it becomes tiring.

Related tools