How to use the decibel meter
- Press Start. Your browser asks for permission to use the microphone. The meter cannot run without it.
- Choose Allow. The big number starts updating right away. That number is the current sound level in decibels.
- Hold the device still with its microphone pointed toward the sound and nothing covering it. On most phones the main microphone is on the bottom edge.
- Measure for at least 30 seconds. A single glance at a jumping number tells you little. The average (Leq) settles as the meter collects more sound.
- Press Stop to freeze the readings. You can export the log as a CSV file or save the graph as a PNG image.
If the meter stays near its lowest value, check that the right microphone is selected in your browser's site settings and that no other app is holding it.
What the numbers mean
A sound level meter never shows just one number, because sound changes from moment to moment. Here is what each readout on this meter tells you.
| Readout | What it is | Use it for |
|---|---|---|
| Current level | The time-weighted sound level right now | Watching a sound as it happens |
| Min | The lowest time-weighted level since Start | The background level of a room |
| Avg (Leq) | The energy average of everything since Start | Comparing with exposure guidance, describing a noisy period with one figure |
| Max | The highest time-weighted level since Start | The loudest event, such as a door slam or a passing truck |
| Peak | The highest instantaneous sample, with no time smoothing | Very short impacts that Max smooths over |
Leq is not a simple average of the numbers on screen. Decibels are logarithmic, so the meter averages sound energy and converts the result back to decibels. Loud moments count for much more than quiet ones: one minute at 90 dB followed by nine minutes at 60 dB gives an Leq of about 80 dB, not 63. Our guide to the decibel explains why.
A, C and Z weighting
Human ears are less sensitive to low and very high pitches than to the middle range. Frequency weighting adjusts the measurement to account for that.
- A-weighting (dBA) cuts low frequencies heavily, about 16 dB at 125 Hz and 26 dB at 63 Hz. It is the default here and the weighting used by NIOSH, OSHA and most noise rules.
- C-weighting (dBC) is nearly flat through the bass range. Use it for music with heavy bass, engines and subwoofers. If dBC reads far higher than dBA, the sound has a lot of low-frequency energy.
- Z-weighting (dBZ) applies no frequency filter at all.
The A and C filters are built from the analog filter definitions in IEC 61672, the international sound level meter standard. That describes how we do the math. It does not make a phone or laptop a certified instrument.
Fast and Slow
Time weighting controls how quickly the current level reacts. Fast uses a 125 millisecond time constant and follows speech and short events. Slow uses 1 second and gives a steadier number that is easier to read; OSHA's workplace limits are written in terms of slow response. Min and Max follow whichever setting you pick. Leq and Peak do not depend on it.
Typical decibel levels
Use these published reference points to sanity-check your reading. The full, sortable decibel chart lists more than 80 sounds with a source for each.
| Sound | Typical level | Source |
|---|---|---|
| Whisper | 30 dBA | ASHA |
| Quiet room | 40 dBA | ASHA |
| Normal conversation | 60–70 dBA | NIDCD |
| Vacuum cleaner | 70 dBA | ASHA |
| Gas lawn mower, subway, motorcycle | 91 dBA | ASHA |
| Concerts, sporting events, headphones at full volume | 94–110 dBA | NIDCD |
| Sirens | 110–129 dBA | NIDCD |
| Fireworks shows | 140–160 dBA | NIDCD |
According to NIDCD, sounds at or below 70 dBA are unlikely to cause hearing loss even after long exposure, while long or repeated exposure at or above 85 dBA can. See safe decibel levels for exposure times.
How accurate is a phone or laptop microphone?
Good enough to tell a 50 dB room from a 70 dB room. Not good enough to replace a calibrated sound level meter. Three things limit it:
- The microphone is built for voices. Device makers tune microphones and their electronics for calls, not for measurement. Sensitivity varies between models, so the same sound can read differently on two devices until each is calibrated.
- There is a floor and a ceiling. Every microphone has its own electrical noise, so very quiet rooms read higher than they are. At very high levels the signal clips and the reading stops rising.
- Browsers process audio. We ask the browser to switch off echo cancellation, noise suppression and automatic gain control, which would otherwise distort levels. Some devices do not honor every request.
For context, when NIOSH researchers tested smartphone sound apps against a laboratory reference with noise from 65 to 95 dB, only four of the ten iOS apps they selected stayed within ±2 dB, and Android results varied too much between devices to draw conclusions. NIOSH's own iOS app was later validated to ±2 dBA on that fixed hardware. A website runs on thousands of device and browser combinations, so we make no such claim.
What the meter does well is relative measurement: is it louder now than an hour ago, does closing the window help, which appliance is noisiest. To improve absolute readings, set the calibration offset against a reference meter. The steps are on the calibration page, our test notes are on the accuracy page, and the math is described in the methodology.
Tips for a good measurement
- Measure where the listener is. Sound falls off with distance, so a reading next to a speaker says little about the level across the room.
- Keep the microphone clear. Remove thick cases, and keep fingers and fabric off the microphone hole.
- Avoid wind and handling noise. Wind across the microphone produces large false readings. Outdoors, shield the device; indoors, rest it on something soft and step back.
- Stay quiet yourself. Your breathing and movement are part of the measurement at close range.
- Let it run. NIOSH's guidance for area measurements is to average for at least 30 seconds at each spot. For traffic or a neighbor's music, several minutes is better.
- Quote Leq and Max, not the flickering number. "Leq 62 dBA over 10 minutes, max 78 dBA" is a result someone else can compare.
- Keep the same setup when you compare readings: same device, same spot, same weighting.
What people use it for
Neighbor and street noise
Run the meter at the affected spot with the windows closed, then open, and note the time, Leq and Max. The 5-minute graph shows how often the noise occurs. Local noise rules differ and enforcement relies on officials' own measurements, so treat your log as a diary, not as evidence.
Classrooms
Set the threshold alarm to a level you consider too loud and put the meter in fullscreen on the projector. The exceedance counter shows how many times the class went over. For a display designed for students, use the classroom noise meter.
Workplaces and workshops
A quick screening tells you whether a task is closer to 75 or 95 dBA. NIOSH offers a useful rule of thumb as well: if you must raise your voice to be heard by someone 3 feet away, the noise is probably around 85 dBA or more. If your screening comes out near that level, ask for a proper noise assessment.
Home theater and music practice
Switch to C-weighting to match speaker levels with test tones, and watch Leq during a rehearsal to see what your ears are getting over the whole session.
Baby rooms and sleep
Check how loud a white-noise machine is at the crib instead of at the machine, and see what a nursery or bedroom sits at overnight. The US EPA identified an indoor average of 45 dB as the level that prevents interference with activities such as sleep and conversation.
Concerts, bars and sports events
NIDCD puts concerts and sporting events at 94 to 110 dBA. Phone microphones often clip at these levels, so a reading that stops climbing probably understates the truth. If the meter shows the high 90s or more, use earplugs.
Privacy: your audio stays on your device
The meter is built on the Web Audio API, a standard part of modern browsers. Sound from the microphone is turned into level values in your device's memory and the raw audio is discarded immediately.
- Nothing is recorded. There is no audio file, on your device or anywhere else.
- Nothing is uploaded. Audio and readings are never sent to us.
- Settings stay local. Your calibration offset is saved in your own browser's storage, so it applies to that browser on that device only.
- Exports are yours. CSV and PNG files are generated in the browser and saved straight to your device.
You can revoke microphone permission at any time in your browser's site settings. If you would like an installed app instead of a web page, see our notes on decibel meter apps.
Sources
- NIDCD: Noise-Induced Hearing Loss
- ASHA: Noise and Hearing Loss Prevention
- CDC/NIOSH: Noise-Induced Hearing Loss
- CDC/NIOSH: Understand Noise Exposure
- CDC/NIOSH: NIOSH Sound Level Meter App
- Kardous and Shaw (NIOSH), Evaluation of smartphone sound measurement applications, 2014
- Kardous and Shaw, A smartphone noise meter app in every pocket? (Acoustical Society of America lay paper)
- OSHA, 29 CFR 1910.95: Occupational noise exposure
- US EPA: EPA Identifies Noise Levels Affecting Health and Welfare
Frequently asked questions
Is this decibel meter free?
Yes. The meter runs in your browser with no account, no install and no time limit. It works on phones, tablets and computers that have a microphone and a current browser.
How accurate is an online decibel meter?
It is an estimate, not a certified instrument. Accuracy depends on your device's microphone and on calibration. Calibrated against a reference meter it can track everyday levels closely; uncalibrated, readings can be several decibels off, and very quiet or very loud sounds are the least reliable.
Does the meter record or upload my audio?
No. Sound is analyzed in your browser's memory to compute levels and is then discarded. Nothing is recorded, stored or sent to a server. Only your settings, such as the calibration offset, are saved in your own browser.
What is a normal decibel level in a house?
A quiet room is around 40 dBA and normal conversation is about 60 to 70 dBA. Appliances such as a vacuum cleaner or blender push levels to roughly 70 to 94 dBA while they run.
How many decibels is too loud?
Sounds at or below 70 dBA are unlikely to harm hearing even after long exposure. Long or repeated exposure at or above 85 dBA can cause hearing loss, and the louder the sound, the less time it takes.
Should I use A, C or Z weighting?
Use A for almost everything: it approximates how the ear responds and is the weighting used by noise guidance and regulations. Use C for bass-heavy sound and peak checks, and Z when you want the unfiltered level.
Can I use these readings for a noise complaint or an OSHA measurement?
Use them as supporting notes, not as proof. Enforcement and workplace compliance measurements call for calibrated sound level meters or dosimeters operated by a trained person.
Updated