White Noise Generator

Generate white, pink or brown noise in the browser and play it for as long as you need. The audio is synthesised on your own device in real time, so there is no download, no streaming and no file sitting on a server.

Three Colours of Noise, Three Different Spectra

The colours are named for the way energy is distributed across frequencies, and the difference is audible once you know what to listen for. White noise sounds hissy and bright. Pink noise sounds more like steady rainfall or a fan. Brown noise is a low rumble, closer to distant traffic or an aircraft cabin. Which one helps depends on what you are trying to mask.

• White noise has equal energy at every frequency, so high frequencies are as loud as low ones. It is the most effective masker of speech, which is why it is the default for an office or a shared room.
• Pink noise has equal energy per octave, which means energy falls as frequency rises. That matches the spectrum of many natural sounds and most people find it easier to listen to for hours.
• Brown noise falls faster still, roughly 6 dB per octave, producing a deep, even rumble. It is the common choice for masking low-frequency sound such as footsteps from the flat above.
• An optional pure tone can be layered on top, adjustable between 20 Hz and 2000 Hz. This exists for tinnitus masking, where a narrow tone near the pitch you hear is more effective than broadband noise alone.

Setting It Up

1. Press play to start. Nothing is downloaded and playback is generated live, so it begins immediately.
2. Choose a colour using the three buttons. You can switch between them while playing without an interruption.
3. Set the volume with the slider, which runs from 0 to 100 and starts at 50. Begin lower than you expect to need — noise is easy to creep upward over an hour.
4. Set the timer if you want playback to stop on its own. It accepts 0 to 480 minutes, and 0 means no timer at all, so the noise continues until you stop it.
5. Optionally raise the tone level to mix a sine wave into the noise, and set its frequency. A level of 0 removes it entirely, which is the default.
6. Turn on mix mode if you want more than one noise colour playing at once. The layers are summed rather than switched between.

The Four Built-In Starting Points

Alternative	Notes
Sleep	Pink noise at 40 percent volume with a 60-minute timer. The timer matters here: the point is to fall asleep to it, not to keep the room noisy all night.
Focus	White noise at 35 percent, no timer. White noise is the strongest masker of speech, so this is the preset for a noisy open-plan office where conversation is the distraction.
Study	Brown noise at 45 percent, no timer. Softer and lower than the focus preset, better suited to reading where a bright hiss becomes tiring.
Tinnitus	Pink noise at 30 percent with mix mode on, layering white and pink together, plus a 4000 Hz tone at a low level. The tone is meant to sit below the noise rather than be the main sound.

How the Noise Is Actually Made

This is the part that distinguishes a real generator from a looping recording. The page fills an audio buffer using the Web Audio API and loops it, which is why it can run for eight hours without a file large enough to matter.

• The buffer holds two seconds of audio at the device's own sample rate and repeats seamlessly. Two seconds is long enough that the repeat is not perceptible as a rhythm.
• White noise is generated sample by sample as a random value between minus one and one. There is no filtering and no correlation between one sample and the next, which is exactly what makes its spectrum flat.
• Pink noise is produced with a recursive filter of the kind published by Paul Kellet — seven interlinked one-pole stages whose outputs are summed and scaled. This is an approximation to true pink noise that tracks the ideal spectrum closely across the audible range and costs almost nothing to compute.
• Brown noise is a single one-pole low-pass with feedback, which is the digital equivalent of integrating white noise. Energy falls about 6 dB per octave, giving the characteristic rumble.
• The visualiser is a live oscilloscope drawn from the time-domain output of the analyser, redrawn on every animation frame. It shows the waveform, not a frequency spectrum.

Using It Without Doing Yourself Harm

Noise is a tool with a sensible dosage, and the failure mode is gradual rather than obvious.

• Keep the level below the point where you would have to raise your voice to be heard over it. Sustained exposure at conversational levels for eight hours is a genuine risk to hearing.
• Do not use it to mask tinnitus permanently at a higher volume than the tinnitus itself. The standard guidance is that masking sound should be at or slightly below the perceived loudness of the sound you are masking.
• Take breaks. Continuous noise for a whole working day is harder on attention than the intermittent noise it is replacing, because there is no quiet interval in which to recover.
• Stop if it worsens your sleep. Some people find broadband noise reduces sleep latency, others find it fragments the second half of the night. The timer exists partly for this reason.

White Noise Questions

Does the page record anything through my microphone?

No, and it never asks for microphone access. All three noise colours are generated mathematically from random numbers — the audio is created rather than captured. The only permission involved is the one your browser requires before any page may produce sound.

Why does the timer offer 480 minutes if I only want to sleep?

Because the same page is used for concentration during a long shift or a study session. The practical settings for sleep are the 30 to 90 minute range; the longer values are there for background masking over a working day.

What does mix mode do that switching colours does not?

Switching replaces one colour with another. Mix mode plays several at once, summing them into a single output. The tinnitus preset uses this to layer white and pink together, which produces a broader mask than either colour achieves alone.

Can I run it for eight hours?

Yes. The audio is a two-second buffer looping in memory, so duration costs nothing in storage or memory and there is no file to run out. Practically, your device will stay awake while it plays, so the battery cost is the real limit on a laptop or phone.

Last verified: October 3, 2026

Launch White Noise Generator

All Tools · All Tools · About · How We Test · Contact · Privacy

Maintained by Huang Bin · Last updated October 3, 2026