White Noise for Travel: Staying Focused on the Go
The cabin is louder than anything you would choose to add
A study that monitored sound continuously on two hundred flights across six aircraft groups found a median level of 83.5 A-weighted decibels, with means across flight phases ranging from 37.6 to over 110 dB(A). Nine of those flights exceeded the occupational eight-hour exposure limit of 85 dB(A) under the approach used by the US National Institute for Occupational Safety and Health.
Hold that against the hearing-safety figures: at or below 70 dB(A) is unlikely to cause hearing loss even after long exposure, and long or repeated exposure at or above 85 dB(A) can cause it. A long-haul cabin is not a quiet place you are adding sound to. It is already the loudest sustained environment most people voluntarily sit in, and the sound you add is a rounding error against it. That reverses the usual framing of travel audio advice: the question is not how much noise to introduce, it is how much of the existing noise you can take away before adding anything at all.
Where the energy actually sits
Nobody has published one-third-octave spectra for commercial cabins, but the same research group measured several aircraft and rail systems both A-weighted and C-weighted, and that pair carries real spectral information. A-weighting discounts low frequencies to approximate human sensitivity; C-weighting counts them. The gap between the two numbers is therefore a measure of how much of the sound lies below the range where speech sits.
Across five wide-body aircraft the cruise figures run from 69.5 dB(A) and 83.7 dB(C) on an A380-800 to 74.9 dB(A) and 87.9 dB(C) on an A350-900 — spans of fourteen and thirteen decibels, with the authors noting dominant peaks below 600 hertz. Four airport express rail systems gave spans from 11.3 to 17.0 decibels, with significant content below 100 hertz.
This is the single most useful fact about travel noise. Thirteen decibels of span means the overwhelming majority of the acoustic energy is in a region that A-weighted figures barely register and that your ear registers as pressure rather than as sound. It explains why earbuds do almost nothing — passive isolation is weakest at low frequencies — and it explains why cancellation helps more than its specification sheet suggests.
What cancellation is doing, and where it stops
Active noise control works by generating a signal that destructively interferes with the incoming one. The tutorial literature on the method is explicit about the constraint: the approach suits low-frequency noise, where the wavelength is long relative to the spacing between microphone and ear, and passive methods are more effective at high frequencies. Broadband and higher-frequency cancellation is hard because the system must predict the waveform far enough ahead to oppose it.
That constraint happens to align perfectly with the measurements above. A cabin's energy is concentrated exactly where cancellation is strongest, which is why the effect in a plane is dramatic in a way it is not in an office full of voices. Spend the money on cancellation for flights and trains; it is the wrong tool for a conversation three rows back, which is high-frequency, intermittent and unpredictable.
What cancellation leaves is a residual that fluctuates as the system tracks the input. Adding a steady layer over it gives the fluctuation something to hide in, which is the real and modest case for playing noise on a plane. Brown noise is the better match because its energy distribution resembles the cabin's own, so it blends rather than sitting on top as a separate hiss.
The level discipline that follows
Because the cabin is loud, every instinct pushes you to turn your own sound up to compete with it, and that instinct is the actual hazard. The cabin noise is largely outside your control; the level in your headphones is entirely within it, and it is the part that stacks on top of an environment already near the occupational limit.
So the rule for a plane is the opposite of the rule for a quiet room. Let the cancellation do the work, add the smallest layer that smooths the residual, and treat any urge to go louder as evidence that the headphones are not sealing rather than that the layer is too quiet. If the sound has to be loud to be audible over the cabin, you are not masking anything — you are adding to the dose.
This site will not work without a connection
It is worth being direct about this, because articles about travel audio routinely promise offline capability that the underlying tool does not have.
This site has no service worker and no web app manifest. Nothing is cached for offline use, so the pages and scripts have to be fetched over the network, and a browser tab left open before you lose signal is not a supported offline mode — it is a tab that may or may not survive being backgrounded. There is also nothing to download: the only audio file on the site is the timer's alarm, since every noise colour is generated in the browser rather than served as a file, and the alarm is not offered as a download.
The generated layers do run entirely on your device once the page has loaded, since they are synthesised in the browser rather than streamed. That is a genuine property and it is not the same as offline support. For a flight, the honest advice is to put long noise files on your device from somewhere that actually provides them, or use an application that generates sound locally and is installed rather than loaded.
Environments, and what changes between them
What differs between travel environments is less the level than where the energy is, and that changes which layer to reach for.
A cabin at cruise and a rail car at speed are the same problem: most of the energy is low, so cancellation plus a brown layer at a low level is the setup, and the span figures above show rail cars are, if anything, more low-weighted than aircraft.
A terminal or a station is a different problem wearing the same clothes. The level is lower but the content is speech and announcements, which is high-frequency, intelligible and therefore the most disruptive category in the reading literature. Cancellation is weak here. A flat white layer covers more of the speech range than brown does, and the useful constraint is that you want announcements to remain audible, so this is the one travel setting where leaving the covering sound incomplete is deliberate.
A hotel room is the closest thing to the home case, and the honest comparison from the figure is the scale of it: the residential row sits around 47.7 dB(A), more than twenty decibels below any of the travel rows. That number is an outdoor one — measured on a seventh-floor balcony with the window open — so it is a ceiling on what the room behind it would read rather than an indoor level, and the gap to a cabin is if anything wider than the figure shows. In a room that quiet, the intrusions are events rather than a floor — a corridor door, a lift, a neighbour — and a low layer that covers onsets is sufficient. A speaker beats headphones here for the same reasons it does at home: nothing presses on your ear for eight hours and you can still hear an alarm.
What to pack
Battery figures for audio playback vary by device, codec and screen state by more than any single number would survive, so this guide gives none, and the storage arithmetic for noise files depends entirely on the codec and bit rate you choose.
What to carry follows from one rule: take whatever lets you avoid turning the volume up. A set of cancelling headphones that seals properly, spare tips if they are in-ear, and the files already on the device rather than a plan to fetch them later. In that order, because the seal is what decides how loud the rest of it has to be.
References
- Zevitas, C. D., Spengler, J. D., Jones, B., McNeely, E., Coull, B., Cao, X., Loo, S. M., Hard, A.-K. & Allen, J. G. (2018). Assessment of noise in the airplane cabin environment. Journal of Exposure Science & Environmental Epidemiology 28(6), 568-578 (two hundred flights, six aircraft groups; mean levels across flight phases 37.6 to over 110 dB(A) with a median of 83.5 dB(A); nine flights exceeded the NIOSH/ACGIH 8-hour TWA limit of 85 dB(A))
- Lee, H. P., Kumar, S., Garg, S. & Lim, K. M. (2022). Assessment of in-cabin noise of wide-body aircrafts. Applied Acoustics 194, 108809 (cruise averages including A380-800 at 69.5 dB(A) and 83.7 dB(C), A330-300ER at 71.3 and 85.6, B787 at 72.7 and 86.9, A350-900 at 74.9 and 87.9; C-weighted levels more than 10 dB above A-weighted, with dominant peaks below 600 Hz)
- Lee, H. P., Kumar, S., Garg, S. & Lim, K. M. (2021). Comparison of Cabin Noise of Airport Express Rail Systems. Acoustics 4(1), 1-13 (four airport express rail systems; the difference between C-weighted and A-weighted Leq ranges from 11.3 to 17.0 dB, with significant low-frequency content below 100 Hz; Taoyuan Airport MRT lowest at 66.4 dB(A) and 81.4 dB(C))
- Kuo, S. M. & Morgan, D. R. (1999). Active noise control: a tutorial review. Proceedings of the IEEE 87(6), 943-975 (active control suits low-frequency noise, where passive methods are least effective; passive methods are more effective at high frequencies)
- Noise-Induced Hearing Loss — National Institute on Deafness and Other Communication Disorders (at or below 70 dBA unlikely to cause hearing loss even after long exposure; at or above 85 dBA can with long or repeated exposure)
- Vasilev, M. R., Kirkby, J. A. & Angele, B. (2018). Auditory Distraction During Reading: A Bayesian Meta-Analysis of a Continuing Controversy. Perspectives on Psychological Science 13(5), 567-597 (intelligible speech and lyrical music were the most distracting conditions tested)
- Yang, T. & Kang, J. (2023). Sound environment in an urban apartment building during and after the COVID-19 lockdown. Frontiers in Psychology 14, 1136201 (a post-lockdown day level of 47.7 dBA measured outdoors at a seventh-floor balcony with the window open, cited here as an outdoor residential facade level, not an indoor one)
- Colored noise: spectral slopes of white, pink and brown noise — IEEE Technology Navigator
Frequently Asked Questions
Is white noise or brown noise better on an airplane?
Brown noise, because its energy distribution resembles the cabin's own. Measurements of wide-body cabins put C-weighted levels thirteen to fourteen decibels above A-weighted ones, with dominant peaks below 600 hertz, which means the sound you are blending with is overwhelmingly low-frequency. A flat white layer adds high-frequency content the cabin does not have, so it sits on top as a separate hiss rather than blending.
Does this site work without an internet connection?
No. There is no service worker and no web app manifest, so nothing is cached for offline use and the pages must be fetched over the network. The generated layers do run on your own device once a page has loaded, because they are synthesised in the browser rather than streamed, but that is not the same as offline support and a backgrounded tab is not a reliable substitute.
Can noise files be downloaded from the site for a flight?
No. The only audio file on the site is the timer's alarm, and it is not offered as a download; the noise itself is generated in your browser rather than served as a file. For a flight, put long files on the device from a source that provides them, or use an application that generates sound locally and is installed rather than loaded.
Should the volume be higher on a plane because the cabin is loud?
The opposite. A median cabin level of 83.5 A-weighted decibels is already close to the 85 dBA figure at which long or repeated exposure can cause hearing loss, and your headphone level is the part of the total you control. Let cancellation reduce the cabin, add the smallest layer that smooths what it leaves, and treat the urge to go louder as a sign the headphones are not sealing.
Why does cancellation work so much better on a plane than in an office?
Because active control suits low-frequency sound, which is where almost all of a cabin's energy sits, while passive methods are more effective at high frequencies. An office's problem is speech: high-frequency, intermittent and unpredictable, which is the hardest case for cancellation and the easiest for a sealed earcup plus a flat masking layer.