🛌

Sleep Environment Noise Checker

This sleep environment noise checker lets you hold your phone or laptop where your head rests at night and measure your bedroom’s noise level from the microphone in real time. The tool compares it against published WHO night-noise guidance, tells you whether the noise is low-frequency rumble (traffic, machinery) or mid/high-frequency sound (neighbours, appliances, fans), and suggests a masking approach.

ℹ This is an uncalibrated estimate, not a certified sound-level meter. A browser/phone mic isn’t calibrated, rolls off at the frequency extremes, and generally cannot capture true infrasound or deep sub-bass. Until you calibrate it against a real meter the level is shown as dBFS (relative, ≤ 0); the dB SPL figure is only an estimate. It is not a substitute for a Type 1/2 meter and is not valid as legal, complaint or compliance evidence. The most trustworthy outputs here are the noise character (low vs high frequency) and before/after comparisons with the same mic and setup — those don’t depend on calibration. Nothing is recorded or uploaded.

Microphone

Input device

Idle — press Start. Make the room as quiet as it is when you sleep.

SPL calibration (shared across noise tools)

Reading on a real meter right now (dB SPL)

Not calibrated — showing relative dBFS. To estimate dB SPL, read the level on a calibrated sound-level meter or app, type it above while measuring, and set it. The offset is stored locally and reused by every noise tool.

Night-noise level

—dBFS

Press Start to measure.

WHO Night Noise Guidelines for Europe (2009) suggest an annual L_{night,outside} of 40 dB(A) and a long-term target of 30 dB(A). The bands below mirror common indoor sleep guidance; they apply only once you have calibrated for dB SPL. Note that the tool measures an unweighted broadband level, whereas the WHO figures are dB(A), so a rumble-heavy bedroom reads higher unweighted than its true dB(A) — treat the comparison as a rough guide.

Noise character

Start measuring to identify the dominant noise type and a masking suggestion.

Current

—

Average (Leq)

—

Background (L90)

—

Loudest (Lmax)

—

Quietest (Lmin)

—

Elapsed

0:00

Masking suggestion will appear here once measuring. As a rule of thumb: steady broadband masking from a white, pink or brown noise source can cover intermittent disturbances by reducing the contrast between quiet and loud moments — brown/pink suit low rumble, white/pink suit higher chatter.

How It Works

When you press Start, the tool requests that your microphone’s automatic gain, noise suppression and echo cancellation be switched off (these would distort any noise measurement); if your OS or browser forces them on anyway, treat the reading as unreliable. It then continuously reads the raw waveform. From each short frame it computes the RMS energy and converts it to dBFS — decibels relative to digital full scale, where 0 is the loudest the system can capture and quieter sounds are negative. dBFS is genuinely meaningful as a relative level: it lets you compare a window open versus closed, a fan on versus off, or tonight versus last night with the same device and placement, giving you a reliable picture of your sleep environment without any calibration. For in-depth logging over hours rather than minutes, pair this with the environmental noise monitor, which records a time-stamped level chart using the same shared calibration.

A browser cannot know your microphone’s real-world sensitivity, so it cannot turn dBFS into a true dB SPL on its own. That is what the calibration control is for. While the tool is measuring, read the level on a calibrated sound-level meter or a reputable calibrated phone app, type that number in, and press Set calibration. The tool stores the offset (known SPL − current dBFS) in your browser under the shared key fd-noise-cal, so every noise tool on the site — this one, the environmental monitor, the noise pollution meter — uses the same calibration. Calibrate once and your dB SPL estimates carry over. The offset never leaves your device.

Alongside the level, an FFT splits the sound into low (20–200 Hz), mid (200 Hz–2 kHz) and high (2–16 kHz) energy. Low-frequency dominance points to traffic rumble, HVAC or machinery — hard to block, best tackled with brown or pink masking noise and sealing gaps around windows or doors. Mid/high dominance points to voices, TVs, fans and appliances — easier to mask with white noise or pink noise. Because consumer mics roll off below roughly 50–100 Hz, the very deepest rumble and any true infrasound (<20 Hz) will read low or be missed entirely; the spectral shape is still a reliable guide to what kind of sleep noise you are dealing with. The L90 statistic gives a useful estimate of the steady noise floor — the background level that persists even when intermittent events are ignored. If you specifically suspect sub-bass hum from a boiler, transformer or distant motorway, the low-frequency noise detector is better suited to that narrower hunt. The level series also yields Leq (the equivalent continuous level, energy-averaged), L90 (the level exceeded 90% of the time — a good proxy for the steady background), and Lmax/Lmin. These are valid relative metrics whether or not you have calibrated.

The verdict bands echo widely cited sleep noise guidance: the WHO Night Noise Guidelines for Europe (2009) recommend an annual night-outside level (L_{night,outside}) of 40 dB(A) with a long-term target of 30 dB(A), below which there are no substantiated effects on sleep. Research links chronic nocturnal noise above these thresholds to sleep fragmentation, elevated cortisol, and cardiovascular risk over time. As an indoor rule of thumb the tool treats roughly <30 as ideal, 30–40 as acceptable, and >40 as likely disruptive. Those WHO numbers are real published values in dB(A), but the tool computes an unweighted (Z-weighted) broadband level — it applies no A-weighting filter — so its calibrated figure is a dB SPL unweighted estimate, not dB(A). Because A-weighting cuts the low end heavily, a bedroom dominated by traffic or HVAC rumble will read higher unweighted than its true dB(A); treat the verdict as a rough guide. The comparison is for guidance only and is only meaningful once you have calibrated for dB SPL — before that the tool shows the relative dBFS level and withholds the verdict rather than inventing a number.

Frequently Asked Questions

Is this an accurate sound-level meter for my bedroom?

No. It is an uncalibrated estimate from a consumer microphone, not a certified Type 1/2 sound-level meter, and it must not be used as legal, complaint or compliance evidence. A browser cannot know your mic’s sensitivity, and phone mics roll off at the frequency extremes. The reading is most useful as a relative level — comparing tonight to last night, or window-open to window-closed with the same device — and for telling low-frequency rumble apart from higher-frequency sound. For a real dB(A) figure, use a calibrated meter.

What are the WHO night-noise numbers you compare against?

The WHO Night Noise Guidelines for Europe (2009) recommend an annual average outdoor night level (Lnight,outside) of 40 dB(A), with a long-term target of 30 dB(A) below which no substantiated sleep effects are expected; levels above 40 dB(A) are associated with adverse effects. This tool uses a simple indoor rule of thumb of under 30 ideal, 30 to 40 acceptable, and over 40 likely disruptive. Important: the WHO figures are A-weighted dB(A), but this tool computes an unweighted (Z-weighted) broadband level — it applies no A-weighting — so its calibrated figure is a dB SPL unweighted estimate, not dB(A). A rumble-heavy bedroom therefore reads higher unweighted than its true dB(A). These are real published values, but our comparison is approximate guidance only, not an official assessment, and it applies only after you calibrate for dB SPL.

Why do I have to calibrate, and what does the shared calibration do?

Until you calibrate, the tool honestly shows dBFS (a relative level ≤ 0), not dB SPL, because a browser cannot know your mic’s sensitivity. To estimate dB SPL, measure with this tool while reading the level on a calibrated meter or app, type that number in and set it; the tool stores the offset (known SPL minus current dBFS) in your browser under the shared key fd-noise-cal. Every noise tool on the site reuses that one offset, so you only calibrate once. The offset stays on your device and is never uploaded.

How does it tell rumble from neighbour noise, and which masking sound should I use?

It splits the spectrum into low (20 to 200 Hz), mid (200 Hz to 2 kHz) and high (2 to 16 kHz) energy. Low-frequency dominance suggests traffic, HVAC or machinery rumble; mid/high dominance suggests voices, TVs, fans or appliances. For low rumble, brown or pink noise masks best; for higher-frequency chatter, white or pink noise works well. The spectral shape is calibration-independent, so it is reliable even before you set an SPL offset. Note that consumer mics roll off in the deep bass and cannot capture true infrasound below 20 Hz.

What are Leq, L90, Lmax and Lmin?

Leq is the equivalent continuous level — the steady level carrying the same total energy as the varying sound since you started, computed as 10 log10 of the mean of 10^(L/10). L90 is the level exceeded 90% of the time, a good proxy for the steady background once short loud events are ignored. Lmax and Lmin are the loudest and quietest readings over the session. They are all valid as relative metrics in dBFS and become dB SPL estimates once you calibrate.

Is my audio recorded or uploaded?

No. The microphone signal is analysed live in your browser to compute the level and spectrum, and is never recorded, saved or transmitted. The calibration offset is the only thing stored, and it stays in your browser’s local storage on this device. The microphone is released when you press Stop or close the tab.

Can I use this to measure traffic or street noise coming through my bedroom window?

Yes — place your phone or laptop near the window (or in the middle of the room for typical sleeping-position noise) and run a session while traffic is at its usual level. The L90 statistic is most informative for traffic assessment, as it captures the steady background floor rather than peak events. Bear in mind that the tool cannot capture very low-frequency tyre rumble below about 50–100 Hz, and the reading is relative (dBFS) unless you first calibrate with a real meter.

What level of bedroom noise is considered too loud for sleep?

The WHO (2009) places the threshold for substantiated sleep effects at 40 dB(A) outdoors at night, with a long-term ideal of 30 dB(A). Common bedroom guidelines suggest keeping the indoor equivalent below about 30–35 dB(A). Intermittent events — trucks, dogs, snoring partners — are more disruptive than steady sounds at the same average level, because they cause cortical arousals (brief wake-ups) even if you don’t remember them in the morning. This tool’s Lmax statistic highlights those peak events.

Does white noise actually help with sleep, and at what volume?

Masking noise works by reducing the contrast between your quiet baseline and sudden loud events, not by being louder than the disturbance. Studies suggest steady broadband masking is effective when it sits roughly 6–10 dB above the background floor (L90) but stays below about 50 dB(A) at the ear to avoid its own sleep disruption. Pink and brown noise, which have more energy in the low frequencies, are often preferred for traffic rumble; white noise is typically better at masking higher-frequency chatter. Keep any masking source well below 50 dB to avoid hearing risk during the night.

Related Tools

← All Noise Analysis Tools