Online Decibel Meter

A decibel meter — also called a sound level meter, noise meter, dB meter, or sound meter — is an instrument that measures how intense sound is at a specific point and reports it in decibels (dB). Professional acousticians carry dedicated hardware that meets the IEC 61672 standard. An online decibel meter brings the same core principle to your web browser: it uses your phone, tablet, or laptop microphone to capture live sound, analyses it in real time, and shows a continuously updating dB reading. No app to install, no account, no hardware to buy — just press Start Measuring and you have a working sound level meter online.

This page is that instrument. The tool at the top is a free noise level meter with A-weighting (dBA), C-weighting (dBC), Z-weighting (dBZ), Fast / Slow / Impulse time response, live waveform and audio spectrum analyzer, plus minimum, average, maximum, peak, and Leq statistics you can export as CSV or JSON. The guide below explains what every number means, how to measure correctly, when a reading is trustworthy, and when you should call in a certified professional instead. If you are in a hurry: allow microphone access, hold the device where people actually listen, wait 30–60 seconds, and quote the average or Leq — not a single flickering instant.

Online meter vs. dedicated hardware: what is the difference?

Both do the same three things: a microphone converts air-pressure fluctuations into an electrical signal, electronics average that signal over a standardized time and frequency filter, and a display shows the result in decibels. The difference is the microphone and the calibration chain. A Class 1 or Class 2 handheld meter uses a laboratory-measured capsule with a known sensitivity, a pistonphone calibrator that produces exactly 94 dB at 1 kHz, and documented corrections for temperature, humidity, and reflections. Your phone uses a tiny MEMS microphone optimised for voice calls, with automatic gain control, noise suppression, and echo cancellation that help calls but alter measurements.

In practice that means: an online dB meter is excellent for everyday questions — Is this bedroom quiet enough to sleep? Which appliance is loudest? Is this concert harming my ears? Did my new curtains actually help? — especially for before-and-after comparisons on the same device. It is not a legal instrument. For workplace compliance, planning disputes, product specifications, or any number that will be challenged in writing, you need a certified meter operated by a competent person. Our accuracy guide, methodology, and validation protocol document exactly where the line sits and how we stay on the honest side of it.

Who uses an online sound meter?

Parents check nurseries and white-noise machines. Renters document thin walls and late-night bass before talking to a landlord. Teachers check whether a classroom hum is drowning out the back row. Office managers hunt down a rattling air conditioner. Musicians and podcasters balance speakers and rehearsal rooms. Drivers compare road, wind, and engine noise at different speeds — often alongside a GPS speedometer to see how cabin noise climbs as speed increases. Travellers test a hotel room in 60 seconds on arrival. For a second opinion on a second device, readers also open our sister project Real Decibel Meter — a free online decibel meter and compare 60-second Leq averages side by side. None of these need laboratory precision; all of them need a fast, honest number with enough context to act on. That is what this tool is built for.

What Do Decibels (dB) Actually Mean?

The decibel is a logarithmic unit that expresses the ratio of a measured sound pressure to a reference pressure — 20 micropascals, roughly the faintest sound a young healthy ear can detect. That is why 0 dB does not mean silence; it means the threshold of hearing. Because human hearing spans an astonishing range — from a mosquito wingbeat to a jet engine, a ratio of more than a trillion to one — the logarithmic scale compresses that range into manageable numbers from about 0 to 140 dB.

Three rules make the scale intuitive. +3 dB doubles sound energy. Two identical 70 dB sources together produce about 73 dB, not 140 dB. +6 dB doubles sound pressure and roughly corresponds to doubling distance outdoors — 76 dB at one metre falls to about 70 dB at two metres. +10 dB sounds roughly twice as loud and represents ten times the energy. Going from a 65 dB office to a 75 dB open-plan floor is therefore not a small step; it is a huge leap in energy and explains why you leave exhausted. For a full beginner walkthrough with worked examples, see What Is a Decibel? and the Sound Level Guide.

How Does an Online Decibel Meter Work?

When you press Start Measuring, your browser asks for microphone permission — a privacy safeguard built into every modern browser. Once granted, audio flows through the Web Audio API entirely on your device. The waveform is sliced into short windows (tens of milliseconds), each window's energy is computed, the selected frequency weighting is applied band by band, the chosen time weighting smooths the result, and the value is converted to decibels against a calibration offset and shown live. The live meter displays the current level plus a level bar, while the statistics panel tracks minimum, average, maximum, peak, and Leq. Switch to the Spectrum tab to run a real-time frequency analyzer and sound frequency analyzer view that shows whether your noise is low rumble, mid-range speech, or high hiss.

Nothing is uploaded, recorded, or sent anywhere. You can verify this by turning on airplane mode after the page loads (on most devices measurement continues) or by inspecting network traffic — there is none during measurement. Read the full details in our privacy policy and measurement methodology. Because phone microphones differ in sensitivity by several decibels, uncalibrated readings are estimates. Measuring against a known 94 dB reference tone and storing the offset with the Calibrate button turns the estimate into a consistent personal instrument. See the calibration guide for the five-minute method.

How to Measure Sound With Your Device Microphone: Step by Step

1. Allow the microphone and get settled. Click Start Measuring, grant permission once, and place the device where people actually listen — ear height at a desk, bedside table, or seat — with the microphone pinhole uncovered and facing the source.
2. Silence yourself. Do not talk, type, or shuffle during the measurement. Your own voice at arm's length adds 10–20 dB and ruins a background-noise reading.
3. Wait for the reading to settle. Watch the live dB value for at least 30–60 seconds. Single flickering instants are meaningless; the average and Leq over a full minute are quotable.
4. Note the context. Write down place, distance to the source, weighting (dBA is default), and duration — for example, “62 dBA Leq, 60 s, 2 m from TV”. Without distance and duration, comparisons are void.
5. Check the spectrum if bass is the problem. Open the frequency analyzer while measuring. Neighbour bass that reads modest in dBA but strong in dBC is real — note both values.
6. Export and compare. Save CSV or JSON, repeat from a second spot, and compare averages. If two positions agree within a few dB, you can trust the picture. Log repeated sessions in the Noise Journal.

For the most representative results, measure at the spot where people sit or sleep, keep phone cases and fingers away from the mic opening, disable battery-saver modes that throttle the browser, and keep the screen on — mobile Safari pauses background tabs quickly. Outdoors, shield the mic from wind with your hand held nearby (without touching) or a foam cover; wind artefacts can add 10 dB or more. Our step-by-step measuring guide and phone and PC guide cover device-specific tricks for iPhone, Android, and laptops.

Understanding Every Reading: Min, Avg, Max, Peak, Leq, RMS

The current value is the live level right now. Minimum marks the quietest moment captured — useful for background noise when no events occur. Maximum marks the loudest event — a door slam, horn, or bass hit. Average reflects the typical level over the session. Leq (equivalent continuous sound level) is the energy average over the whole session and is the metric professionals quote for exposure, because it correctly weights loud moments: one hour at 70 dB plus one hour at 50 dB is not 60 dB but about 67 dB Leq — the loud hour dominates. Peak catches the sharpest instantaneous pressure, which matters for impact sounds the average would swallow. RMS (root mean square) tracks effective loudness over a short window and follows perceived volume better than raw peaks.

Letters matter as much as numbers. dBA (A-weighting) filters the signal to mirror human hearing — damping deep bass and extreme treble — and is the standard for health, sleep, and neighbourhood questions. dBC (C-weighting) stays much flatter and preserves bass, so it is the right choice for concerts, engines, machinery, and neighbour subwoofers. dBZ (Z-weighting) applies no correction at all and shows the raw flat signal for acoustic data. See dB vs dBA, dBA meter, dBC meter, and Leq meter for dedicated tools and deeper comparisons.

Time response matters too. Fast (125 ms) follows fluctuations closely and suits most everyday checks. Slow (1 s) smooths the display and suits steady background noise. Impulse grabs brief transients like claps or impacts. If the maximum is much higher than the average, you are hearing occasional loud events rather than steady noise — both numbers together tell the story.

Why Microphone Permission Is Required — and What Happens to Your Audio

Browsers are required to ask before any website can access your microphone — this is a privacy safeguard built into every modern browser, not a choice we made. The permission prompt appears because this decibel meter reads live audio from your device to compute loudness. All processing happens locally via the Web Audio API: your microphone audio never leaves your device, nothing is uploaded, recorded, or listened to, and no network request is involved in measurement. Revoking permission instantly stops measurement. You can read the full details in our privacy policy and verify the claim by measuring with Wi-Fi off after the page has loaded.

How Accurate Is a Microphone Decibel Meter?

We do not make unrealistic claims. A consumer microphone — especially the small MEMS capsules inside phones and laptops — is not a professional measurement instrument. Uncalibrated, two phones measuring the same sound commonly disagree by 5–10 dB because of different capsule sensitivity, frequency response, and hidden processing: automatic gain control (AGC) that rides levels up and down, noise suppression that erases steady hum, and echo cancellation that fights continuous tones. Laptop fans alone add 35–45 dB of background. Room acoustics, mic direction, phone cases covering the pinhole (up to −10 dB error), and wind outdoors all shift readings further.

That sounds discouraging, but it misses the point: used carefully, this tool gives consistent, useful estimates for reference, education, and before-and-after comparisons — especially on the same device after calibration. Absolute accuracy without calibration is unknown; relative accuracy on one device over 30–60 second Leq windows is genuinely good. The decibel meter accuracy guide quantifies every factor, and the validation protocol shows how we test the signal chain so the math itself is defensible.

How to Improve Accuracy: Calibration and Technique

• Follow the step-by-step measuring guide for a repeatable workflow — same place, same distance, same duration.
• Use the Calibrate button with a known reference: play a 94 dB calibration tone from a trusted source or compare against a trusted second meter at the same position, note both values over 30–60 seconds, and store the offset. See the calibration guide.
• Disable automatic gain control, noise suppression, and echo cancellation if your device or browser exposes them; on laptops prefer an external USB measurement microphone for quiet rooms.
• Measure iPhone vs Android fairly: iPhones resemble each other more closely, Android devices scatter widely by maker and case — always note which device you used and never compare single instants across devices.
• Take several measurements from different positions, quote Leq or average rather than peaks, and export CSV/JSON so outliers are visible instead of remembered.

Chrome and Edge currently give the most stable Web Audio chain; Safari on iPhone and iPad works well but asks permission more strictly and pauses background tabs sooner, so keep the screen on; Firefox measures reliably with permissions tucked behind the padlock icon. Keep browsers updated, use a single microphone tab, and reload after changing permissions. For professional or regulatory work — workplace exposure assessments, planning evidence, product specs — use a certified sound level meter that meets IEC 61672 and have measurements done by qualified professionals. Bring your online readings along anyway; they make the professional survey faster because you already know where and when it gets loud.

What to Read Next

New to sound measurement? Start with the Sound Level Guide and dB vs dBA, learn the measuring workflow, check the noise exposure guide and exposure calculator, explore all tools, browse the Learn hub and blog, or get answers in the FAQ — the full site index is in the footer below. Learn how the numbers are produced in Methodology, who stands behind the site on About and Editorial Policy, and how to reach a human via Contact. Prefer German? Try the Dezibelmesser — Dezibel messen ohne App.

About this guide: maintained by the DecibelMeter editorial team, last reviewed for NIOSH and WHO reference values. For corrections or device reports, contact us — real measurement reports improve the guides for everyone.

Adding sound sources: decibels never add directly

The single most useful rule: two equal sources add +3 dB, not double the number. Two identical 70 dB fans side by side produce about 73 dB together. Ten identical fans produce 80 dB — ten times the energy sounds only twice as loud. And when one source is 10 dB or more louder than everything else, the total is essentially the louder source alone: an 80 dB air conditioner plus a 65 dB refrigerator still reads about 80 dB, because the quieter machine contributes less than half a decibel.

This explains three everyday puzzles. First, why one loud appliance dominates. Measure your kitchen at 71 dBA, switch off the range hood, and it drops to 58 dBA — the hood was the entire problem, and the refrigerator you suspected was innocent. Fix sources in order of loudness; silencing the second-loudest thing first changes almost nothing. Second, why traffic plus neighbours equals whichever is louder. If road noise at your window is 62 dBA and the neighbour's music through the wall is 55 dBA, closing the window (cutting traffic to 50 dBA) reveals the music at 55 dBA — you did not create a new problem, you unmasked one. Third, why "just a bit louder" exhausts you. A 68 dBA open-plan office carries roughly two and a half times the sound energy of a 64 dBA one; small numbers, large fatigue.

The distance rule: 6 dB per doubling outdoors, less indoors

Outdoors, away from walls, sound from a small source spreads over an ever-larger sphere, so every doubling of distance drops the level about 6 dB. A mower reading 84 dBA at 2 metres reads roughly 78 at 4 metres, 72 at 8 metres, and 66 at 16 metres. Measure once at a known distance and you can estimate any other distance without walking there: note "78 dBA at 4 m" and the neighbour's fence at 16 m gets about 66 dBA.

Indoors the rule weakens because walls reflect sound back — expect more like 3–4 dB per doubling in a furnished room and almost no drop at all in a small tiled bathroom, where reverberation dominates. That is why a TV sounds nearly as loud from the sofa as from the doorway, and why soft furnishings (rugs, curtains, bookshelves) genuinely lower levels: they remove reflections, not just echoes. When you document a reading, always write the distance — "66 dBA at 1 m" and "66 dBA at 4 m" describe completely different situations, and only the first tells anyone what the source itself produces.

Perceived loudness: the 10 dB step

Energy math (+3 dB = double energy) and human perception diverge: a sound must rise about 10 dB to feel twice as loud. A 50 dBA suburb feels roughly twice as loud as a 40 dBA one, though it carries ten times the energy. Use this when setting goals. Dropping a bedroom from 48 to 45 dBA is a real energy cut (half the energy) but barely noticeable; dropping it from 48 to 38 dBA transforms sleep. Aim reduction projects at 8–10 dB improvements — new seals plus a rug, not seals alone — and verify with before-and-after Leq in the Noise Journal. Small single fixes disappoint; stacked fixes delight.

A weighting is a frequency filter applied before the meter computes decibels. A-weighting copies the human ear at moderate levels: it heavily discounts deep bass and extreme treble and passes speech frequencies nearly untouched. C-weighting stays almost flat, keeping bass, and matches how we perceive very loud sounds. Z-weighting ("zero") applies no filter at all and reports raw physical energy. The table below shows the standard corrections each weighting applies at key frequencies — negative means that frequency counts less toward the final number.

Now the practical consequences. Use dBA for sleep, offices, classrooms, neighbourhood complaints, headphone safety, and anything compared against health guidance — nearly every regulation speaks dBA. Use dBC when bass dominates: concerts, engines, generators, neighbour subwoofers, wind turbines. The classic trap is a neighbour's bass that reads a modest 52 dBA but 71 dBC — quoted alone, the dBA number makes your complaint look frivolous; quoted together, the 19 dB gap proves a low-frequency problem the A-filter hid. Use dBZ when you want the unvarnished physics, for example comparing two rooms' raw energy or checking a spectrum peak against a linear scale.

Two cautions. First, never compare a dBA number with a dBC number — a "quieter" 60 dBC room can easily be louder-feeling than a 65 dBA one if the energy sits in speech frequencies. Always state the letter. Second, phone microphones under-read deep bass even in dBC or dBZ mode, because the capsule itself rolls off below ~100 Hz; treat low-frequency phone readings as minimums. The dBA meter and dBC meter tools lock the weighting for you, and dBA vs dBC: which to use walks through concert, sleep, and traffic scenarios.

Occupational limits look confusing because each authority expresses the same physics differently. The two dials are the criterion level (the level allowed for 8 hours) and the exchange rate (how much extra level halves the allowed time). A 3 dB exchange rate reflects equal energy and is the more protective, scientifically grounded choice; a 5 dB rate is a historical compromise. For personal decisions, use the strictest widely respected reference — NIOSH — and treat anything stricter as a bonus.

Translate this into daily life with the noise exposure calculator: enter your measured Leq and duration to get a dose percentage against the NIOSH budget. A concrete example — 2 hours of rehearsal at 94 dBA uses the full daily budget (94 dBA allows about 1 hour, so 2 hours is roughly 200%), which is why musicians who rehearse unprotected lose high frequencies first and notice it as "muffled cymbals" years later. Office workers face the opposite trap: 70 dBA feels safe — and is, for hearing — but studies consistently link sustained 65 dBA+ office noise to fatigue and errors, a performance problem rather than a hearing one. The noise exposure guide and safe limits explainer cover dose math country by country, and the 3 dB rule makes the halving logic visual.

Impulse noise — gunshots, fireworks, hammer strikes, balloon pops near children — plays by peak rules, not averages: peaks above 135–140 dBC can injure instantly no matter how short. Never judge fireworks or firearms by Leq or average; watch the peak readout, keep real distance, and wear rated protection. For children, apply every limit earlier: smaller ear canals receive higher doses from the same sound, so cap toys and tablets conservatively and prefer over-ear limits around 75–85 dB maximum.

Work one room at a time with the same routine: silence everything, record a 60-second baseline Leq, then switch devices on one by one and note how much each adds. The single biggest contributor is your project — everything else waits. Parents should pay special attention to the nursery row: paediatric guidance keeps infant sound exposure, including white-noise machines, at or under 50 dB at the crib, with the machine as far from the mattress as the room allows and on the lowest effective setting overnight. A 65 dB "shusher" at 30 cm is not soothing; it spends a baby's daily budget before breakfast.

Renters get the most leverage from reversible fixes. Measure the bedroom three nights — windows closed, windows ajar, windows open — and you will learn whether the pane or the frame leaks. Temporary window-seal tape, a draught-stopper under the door, a thick rug on bare floors, and bookshelves against the noisy wall each buy 2–5 dB; stacked, they buy sleep. Document before-and-after Leq pairs so the landlord sees numbers, not adjectives — the full playbook from free fixes to renovation is in how to measure and reduce home noise, with neighbour-specific tactics in how loud is too loud for neighbours.

1. Your finger or case covers the microphone. The classic: a bulky case or your grip sits over the mic pinhole and everything reads 10–20 dB low. Fix: uncover the pinhole, remove thick cases, and hold the phone by its edges with the mic facing the source.
2. Automatic gain or noise suppression is active. Some phones and browsers ignore the page's request to disable processing, so steady hums slowly fade from the reading while speech pumps it up and down. Fix: check the Device and Microphone panel, try another browser (Chrome or Edge first), and distrust any reading that drifts on a perfectly steady sound.
3. You are measuring yourself. Typing adds ~50 dBA at the keyboard, breathing near the mic adds low-frequency energy, and a held phone picks up handling thumps. Fix: set the phone down, step back a metre, stay silent for the full 60 seconds.
4. Wind outdoors. Even a light breeze across an exposed mic capsule creates turbulence that reads as 60–75 dB of "noise" that no ear would report. Fix: shield with your body, measure from a doorway, or cover the mic with foam — never measure gusts and quote them as place levels.
5. Wrong weighting for the question. Bass-heavy problems read deceptively low in dBA (see the −26 dB correction at 63 Hz above). Fix: when bass, engines, or subwoofers are the complaint, quote dBC alongside dBA and say both numbers every time.
6. Too close to a wall or corner. A mic within centimetres of a hard surface picks up pressure build-up that can add 3–6 dB versus the middle of the room. Fix: keep 30 cm+ from walls for room readings (unless you are deliberately logging neighbour transmission at the party wall — then say so).
7. Quoting an instant instead of an average. Live values flicker with every syllable and footstep; a screenshot of "82 dB!" during a door slam describes the slam, not the room. Fix: quote 30–60 second Leq or average, plus minimum for background and maximum for events.
8. Background contamination you stopped hearing. Your brain filters the fridge, the HVAC, the distant road; the meter does not. Fix: measure with suspects switched off one at a time — the drop tells you exactly what each contributed.
9. Battery saver or background tab throttling. Power-saving modes and backgrounded mobile tabs starve the audio pipeline, freezing or flattening the display. Fix: keep the tab foreground, screen awake, charger handy, one microphone tab only.
10. Expecting the wrong number. "Quiet" rooms with a humming PC easily sit at 45 dBA; libraries with HVAC run 40+. Fix: sanity-check against the common levels chart and everyday examples before concluding the meter lies — a 62 dB library usually means a blocked mic or a forgotten fan, both fixable in seconds.

Still suspicious after all ten? Calibrate against a known reference with the Calibrate button (see the calibration guide), then cross-check with a second device running our sister meter side by side for 60 seconds. Two independent setups agreeing within a few dB is trustworthy; two disagreeing setups mean at least one has a setup problem from this list — find it before quoting either number. Device-specific quirks for iPhone, Android, and laptops live in how to use an online meter and accuracy.

Every log entry needs seven fields — miss one and the entry is arguable. When (date, start–end time), where (room and mic position, e.g. "bedroom, pillow height, 30 cm from party wall"), device (phone model and browser, plus calibration offset if set), settings (dBA or dBC, Fast/Slow), duration (Leq needs a stated window — "Leq 15 min"), numbers (Leq, minimum, maximum), and conditions (windows open/closed, appliances on, weather, what the source was doing). One line per session, same format every time. Two weeks of this beats any single dramatic screenshot.

Example entry: "Oct 14, 22:10–22:25, bedroom pillow position, Pixel 8 / Chrome, dBA Slow, Leq 15 min 54 dBA, min 41, max 71 (three bass hits), windows closed, neighbour music with heavy bass, dry night." Pair it with the matching dBC Leq (say 68 dBC) and the 14 dB gap tells the story before you say a word. Export each session as CSV or JSON from the tool and keep the files — filenames with dates sort themselves. Log repeats in the Noise Journal so the pattern across nights is visible at a glance.

Stay honest about limits — it makes you more credible, not less. Write "phone estimate, uncalibrated, for screening only" on shared logs, never claim legal-grade precision, and never edit numbers to look worse; anyone who checks a doctored log dismisses the whole case. Present the log as a briefing for the professional or mediator ("here is when it peaks, here is where to place certified kit"), not as a verdict. For neighbour situations specifically, apartment limits explained covers day/night thresholds and the logging routine mediators respect, and why readings jump explains which statistic to quote for fluctuating noise.