Noise generator
Before you start: Noise can sound intense even when a tone seems quiet. Start with low device volume and a short duration. Stop if uncomfortable. Do not use continuous playback to assess safe exposure.
Noise generator controls
Choose a noise spectrum and keep the seed unchanged when you need the same repeatable reference pattern.
Loading audio controls.
Create repeatable white, pink or brown noise without uploading audio or relying on a streamed file.
Interpretation: A seeded noise reference can be repeated across path changes to expose obvious dropouts, rattles or channel differences.
Limit: Noise playback does not calibrate a room, prove speaker frequency response or measure physical sound pressure.
Safety and privacy: No sound starts automatically. Signals and analysis stay in your browser by default. Browser dBFS is not physical SPL.
How to use this test
- Choose the spectrum: White noise is equal by frequency, pink is balanced by octave and brown emphasizes lower frequencies.
- Keep the seed: Use the same numeric seed when the comparison requires the identical generated pattern.
- Compare at one level: Do not change browser or device volume between the paths you are comparing.
Choose the signal level: New references start at -40 dBFS peak. Select up to 0 in 5 dB steps, then press Play. This is WaveLocus's selectable range, not an industry standard or a physical-volume measurement. Zero is full-scale digital output; lower device volume before playback. Noise also offers separately labelled source-gain and RMS modes with a peak guard.
Reference and scope: SMPTE specifies -21.5 dBFS RMS for a particular calibrated pink-noise waveform with additional bandwidth, crest-factor and uniformity requirements. This general-purpose generator is not that calibrated waveform. Do not treat selecting a number as calibration. SMPTE reference-noise specification.
Why noise needs its own level control
White, pink and brown noise have different RMS and crest factors. Choose source gain, source peak target or source RMS target explicitly. Targets describe the finite source before fades and channel attenuation; the rendered readout includes fades. The peak guard can reduce gain below the requested target. SMPTE specifies −21.5 dBFS RMS for a particular pink-noise reference, with bandwidth, crest-factor, period and uniformity requirements. This general-purpose generator has not established compliance with that waveform. Selecting the same level does not establish calibration.
Source: SMPTE reference-noise specification. Read dBFS vs dB SPL to distinguish digital signal levels from physical sound pressure. Use the speaker test to record whether a broadband reference reveals a repeatable dropout or rattle.
How to interpret your noise generator results
Peak and RMS differ: Peak describes the largest sample; RMS averages squared samples over the reference, including fades. Different waveforms and noise colours can have different RMS at the same source gain.
A peak guard limits the requested target: The displayed measured output takes priority over the requested setting. Do not treat an unreached RMS target as a successful match.
Noise colours sound different: Their spectra differ by design. The Nyquist boundary is not a measured flat passband, and these references are not calibrated SMPTE noise.
Source: REW level-setting procedure.
Frequently asked questions
What is the difference between white, pink and brown noise?
White noise has equal power per hertz. Pink noise reduces power toward higher frequencies, with equal power per octave. Brown noise is weighted more strongly toward low frequencies.
Why keep the noise seed unchanged?
The seed makes the generated reference repeatable. Use the same colour, seed, level and duration when comparing one change in your audio path.
Can I leave the noise playing overnight?
This generator provides bounded, short audio-test references rather than continuous background playback. It stops at the selected duration. Its digital level setting does not establish safe listening exposure.