The deep rumble of a powerful waterfall, distant thunder, and the muffled world inside a car — what do they have in common?
In one sentence: Brown noise is a random signal whose power decreases as the square of frequency — specifically, it falls off as 1/f² — meaning the lower the frequency, the stronger the energy, resulting in a deep, rumbling, "thunderous" sound that is even darker and smoother than pink noise.
While white noise is static and pink noise is like steady rain, brown noise is the deep, rumbling roar of a powerful waterfall or the low, enveloping hum inside a moving vehicle. It gets its name not from the color brown, but from Brownian motion (named after botanist Robert Brown) — the random, jittery movement of microscopic particles in a fluid. If you track the position of such a particle over time and turn that movement into sound, you get exactly this spectrum.
Mathematically, brown noise has a power spectral density (PSD) that is inversely proportional to the square of the frequency:
where C is a constant. This means that for every doubling of frequency (an octave), the power drops by a factor of four — a steep roll-off of 6 dB per octave. In the time domain, brown noise looks like a "random walk" — it drifts slowly up and down, creating smooth, sweeping curves with no sharp edges, making it the smoothest and deepest of the three.
The plot below shows a short segment of simulated brown noise in the time domain (top), together with its estimated power spectral density (bottom). Notice the smooth, drifting waveform — a stark contrast to the chaotic jaggedness of white noise — and the spectrum, which drops steeply as frequency increases.
Figure 1: Top: a short segment of simulated brown noise in the time domain — smooth, drifting, with no sharp jumps. Bottom: its estimated power spectral density, showing the characteristic 6 dB/octave downward slope (1/f² roll-off).
Brown noise sits at the extreme low-frequency end of the noise color spectrum, making it the "deepest" of the three main types:
- White noise: Flat PSD — equal power at every frequency. Sounds harsh and hissy, like TV static.
- Pink noise: PSD ∝ 1/f — 3 dB/octave roll-off. Sounds balanced, smooth, and natural, like rainfall.
- Brown noise: PSD ∝ 1/f² — 6 dB/octave roll-off. Sounds deep, rumbling, and powerful, like a waterfall or thunder.
Why is brown noise so captivating?
Brown noise has a primal, soothing quality that makes it perfect for deep relaxation and masking low-frequency disturbances:
- Deep sleep and relaxation: The low, rumbling character of brown noise is often compared to the sound heard inside the womb — which is why many people find it extremely calming and effective for falling asleep and staying asleep.
- Masking low-frequency noise: Brown noise is excellent at covering up low-frequency disturbances like traffic rumble, HVAC systems, or neighbour's bass — sounds that white and pink noise struggle to mask effectively.
- Subwoofer and bass testing: Because it concentrates energy in the lowest frequencies, brown noise is the standard test signal for evaluating subwoofers and low-frequency response in audio systems.
- Relaxation and focus: Many people use brown noise as a background sound for studying, working, or meditating, as it provides a consistent, non-intrusive "cocoon" of sound that blocks out distractions.
- Data simulation: Brown noise is often used in financial modeling (stock price simulations) and statistical physics, as the random walk (Brownian motion) is a fundamental model for many natural and economic processes.
Practical limits and common pitfalls
Like its lighter cousins, brown noise comes with its own set of caveats:
- Infinite energy at low frequencies: Mathematically, 1/f² becomes infinite as f approaches zero — meaning a perfect brown noise signal would have infinite energy at DC (0 Hz). In practice, brown noise is always band-limited and high-pass filtered to prevent this.
- Not suitable for all speakers: Because of its heavy bass emphasis, playing brown noise at high volume through small speakers or headphones can cause distortion or even damage — always start at a low volume.
- Generating true brown noise is computationally easy but requires careful normalisation: Unlike pink noise (which requires complex filter banks), brown noise can be generated simply by integrating (cumulatively summing) white noise — but this often creates a DC offset that must be removed and requires scaling to prevent clipping.
- It's not a magic sleep cure: While brown noise helps many people fall asleep, it doesn't work for everyone — some people find the rumbling distracting rather than relaxing. Experiment to find what works for you.
In machine learning and audio processing, brown noise is sometimes used as a stress test for models handling very low-frequency content, and it appears in generative models as a source of smooth, continuous variation — representing the "slow drift" of natural processes, from temperature changes to stock market fluctuations.
The takeaway: brown noise is the deep, grounding anchor of the noise family — not as hissy as white, not as airy as pink, but rich, rumbling, and primal. It reminds us that some of the most powerful sounds are the ones we feel in our chests as much as we hear with our ears.