Updated
What a browser decibel meter can and cannot tell you
Your microphone does not report decibels. It reports a digital signal, and the loudest signal it can capture is a fixed ceiling called full scale. What this page measures exactly is how far the sound sits below that ceiling. To turn that into the familiar sound level meter number, the meter adds a calibration offset, the distance between full scale and real sound pressure for your particular device.
That offset is the honest weak spot of every phone and laptop decibel meter. It depends on the microphone capsule, the gain the operating system applies, the case and even the angle of the phone. This page starts from a round assumption of 100 dB, which is a starting point and not a measurement of your microphone. Until you calibrate, treat the number as an estimate that can be 10 dB or more away from a professional instrument, and trust comparisons more than absolute values: whether the room got louder after lunch, which corner is quieter, whether a fan setting changed anything.
Within one device the meter is consistent. The same sound at the same distance gives the same reading tomorrow, so you can track changes, set a classroom limit and compare before and after. What it cannot do is replace a certified instrument for a legal noise complaint, a workplace assessment or a hearing conservation program.
Decibels in plain numbers
A decibel is a ratio on a logarithmic scale, which is why decibels behave strangely if you expect them to add up like meters or pounds. Every 10 dB step means ten times the sound energy. Every 3 dB means twice the energy, which is why two identical speakers playing together read about 3 dB higher than one, not twice the number. Every 6 dB doubles the sound pressure, the quantity a microphone actually responds to.
Your ears judge loudness differently again. A rise of about 10 dB usually sounds roughly twice as loud, and a change of 1 dB is hard to notice unless you switch back and forth. So when the meter climbs from 55 to 65, the room did not get a little louder; to most listeners it doubled.
Sound pressure level is measured against a reference of 20 micropascals, about the faintest tone a young ear can hear at 1 kHz, which is defined as 0 dB. Ordinary conversation at a meter or so is usually quoted around 60 dB, a busy restaurant in the 70s and a loud concert above 100. Those are rough landmarks; the meter shows what your room is doing right now.
A or Z, Fast or Slow: the settings explained
Weighting decides how much each frequency counts. A-weighting follows the ear at moderate levels: it turns deep rumble and the very top of the range down before measuring, so a truck idling outside counts less than the same energy in the voice range. Most workplace and community noise limits are written in dB(A), so A is the default. Here the A curve is built from the formula in IEC 61672, and the digital filter stays within about 0.3 dB of it from 20 Hz to 10 kHz.
Z-weighting is flat, every frequency counts the same. It is useful when you care about bass, such as a subwoofer, an air handler or a car audio system, because a dB(Z) reading higher than dB(A) by a wide margin tells you the energy sits in the low end.
Response is how quickly the number follows the sound. Fast uses a time constant of 0.125 seconds and catches claps, door slams and the peaks of speech. Slow uses one second and gives a calmer number that is easier to read across a room. The Average shown under the readout is the equivalent continuous level, often written Leq: the steady level that would carry the same energy as everything since you pressed Start or Reset. It is the figure noise limits usually refer to, and it is always closer to the loud moments than a simple average of the readings would suggest.
How to calibrate it against a real meter
If you can borrow a sound level meter, five minutes of calibration turns this page into a far better tool. Play steady pink noise or a constant fan sound, hold your phone or laptop microphone right next to the reference meter, set both to A-weighting and Slow, and change the Calibration offset until the two numbers agree. The offset is remembered on this device, so you do it once per microphone.
Check two levels if you can, one quiet and one loud, because cheap microphones can drift at the ends of their range. If the readings agree at 50 dB but not at 90, trust this meter only in the range where they agreed. A calibration is tied to one microphone, one browser and one position: a headset, a USB microphone or a phone case changes it, and so can a system update.
The microphone settings matter too. The page asks the browser to switch off echo cancellation, noise suppression and automatic gain, all of which would bend the reading. Most desktop browsers honor that request; some phones keep part of their processing anyway, which is another reason a calibrated reading is worth more than a default one.
Using it as a classroom noise monitor
Teachers use a classroom noise monitor or a noise meter for classroom work time to make volume visible instead of repeating requests to settle down. Turn on Classroom display and the number grows large enough to read from the back of the room. Under the Noise limit the reading is green, within 5 dB of the limit it turns amber, and above the limit it turns red while the page counts the seconds spent over it.
Set the limit from your own room rather than from a chart. Let the class work at a volume you are happy with for a minute, note the Average, and set the limit 5 to 8 dB above it. A limit far below the natural working level only ever shows red, and children stop looking at it. Because the comparison is with the same room and the same device, this works perfectly well without calibration: a noise level meter for classroom use needs consistency, not a certified number.
Place the laptop or tablet where it hears the room, not right beside the loudest table, keep it plugged in, and set the screen not to sleep in the device settings, since the page cannot stop your system from dimming. Reset at the start of each activity so the time over the limit belongs to that lesson.
Loud enough to matter: levels and hearing
Hearing damage depends on both level and time. The Numbers tab shows the recommended exposure limit published by NIOSH, the US occupational health institute: 85 dB(A) averaged over an 8-hour day, with the allowed time halving for every 3 dB more. By that rule 94 dB(A) is a one-hour level and 100 dB(A) a fifteen-minute one. These are workplace guidelines for adults, and they assume a properly calibrated measurement.
That last point is why this page never declares a level safe. An uncalibrated reading of 80 could be 75 or 90 in reality, which is the difference between harmless and a real risk over a long shift. If a sound is loud enough that you need to shout to be heard at arm's length, treat it as too loud regardless of what any phone says, and use hearing protection.
For sustained loud environments, such as a workshop, a band rehearsal space or a factory floor, the decibel reader on a phone is a fine first check that tells you a proper assessment is worth doing. It is not the assessment, in the same way that a simple sound meter is not a dosimeter worn for a full shift.
Phones, laptops and the edges of the range
Built-in microphones are designed for voices, not measurement, and every online decibel meter, sound decibel meter app or dB meter built into a phone runs into the same physics. In a very quiet bedroom the meter may never drop below the high 20s or 30s because the microphone and its amplifier add their own faint hiss, so a library reading of 35 can partly be the device listening to itself. At the other end, a very loud sound can reach full scale and clip; the number then stops rising even though the sound keeps getting louder.
Laptop microphones often sit near the fan and keyboard, so typing while you measure adds clicks to the reading. Phones pointed at a sound read differently from phones lying flat. For the most repeatable numbers, keep the device still, the microphone uncovered and pointed the same way every time, and avoid touching it during a measurement.
On iPhone and iPad the meter works in Safari and other browsers once you allow the microphone. If the page reads nothing at all, the permission was denied or another app holds the microphone; close calls and recording apps and reload.
Where the sound goes
Nowhere. The microphone signal is filtered and measured inside your browser and is never recorded or uploaded. The page keeps only two numbers in this browser's storage, your calibration offset and your noise limit, so the meter opens the way you left it.
Frequently asked questions
How accurate is a phone decibel meter?
Uncalibrated, expect it to be within about 10 dB of a real sound level meter on a good day and further off on some devices, because every microphone and operating system adds its own gain. It is consistent, though: the same sound gives the same number on the same device. Calibrate the offset against a reference meter and the reading usually comes within a few decibels in the middle of the range.
What is the difference between dB, dBA and dBZ?
dB on its own is just the unit. dB(A), often written dBA, means the sound was A-weighted first, so low rumble and the highest frequencies count less, roughly as the ear hears at moderate levels. dB(Z) is flat, every frequency counts fully. A reading in dB(Z) far above dB(A) means most of the energy is bass. Noise limits are usually given in dB(A).
What decibel level is too loud?
By the NIOSH recommendation, 85 dB(A) averaged over 8 hours is the limit for a working day, and every 3 dB more halves the time, so 100 dB(A) is a 15-minute level. A simpler test: if you have to shout to be understood at arm's length, it is too loud for long. An uncalibrated phone reading cannot settle this on its own.
How do I use this as a classroom noise monitor?
Start the meter, let the class work at a volume you accept for a minute, then set the Noise limit about 5 to 8 dB above the Average shown. Turn on Classroom display for big numbers. Green means under the limit, amber within 5 dB of it, red over it, and the page counts how long the class spent above the limit.
Why does a silent room read 30 or 40 dB?
Partly because few rooms are really silent: fridges, ventilation and distant traffic add up. And partly because the microphone and its amplifier add a faint hiss of their own, which sets a floor the meter cannot go below. The floor differs between devices, which is one more reason to compare readings on the same device rather than across phones.
Can I use it as an SPL meter for speakers or car audio?
For relative work, yes: comparing positions, settings or before and after a change, using Z-weighting if bass matters. For absolute SPL figures, calibrate first, and remember built-in microphones roll off at the lowest frequencies and can clip at high levels, so a loud subwoofer may read lower than it really is.
Is the sound recorded or uploaded?
No. The microphone signal is analyzed in your browser while the meter runs and is never recorded or sent anywhere. Only the calibration offset and the noise limit are stored, in this browser on this device.