White noise is sound whose power spectral density is flat across every frequency the audio system can reproduce, with the same average energy per band whether that band sits at 200 Hz, 2,000 Hz, or 20,000 Hz. That flatness across the spectrum is the defining property, the same idea that makes white light appear as the combination of every visible color at once. A white noise generator therefore produces a continuous stream of uncorrelated samples, hands them to a speaker or headphone output, and lets the listener hear a steady, featureless hiss. Digital systems cannot deliver infinite bandwidth, so what you actually get is a practical band-limited approximation that stays flat only across the frequencies the device can resolve. The White Noise Generator from Lizely plays exactly that kind of bounded digital approximation directly inside your browser tab, with a slider that moves from 0% to 100% and a single click that ends playback without any fade or delay.
What "White" Actually Means in a Noise Signal
Engineers borrowed the word "white" from optics because the spectrum behaves the way sunlight does after it has been broken into colors and recombined. White light carries roughly equal energy per wavelength across the visible band; white noise carries roughly equal energy per frequency across the audible band. The technical phrase for this is a flat power spectral density, and it is the same idea that National Instruments describes when referring to wide-band noise as flat in the frequency domain, and that MathWorks documents as the defining property of a band-limited white noise source.
Flatness does not mean loudness. A perfectly white signal can be very quiet or very loud; the shape of the spectrum is what counts. It also does not mean every sample is the same number. The samples in white noise jump around from one instant to the next; the only constant is how much average energy sits in each frequency bin once you average over time. That averaging step is what separates a true white-noise generator from a random tone generator, and it is why any implementation that repeats the same sample, or that uses a stepped pattern, is not actually producing white noise at all.
How a Digital White Noise Generator Builds Its Signal
A digital white-noise generator is a small pipeline with three jobs. First, it asks the browser for the current sample rate, which the Web Audio API exposes through the AudioContext. The Lizely tool accepts anything from 8,000 Hz up to 192,000 Hz so it can run on common hardware without complaint. Second, it allocates exactly two seconds of mono Float32 storage and fills that buffer with pseudo-random numbers drawn uniformly from -1 inclusive to 1 exclusive. Third, it hands the buffer to a looping AudioBufferSourceNode that plays the same samples over and over through a single GainNode you control with the slider.
At the highest accepted rate, that fixed two-second budget works out as follows:
192,000 samples/second × 2 seconds = 384,000 samples 384,000 samples × 4 bytes per Float32 = 1,536,000 bytes ≈ 1.5 MiB
Capping the buffer at two seconds is a deliberate safety choice: it prevents an unusual AudioContext configuration from asking the browser for an unbounded chunk of memory. The signal is generated with a fresh 32-bit seed every time you press Start, using the three-shift xorshift32 recurrence described by George Marsaglia in the Journal of Statistical Software. The generator is fast and repeatable for audio purposes but is not cryptographically secure, so it should not be used to make keys, tokens, or anything that needs unpredictability against an attacker.
How to Use the Lizely White Noise Generator
Because the tool is browser-local and starts at a conservative level, the workflow stays small. Three steps are enough:
- Set your device output low, then leave the tool at its conservative 15% starting volume or reduce it further.
- Select Start white noise and adjust the 0–100% slider only as needed.
- Select Stop before leaving the page or whenever the sound feels uncomfortable.
There is no file to download, no account to create, and no microphone permission requested. Start creates exactly one AudioContext, one looping AudioBufferSourceNode, and one GainNode. Stop tears all three down and closes the context, so repeated presses cannot stack overlapping players. If the browser refuses to start audio for any reason, the tool surfaces an error and leaves no stale playing state behind.
Uniform vs Gaussian Amplitude: Two Flavors of White
The Lizely tool produces what is technically called bounded uniform white noise: every sample is drawn from a flat distribution between -1 and 1. Many audio references describe white noise as if it were Gaussian, meaning the amplitudes cluster around zero and the tails taper off toward infinity. Both can be white when their samples are uncorrelated and their expected power spectral density is flat, but they sound and measure differently.
| Property | Uniform white noise | Gaussian white noise |
|---|---|---|
| Sample range | Bounded, e.g. -1 to 1 | Theoretically unbounded |
| Histogram shape | Flat across the range | Bell curve, peak at zero |
| Mean-square power (unit scale) | 1/3 | 1 |
| Sounds white? | Yes, when samples are uncorrelated | Yes, when samples are uncorrelated |
| Easy to generate in a browser | Yes, one multiply per sample | Requires a Box–Muller transform |
The Lizely fixture suite confirms range, near-zero mean, uniform mean-square power of 1/3, RMS equal to sqrt(1/3), and a low lag-one product, which is what you would expect from an uncorrelated uniform source. The choice of distribution mostly matters when the noise is feeding a downstream process that assumes Gaussian statistics, such as some perceptual models; for casual listening, masking, or routing tests, uniform white noise sounds effectively identical to Gaussian white noise.
The Real Limits of Band-Limited Browser White Noise
Real audio systems cannot deliver infinite bandwidth. The highest sample rate accepted by the tool, 192 kHz, places the Nyquist limit at 96 kHz, which is well above human hearing but still finite. The lowest accepted rate, 8 kHz, gives a Nyquist limit of 4 kHz, which means everything above that frequency simply does not exist in the signal. In every case the generator remains flat across the band it can actually reproduce, but a browser generator is not a substitute for a certified acoustic test instrument. Readers who want to dig into how close a browser implementation really gets can read how accurate a browser white noise generator is in a separate guide.
Practical limits also come from the playback chain. The volume slider controls only this page's linear GainNode; your operating system mixer, browser audio pipeline, speakers, headphones, amplifier, and the acoustics of the room all sit downstream of that node and shape what your ears actually receive. A change in any one of those stages can move the perceived loudness up or down without the slider moving at all, so the slider value is a tool setting, not a measurement of sound pressure.
Listening Safely With the Volume Slider
The default 15% gain exists because loudness is easy to underestimate, especially with headphones. A low starting value gives you room to raise the slider before the sound reaches a level that is uncomfortable. From there, the slider is the only gain stage this page exposes, and it accepts any value between 0% and 100%.
| Slider position | What it does |
|---|---|
| 0% | Silences the output by zeroing the gain stage |
| 15% (default) | Conservative starting tool gain, not a safety guarantee |
| 100% | Maximum tool gain, not endorsed as safe |
None of those positions tell you how loud the result will be in your specific setup. Start low, raise the slider only as needed, lower the device output or press Stop if the sound feels uncomfortable, and follow the guidance supplied with your audio equipment. The tool is a simple controllable signal source, not medical equipment, hearing protection, or a certified acoustic test instrument, and it does not claim to diagnose, prevent, or treat sleep, hearing, tinnitus, anxiety, or any other health condition.
For a deeper look, see White Noise Generator for Beginners: Quick Start Guide.