Is white noise safe for long-session hearing?

At moderate levels (typically <50 dB SPL at the pillow) the risk is low. The danger is sustained high-volume exposure — start at low volume and increase only to mask intrusions.

What does the sleep timer do?

Counts down 15/30/60 min from the moment audio actually starts playing — arming while idle then pressing Play later resets the countdown. Audio fades over the last 4 seconds. Closing the tab cancels it.

Difference between white, pink, brown noise?

White = flat power per Hz; pink = −3 dB/octave (1/f); brown = −6 dB/octave (1/f²). White is brightest, pink balanced, brown rumbliest. Each colour has its own generator.

Does it work offline?

Yes — once the page is loaded, all processing runs on Web Audio in your browser. Noise, sleep timer, and WAV download all work fully offline.

☁️

White Noise Generator

Free, browser-only white noise generator with a sleep timer (15 / 30 / 60 min), live FFT spectrum showing the characteristic flat power profile, and WAV download at the browser's native sample rate. Useful for sleep, focus, studying, infant soothing, and as a test signal for audio equipment.

Playback

Volume 30 %

Start moderate — white noise is much fuller than a single tone at the same setting. Slider takes effect immediately.

Idle — press Play.

Stream info

Sample rate

—

browser-chosen, typically 48 kHz

Average spectrum

—

mean dB across all FFT bars

Analyser noise floor

—

Web Audio AnalyserNode.minDecibels — bars below this clamp to the floor

Sleep timer

Time remaining

—:——

Audio will fade out over the last 4 seconds so you aren't jolted by an abrupt cutoff.

Download WAV

Length

10 s 30 s 60 s

Saved as 16-bit PCM mono at the audio context's sample rate. 30 s ≈ 3 MB. The 60 s option briefly freezes the UI (~250 ms) while filling the buffer.

Generated locally — never uploaded.

Live FFT spectrum — 48 bars, linear frequency (0 → Nyquist)

White Noise — What It Is and How to Use It

“White” noise is the audio analogue of white light: a signal whose power is uniformly distributed across frequency. In practice, that means each sample is an independent random number with no correlation to the next, drawn from a uniform distribution between −1 and +1. The spectrum is statistically flat at every frequency from DC up to the Nyquist limit set by your browser's sample rate (usually 24 kHz at 48 kHz audio).

Why it's used

Sleep, focus, and concentration — the broadband hiss masks intermittent intrusions (traffic, talking, snoring, distant TV) so the brain's pattern-detection circuitry stops snapping awake at every sudden onset.
Infant soothing — mimics the diffuse low-frequency rush babies hear in utero. Use at modest volume and from a safe distance.
Speaker / room / measurement testing — a flat broadband signal is the cleanest input for impulse-response, RT60, and frequency-response measurements; it excites every band evenly so the room's behaviour is what you're measuring, not the test signal's.
Audio-tool calibration — feed this generator into a real-time analyser elsewhere on this site to verify the analyser's calibration: a true white-noise input should produce a flat (or +3 dB/octave on log frequency) display.

Why the bars look noisy even though it's “flat”

The spectrum is flat in expectation only. Any finite window of a random signal shows random bar-to-bar variation — the average over time is flat, but each individual frame is noisy. The FFT analyser smooths slightly between frames (smoothingTimeConstant = 0.6), so what you see is partially time-averaged; the gold dashed reference line shows the per-frame mean across all bars, which lands close to the same value frame after frame and demonstrates the underlying flatness more clearly.

White vs pink vs brown noise

This page generates white noise. Two close relatives are pink noise (1/f power, −3 dB/octave) and brown / red noise (1/f² power, −6 dB/octave). Pink noise sounds more “balanced” to the ear because human hearing is roughly logarithmic in frequency; brown noise is rumblier and is often preferred for ADHD focus and deep sleep. Each colour has its own generator in this category.

About the WAV download

The download button generates the requested duration on demand, encodes it as 16-bit PCM mono at the audio context's sample rate (usually 48000 Hz on desktop), and triggers a save through a Blob URL. Nothing is uploaded — the file is built in your browser and handed straight to your OS download dialog. Loop it in your media player of choice for indefinite playback.

Frequently Asked Questions

Is the audio loop noticeable?

Internally the tool plays a 2-second buffer of independent random samples on loop. Statistically the loop is detectable; perceptually it isn't — white noise has no salient features for the brain to lock onto, so the cycle isn't audible. For maximum freshness you can download a 60-second WAV and loop it in a player.

Is white noise safe for hearing during long sessions?

At moderate levels, yes — many sleep clinicians recommend < 50 dB SPL at the pillow. The risk with long sessions is chronic exposure at high volume, which can accelerate hearing fatigue. Start the slider low, increase only as needed to mask intrusions, and place the source a metre or so away from your ear rather than next to it.

Should I use white noise for an infant?

It can help — pediatricians often suggest white or pink noise to mimic the in-utero environment — but keep the source at least a metre from the crib, keep the level at or below ~50–55 dB SPL (no louder than the noise it's masking), and don't run it continuously through the night. Consult your pediatrician for specifics, especially for newborns.

What does the “sleep timer” actually do?

It counts down from 15, 30, or 60 minutes from the moment audio actually starts playing. Arming the timer while idle, then pressing Play later, resets the countdown to the full duration — the timer measures listening time, not wall-clock time since arming. When it reaches zero the audio fades out smoothly over 4 seconds rather than cutting off abruptly. The countdown runs on the browser's wall clock so a brief tab-suspend doesn't drift it; closing the tab cancels it entirely.

Why does the spectrum slope up at high frequencies on some analysers?

A spectrum analyser with a log-frequency X axis bins more linear-Hz into each higher band, so flat power-per-Hz looks like a +3 dB/octave rise. This tool uses a linear-frequency axis so genuine white noise looks flat. Both are correct; they're showing different things.

What's the difference between this and the “Pink” or “Brown” noise generators?

White is flat power per Hz; pink falls off at −3 dB/octave; brown falls off at −6 dB/octave. To the ear, white is the brightest (hissier), pink is balanced (rain-like), and brown is the deepest (rumble). Try all three to see which masks your room's intrusions best.

Does this work offline?

Once the page is loaded, yes — everything runs on Web Audio in your browser. You can put the tab on aeroplane mode and the noise + sleep timer + WAV download all continue to work. Nothing is uploaded or fetched after page load.

What sample rate is the WAV?

Whatever the browser's AudioContext.sampleRate reports — usually 48000 Hz on desktop and many mobile devices, occasionally 44100 Hz. The exact rate is embedded in the “Sample rate” readout once you press Play.

Related Tools

← All Wave Generators