> ## Content Index
> Fetch the complete content index at: https://soundforgepro.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Brown Noise Generator: Deep, Continuous Brown Noise Online
- URL: https://soundforgepro.com/brown-noise-generator/
- Published: 2026-10-28T21:26:56.000Z
- Updated: 2026-10-28T21:26:56.000Z
- Description: Brown noise generated live in your browser, not looped: four filter depths, stereo or mono, a capped level, a sleep timer with a five-minute fade and a WAV loop save. Below it: what brown noise is in numbers, why it seems to swell, why phones play so little of it, and how loud to keep it.
- Author: Erick Finn
- Tags: Audio Tools, Audio Fundamentals

Brown noise is random noise whose power drops 6 dB with every octave you go up, so it sounds like a low rumble or a distant waterfall instead of hiss. The generator below makes it live on your device, sample by sample, so there's no loop to notice. Pick one of four depths, stereo or mono, and an optional sleep timer that fades out over five minutes. Start with your system volume low and stop raising it once the sound you're trying to cover blends in.

Browser tool · brown noise

## Generate brown noise

Made live on this device, sample by sample, so it never repeats. No microphone, no upload. Left and right are independent unless you pick Mono.

**Turn the system volume down before Play.** Tool level starts at 5%, caps at 25% and is not dB SPL. For sleep, stop raising it once the sound you want to cover blends in, keep the speaker off your pillow, and never put a speaker inside a crib.

**+6 dB shelf, 1 kHz**Light **\-6 dB per octave**Standard **Low-pass 500 Hz**Deep **Low-pass 200 Hz**Deeper 

Depth 

**Standard**\-6 dB/oct

1001k10kHzdB 

90.1% of audible power below 250 HzdBA reads 10.5 dB under dBCTimer off

Playback**Stopped**

Press Play. Depth, channels, timer and level change live while it plays.

PlayStop

Channels

Stereo · independent L/RMono · same in both ears

Sleep timer · 5-minute fade-out

Off15 min30 min45 min60 min90 min2 h

Lowest note your speaker plays

Full range40 Hz60 Hz100 Hz150 Hz200 Hz

Full range counts the audible band, 20 Hz to 20 kHz. Another 36.1% of this setting sits below 20 Hz, under hearing. Pick the lowest note your speaker plays to see how much audible power is left.

**Keep it playing:** leave this tab open and the device on a charger. On iPhone and iPad, locking the screen or switching apps pauses web audio, so keep the screen on or save a WAV loop below.

Tool level

5%

Loop length

30 s60 s

Sample rate

44.1 kHz48 kHz

Save WAV loop 

16-bit stereo WAV, Standard depth: 11,520,044 bytes (11.5 MB). The last sample runs straight into the first, so it repeats without a gap. Peaks stay below -1 dBFS; start your player quietly.

Ready. Pick a depth, then press Play.

Gaussian noise through a 10 Hz leaky integrator and a 15 Hz high-pass, falling about 6 dB per octave above about 60 Hz. Depth uses Web Audio filters. Readouts come from the filter math at 48 kHz, not from your speaker. Not calibrated output or an SPL reading. Press Escape to stop.

## How to use the brown noise generator

1. **Lower the volume first.** Turn the phone or computer volume down before you press anything. The tool level opens at 5% and can't go past 25%. It's a digital level, not a reading in dB SPL.
2. **Choose a depth.** Standard is plain brown noise. Light adds a treble shelf for small speakers, and Deep and Deeper cut the top end with a low-pass filter. The curve below the depth buttons redraws for each one.
3. **Press Play.** The sound rises over three seconds. Stop, or the Escape key, takes it down in 0.4 s.
4. **Pick stereo or mono.** Stereo sends a different noise stream to each ear. Mono sends the same one to both.
5. **Set the sleep timer.** Choose 15 to 90 minutes or 2 hours. The last five minutes fade slowly to silence, then playback stops.
6. **Save a WAV loop if you need one.** Pick 30 or 60 seconds and 44.1 or 48 kHz, then press Save WAV loop. The file uses the depth and channel setting you have selected.

Depth and channel changes crossfade in 0.3 s while the noise plays, and level changes glide over 80 ms instead of jumping. A new timer choice counts from the moment you pick it. Pick one during the final fade and the sound climbs back to your level over two seconds. The speaker setting changes only the readout, never the sound: it shades the part of the curve your speaker can't play and tells you how much audible power is left. When a depth other than Standard is selected, a dashed line shows Standard for comparison. Depth, channels and speaker are remembered on this device; timer and level start fresh on every visit.

## What brown noise is, in numbers

Brown noise follows a 1/f² power curve. The [University of Illinois acoustics lecture notes on noise](https://courses.physics.illinois.edu/phys406/sp2017/Lecture%5FNotes/P406POM%5FLecture%5FNotes/P406POM%5FLect15.pdf?ref=soundforgepro.com) give its slope as 20 dB per decade, which works out to 6.02 dB per octave: every time the frequency doubles, each hertz carries a quarter of the power. The same notes tie the name to Brownian motion, the random walk of pollen grains in water, so "brown" refers to the physics and not a color. Some generators call it red noise; the two names mean the same signal.

This generator feeds Gaussian random numbers through a 10 Hz leaky integrator and then a 15 Hz high-pass filter. Those two corners keep the signal from drifting away from zero. From 100 Hz to 4 kHz the slope sits within about 0.1 dB per octave of the ideal 6.02\. Lower down it flattens to 5.4 dB per octave between 40 and 80 Hz, and the spectrum peaks near 12 Hz. Near the top it eases to 4.8 dB per octave between 8 and 16 kHz, where the level is already about 40 dB below 100 Hz.

That peak has a cost. About 36% of the Standard setting's total power sits below 20 Hz, under the 20 Hz to 20 kHz band that the [audio frequency chart](https://soundforgepro.com/audio-frequency-chart/) treats as the hearing range. In a listening study covered further down, none of 40 listeners heard a 20 Hz tone on any of four playback devices. So the readouts in the tool count shares of the audible band only, and list the infrasonic part separately.

## Light, Standard, Deep or Deeper: what each depth changes

Each depth is a named filter, not a mood label. Deep and Deeper use the Web Audio low-pass filter, which the [W3C Web Audio specification](https://www.w3.org/TR/webaudio-1.1/?ref=soundforgepro.com) defines as a second-order filter with a 12 dB per octave rolloff. Here its Q is set for a flat Butterworth response, 3 dB down at the corner. Light uses a high shelf that adds 3 dB at 1 kHz and about 6 dB at the top of the band.

__Computed from this generator's filter math at 48 kHz, with shares of audible power (20 Hz to 20 kHz) and flat playback. Loudness change is K-weighted at the same tool level.__
| Depth    | Filter                                | Audible power below 250 Hz | Audible power above 1 kHz | Loudness vs Standard | dBC minus dBA |
| -------- | ------------------------------------- | -------------------------- | ------------------------- | -------------------- | ------------- |
| Light    | High shelf at 1 kHz, +6 dB at the top | 84.4%                      | 7.7%                      | +1.1 dB              | 7.6 dB        |
| Standard | None, -6 dB per octave                | 90.1%                      | 2.4%                      | 0 dB                 | 10.5 dB       |
| Deep     | Low-pass 500 Hz, 12 dB/oct            | 95.1%                      | 0.03%                     | \-0.8 dB             | 14.9 dB       |
| Deeper   | Low-pass 200 Hz, 12 dB/oct            | 99.2%                      | 0.001%                    | \-1.7 dB             | 19.6 dB       |

I'd start with Standard, because it's the true 1/f² curve and the other three are measured departures from it. Deep and Deeper strip out nearly everything above 1 kHz, the faint hiss that remains in Standard, while the overall loudness drops by less than 2 dB. Light goes the other way: it keeps a little more texture in the range small speakers reproduce well.

Brown noise is a poor match for voices. Even Standard puts only 2.4% of its audible power above 1 kHz, while speech detail runs from about 500 Hz to 2 kHz and intelligibility sits at 2 to 4 kHz on the frequency chart. For chatter through a wall or an open office, the [white noise generator](https://soundforgepro.com/white-noise-generator/) fits better, since white noise keeps as much power per hertz at 3 kHz as at 100 Hz. Brown suits low, steady sources: traffic, a fridge compressor, an air handler or a neighbor's bass.

## Why brown noise seems to swell and fade

Listeners often hear brown noise rise and dip even when nothing changes. Mind Amend, whose brown noise uploads have tens of millions of views, explains in the [notes for its 12-hour Super Deep Smoothed Brown Noise video](https://www.youtube.com/watch?v=P48QELwruQs&ref=soundforgepro.com) that low frequencies sometimes bunch together at random, and that this "can create an illusion of the volume dipping when it hasn't changed at all."

The reason is how few low-frequency components carry the power. Two minutes of this generator's output, read with the K-weighting and 400 ms windows of a [LUFS meter](https://soundforgepro.com/lufs-meter/), wanders over a 1.3 dB span for Standard and 1.9 dB for Deeper (5th to 95th percentile). White noise under the same reading spans 0.15 dB, because its power is spread across thousands of hertz that average out. Over 3-second windows Standard settles to a 0.5 dB span.

Here the swell is only that random wobble. The generator has no loop point and nothing fades on a schedule except the sleep timer. Recordings can add a real one: the [notes for Mind Amend's remastered 8-hour track](https://www.youtube.com/watch?v=RqzGzwTY-6w&ref=soundforgepro.com) say the original had an unintentional fade in and out every hour.

## Why brown noise sounds thin on a phone or laptop

Small speakers lose exactly the band brown noise lives in. A [2020 microspeaker study in Applied Sciences](https://doi.org/10.3390/app10248902?ref=soundforgepro.com) states that the limited space in modern devices "leads to poor acoustic performance, especially at low frequencies." With 90.1% of Standard's audible power below 250 Hz, a phone speaker plays only the small top slice of the signal.

### Test results: low tones on four devices

The [HALT headphone and loudspeaker test study (Behavior Research Methods, 2023)](https://pmc.ncbi.nlm.nih.gov/articles/PMC9113065/?ref=soundforgepro.com) played 20, 60, 100 and 140 Hz tones at -1 dBFS true peak to 40 listeners on four devices and asked each one to count the tones heard. The authors describe the setup as a best case, with the devices "pushed to the limit."

__Listeners out of 40 who heard each tone, from Table 6 of Wycisk et al. 2023\. Tones at -1 dBFS true peak, between sections of loud pink noise.__
| Playback device                                      | 20 Hz | 60 Hz | 100 Hz | 140 Hz |
| ---------------------------------------------------- | ----- | ----- | ------ | ------ |
| Studio headphones (Beyerdynamic DT 770 Pro, 250 ohm) | 0     | 40    | 40     | 40     |
| Studio loudspeakers (Yamaha HS8M)                    | 0     | 40    | 40     | 40     |
| Low-quality no-name earbuds                          | 0     | 26    | 39     | 40     |
| Laptop speakers (2015 13-inch MacBook Pro)           | 0     | 5     | 38     | 40     |

Across the four devices, the 60 Hz tone reached a median of 33 of 40 listeners, with a range of 5 to 40\. At 100 Hz the median was 39.5, range 38 to 40\. Nobody heard 20 Hz on any device and everybody heard 140 Hz. It's one study, so treat the counts as a clear pattern rather than a spec: the laptop is the outlier, and the authors note its A-weighted level ran higher than its unweighted level, a sign that most of its output sat between 1 and 6 kHz.

Put those limits into the speaker readout. If a speaker stops near 100 Hz, Standard keeps 24.6% of its audible power, 6.1 dB down, and Deeper keeps 13.2%, 8.8 dB down. Earbuds that reach 60 Hz leave Standard with 40.3%. My pick for a phone or laptop is Light: its shelf adds energy where those speakers work, and it keeps 29.4% of its sound above a 100 Hz limit. To find where your own speaker gives out, step through the tones in the [bass test](https://soundforgepro.com/bass-test/) and set the speaker chip to match.

## How loud should brown noise be?

Set it by ear at the lowest level where the sound you want to cover stops standing out. For a number, put the [decibel meter](https://soundforgepro.com/decibel-meter/) at pillow height; its guidance is to "use A (dBA) for everyday noise and hearing questions, C (dBC) for bass-heavy noise and peaks."

Expect brown noise to read low in dBA. The [IEC 61672 weighting table published by NTi Audio](https://www.nti-audio.com/en/support/know-how/frequency-weightings-for-sound-level-measurements?ref=soundforgepro.com) shows A-weighting turning 63 Hz down by 26.2 dB, against 0.8 dB for C-weighting. With full-range playback, Standard reads 10.5 dB lower in dBA than in dBC, and Deeper reads 19.6 dB lower. Small speakers shrink that gap because they never play the deepest part. Compare dBA readings with hearing limits, which are set in A-weighted terms.

The [NIOSH recommended exposure limit](https://www.cdc.gov/niosh/noise/about/index.html?ref=soundforgepro.com) is 85 dBA averaged over an eight-hour workday. The [WHO safe listening guidance](https://www.who.int/news-room/questions-and-answers/item/deafness-and-hearing-loss-safe-listening?ref=soundforgepro.com) allows 80 dB for up to 40 hours a week, and eight hours a night for seven nights comes to 56 hours. I'd keep overnight noise far below both figures, at a level someone beside the bed could talk over without raising their voice.

For a baby, the American Academy of Pediatrics lists [sounds that remind babies of the womb](https://www.healthychildren.org/English/ages-stages/baby/crying-colic/Pages/Calming-A-Fussy-Baby.aspx?ref=soundforgepro.com) among calming options, and published machine tests set the conditions. In a [Pediatrics study of 14 infant sleep machines](https://pubmed.ncbi.nlm.nih.gov/24590753/?ref=soundforgepro.com), every one went past the 50 dBA hospital nursery limit at 30 cm on maximum volume. A [2021 test of 14 machines and phone apps](https://pubmed.ncbi.nlm.nih.gov/33992973/?ref=soundforgepro.com) found 9 over 85 dBA at 10 cm on maximum, none over at 30 or 100 cm, and none at minimum volume. Its authors advise avoiding the maximum setting and keeping the device well outside the crib or at least 30 cm from the child.

## Stereo or mono brown noise

In stereo, each ear gets its own random stream. Over 40 seconds of output the left and right channels correlate at under 0.08, close to fully independent. A [study of perceived width in the Journal of the Acoustical Society of America](https://pubmed.ncbi.nlm.nih.gov/24907818/?ref=soundforgepro.com) found that noise sounds wider as the match between the ears drops, and that listeners with poorer binaural fine-structure thresholds, which tracked age, heard less of that widening.

I'd use stereo on headphones or a pair of speakers, where the wide, enveloping sound covers noise from every side. Mono suits a single speaker, which mixes the channels anyway, and anyone who finds the wide version distracting. The [left and right speaker test](https://soundforgepro.com/left-right-speaker-test/) confirms which side is which before you settle in.

## What the research says about brown noise

Brown noise has far less study behind it than its popularity suggests.

- **Attention and ADHD.** A [2024 meta-analysis of 13 studies with 335 participants](https://pubmed.ncbi.nlm.nih.gov/38428577/?ref=soundforgepro.com) found a small benefit of white and pink noise on task performance in people with ADHD or its symptoms (g = 0.249), and no brown noise studies at all. A [2025 study of 26 adults](https://pmc.ncbi.nlm.nih.gov/articles/PMC12576875/?ref=soundforgepro.com) then tested brown noise on a visual attention task and found no significant effect on attentional efficiency, while slow beat music helped.
- **Arousal.** A [2025 pupil study](https://pubmed.ncbi.nlm.nih.gov/41110471/?ref=soundforgepro.com) analyzed 31 adults listening to white, pink and brown noise and found no difference in pupil size, which the authors use as a marker of autonomic nervous system activity.
- **Sleep.** A [2021 systematic review](https://pubmed.ncbi.nlm.nih.gov/33007706/?ref=soundforgepro.com) rated the evidence for continuous noise improving sleep as very low quality and warned it could also harm sleep and hearing. A [2022 review of 34 studies and 1,103 people](https://pmc.ncbi.nlm.nih.gov/articles/PMC9163611/?ref=soundforgepro.com) counted positive findings for white, pink and mixed audio, with no brown noise category.
- **Tinnitus.** A [Cochrane review of 8 studies and 590 participants](https://pubmed.ncbi.nlm.nih.gov/30589445/?ref=soundforgepro.com) found no evidence that sound therapy beats a waiting list, placebo or education. The [NIDCD tinnitus page](https://www.nidcd.nih.gov/health/tinnitus?ref=soundforgepro.com) says to see a primary care doctor first, and describes tabletop and phone sound generators as an aid for relaxation or sleep.

So brown noise is a masking and comfort tool. Use it because it hides a sound you don't want and you like how it sounds, not as a treatment.

## Live brown noise vs a loop, and the WAV file

The player runs in an AudioWorklet, which [MDN describes](https://developer.mozilla.org/en-US/docs/Web/API/AudioWorklet?ref=soundforgepro.com) as custom audio code running in a separate thread with very low latency. It draws new random samples for as long as it plays, so no stretch ever repeats.

Loops are a different case. In a [study of repeated noise](https://pubmed.ncbi.nlm.nih.gov/15620225/?ref=soundforgepro.com), untrained listeners could tell a repeating noise from fresh noise with cycles up to 20 seconds long, with performance falling most between 5 and 10 seconds. That's why the WAV comes in 30 or 60 seconds, past that 20-second mark. The file is filtered as a closed circle, so its last sample leads straight into the first with no step at the joint. Set your player to repeat one track.

Size is sample rate times channels times 2 bytes times seconds, plus a 44-byte header. A 60-second stereo file at 48 kHz is 11,520,044 bytes, about 11.5 MB; 30 seconds of mono at 44.1 kHz is 2,646,044 bytes. The [audio file size calculator](https://soundforgepro.com/audio-file-size-calculator/) handles other lengths and formats. The tool level slider doesn't apply to the file: it's saved at a fixed level with peaks kept below -1 dBFS, so start the player quietly.

## Limits of this brown noise generator

- The readouts come from the filter math at 48 kHz, not from your speaker or room. The tool level is a digital level, not an SPL reading.
- The speaker setting is a calculator. It changes the numbers and the shaded curve, never the sound.
- On iPhone and iPad, [MDN notes](https://developer.mozilla.org/en-US/docs/Web/API/BaseAudioContext/state?ref=soundforgepro.com) that leaving the page or turning off the screen interrupts web audio. Keep the screen on, or save a WAV loop and play it in a music app.
- The live generator needs a secure HTTPS page. Where it can't start, the tool plays a 60-second stereo loop of Standard brown noise through the same filters, so depth, channels and level still work.
- Sound starts only from a tap or click, because browsers apply [autoplay rules](https://developer.mozilla.org/en-US/docs/Web/Media/Guides/Autoplay?ref=soundforgepro.com) to audio started any other way.

My verdict: Standard brown noise in stereo, through a speaker with a real woofer placed across the room, at the lowest level that hides the rumble you're fighting. Switch to Light on a phone or laptop, use Deep when the faint hiss bothers you, and let the timer fade it out after you fall asleep.

## Brown noise generator FAQ

### Is brown noise better than white noise for sleep?

The 2022 sleep review had no brown noise category at all, and the 2021 review rates the evidence for any continuous noise as very low. Brown is better at covering low rumble such as traffic or a neighbor's bass, and it has far less hiss, with 2.4% of its audible power above 1 kHz. White covers voices better.

### Why does brown noise sound like it's getting louder and quieter?

Its power sits in a few low frequencies that bunch together at random, so its short-term level wanders. On this generator the 400 ms loudness spans 1.3 dB at Standard against 0.15 dB for white noise. Nothing in the tool moves the level on its own except the sleep timer's fade.

### Is brown noise the same as red noise?

Yes. Both names describe noise with a 1/f² spectrum that falls 6 dB per octave. "Brown" comes from Brownian motion, and other generators label the same signal red.

### How loud should brown noise be for sleep?

Only loud enough that the sound you want to hide stops standing out. WHO's safe-listening figure of 80 dB for up to 40 hours a week is already exceeded by eight hours a night at that level, so stay far below it, and for a baby keep the device outside the crib and off maximum volume.

### Does brown noise help with ADHD or focus?

There isn't evidence for it yet. The 2024 ADHD meta-analysis found a small effect for white and pink noise but no brown noise studies, and a 2025 attention study found no significant effect of brown noise.

### Can brown noise help tinnitus?

It can mask tinnitus while it plays, but a Cochrane review found no evidence that sound therapy beats a waiting list, placebo or education. See a doctor first, as the NIDCD advises, and treat the noise as a relaxation aid.

### Why does brown noise sound weak on my phone?

About 90% of its audible power sits below 250 Hz, and phone speakers barely play that range. Choose Light, use headphones, or play it through a speaker with a woofer.

### Can I save brown noise to play offline?

Yes. Save a 30- or 60-second WAV loop at 44.1 or 48 kHz in the depth and channel setting you have selected, then play it on repeat. A 60-second stereo file at 48 kHz is about 11.5 MB.