Noise Generator — White to Grey

This noise generator plays white, pink, brown, blue, violet and grey noise from an AudioWorklet in your browser, with a dBFS level slider, a live readout and a 2-second WAV export. Use it to test speakers, headphones and rooms, to mask background sound, or to hear how spectral slope changes timbre.

How to use it

1. Pick a color. Six buttons select white, pink, brown, blue, violet or grey, and the hint line under them describes the current choice.
2. Set the level. The slider starts at 30%, which the readout shows as −10.5 dBFS. Leave headroom rather than pushing it up, especially for the high-frequency colors.
3. Press Play Noise. Audio starts after a 20 ms fade, the state switches to Playing, and the third readout shows the device sample rate, for example 48.0 kHz.
4. Compare colors while the sound runs. Switching color mid-stream changes the spectrum immediately, which is the quickest way to hear what a 3 dB or 6 dB per octave tilt does.
5. Download a clip if you need one. Download WAV renders 2 seconds of the current color at the current level and saves it as a mono 16-bit WAV file.
6. Press Stop to end playback. The processor fades out and is released; nothing is recorded or uploaded at any point.

The six noise colors and what their slopes mean

Noise is defined by how its energy is distributed across frequency, and the color names are shorthand for that distribution. Each slope below is the amplitude change per doubling of frequency.

Color	Slope	Energy description	Typical use
White	0 dB/octave	Equal energy per hertz, flat spectrum	Reference signal for analysers, bright masking
Pink	−3.0 dB/octave	Equal energy per octave	Room and speaker checks, balanced masking
Brown	−6.0 dB/octave	Leaky-integrated white noise	Masking low rumble, subwoofer checks
Blue	+3.0 dB/octave	First difference of white noise	Tweeter and high-frequency driver tests
Violet	+6.0 dB/octave	Second difference of white noise	Top-octave driver work and high-frequency masking
Grey	Fixed 40/60 white and pink blend	Rough loudness weighting	Masking when white is harsh and pink is dull

The octave arithmetic explains the character. The audible band from 20 Hz to 20 kHz spans log2(20000 / 20) = 9.97 octaves, so pink noise spreads its power evenly across roughly ten octaves while white noise spreads the same total power evenly across 19,980 hertz of bandwidth. About 95% of white noise’s energy therefore sits above 1 kHz, which is why white hisses and pink sounds balanced when both are played at the same RMS level.

Crest factor and why noise needs headroom

Noise is a random signal, so its peaks sit well above its average level: Gaussian noise has a crest factor near 12 dB, which means peaks around four times the RMS value. A noise signal with an RMS of −20 dBFS will therefore peak near −8 dBFS, and that 12 dB gap is the headroom a noise generator needs to avoid clipping. The level slider is a gain control in dBFS, not an SPL calibration: 30% on the slider reads −10.5 dBFS because 20 × log10(0.3) = −10.5, and the same setting produces very different perceived loudness in white and violet because the ear’s sensitivity changes so much across the band.

Choosing a color for the job

• Pink noise for rooms and speakers. Equal energy per octave matches the way we hear, so a pink noise sweep through a room shows tonal problems clearly on a spectrum analyser; deviations from a downward-sloping line point at resonances and absorption problems.
• White noise for flat-per-hertz work. Use it when the measurement is defined in hertz or with an FFT that averages many frames, and for straightforward masking of broadband sound.
• Brown noise for low-frequency work. Its tilt puts most energy in the bass, which suits masking traffic or rumble and checking that a subwoofer is producing output without bottoming out.
• Blue and violet for high-frequency drivers. These colors concentrate energy where tweeters live, so they reveal a dead tweeter quickly. Because the ear is sensitive up there, set the slider low before pressing Play.
• Grey for comfortable masking. The fixed blend approximates the ear’s weighting, so it sounds more even than plain white for speech masking and long listening sessions.

Accuracy, limits and troubleshooting

• The channels are identical. The worklet generates one channel and copies it to every output, so the stereo image sits in the center with no decorrelation between left and right.
• Brown noise is a leaky integrator. The one-pole recurrence flattens below a corner set by the sample rate, around 70 Hz at 44.1 kHz and 76 Hz at 48 kHz, instead of rising forever as an ideal 1/f² source would. The slope above that corner is the expected 6 dB per octave.
• Grey is an approximation. It is a 40% white and 60% pink blend, which tracks speech and music masking well but is not a measured equal-loudness contour.
• Sample rate sets the ceiling, not the slope. At 44.1 kHz the top of the spectrum is 22.05 kHz, and the blue and violet tilts continue up to that limit.
• Color changes are instant. That is useful for A/B comparison and it also means a loud color can arrive without warning, so keep one hand on Stop.
• Device processing colors the result. Operating system enhancements, Bluetooth codecs and speaker DSP all reshape the extreme top and bottom. Before blaming the generator, listen on wired headphones.
• Not a calibrated level source. All readings are dBFS, relative to full scale. For acoustic levels use a meter and a calibrated reference.
• Hearing care. High-frequency colors put their energy where hearing is most sensitive, so start low with blue and violet and never raise the level just to make a tone audible. If you are concerned about your hearing, see an audiologist.

Related tools

• Frequency Sweep Generator: glide a sine across the band to find the exact frequency where a driver gives up.
• Audio Spectrum Analyzer: watch the six slopes appear in real time and check how a room reshapes pink noise.
• Sleep Sounds: longer-form ambient textures if the goal is background sound rather than testing.
• Learn: colors of noise explained: the full reference on white, pink, brown, blue and violet spectra with the underlying formulas.

Aliases covered on this page: noise generator, white noise generator, pink noise generator, brown noise generator, and blue noise generator.