Decibel Meter
Measure approximate ambient sound levels using your microphone — with an honest statement that a phone microphone is not a calibrated instrument.
Approximate, uncalibrated reading — your audio is analyzed live and never recorded.
| ~30 dB | Whisper, quiet library |
|---|---|
| ~50 dB | Quiet home, light rain |
| ~60 dB | Normal conversation |
| ~70 dB | Busy traffic, vacuum cleaner |
| ~85 dB | Hearing damage risk with long exposure |
| ~100 dB | Motorcycle, hand drill |
| ~110 dB | Rock concert, chainsaw |
| 120+ dB | Threshold of pain — jet takeoff nearby |
Averaging decibels is not averaging loudness
A decibel is a logarithm, and adding logarithms up and dividing by how many there were does not give the logarithm of the average. Take five seconds of a room: four quiet at 60 dB and one loud at 90.
| How it is worked out | Answer |
|---|---|
| Mean of the numbers | 66 dB |
| Mean of the energy | 83 dB |
17 decibels apart, and the second is the one that matters. Ten decibels is ten times the energy, so a 90 dB second carries a thousand times as much as a 60 dB one and dominates the exposure however many quiet seconds surround it. That is exactly why noise-exposure limits are written as an energy average, and the running figure above is now one.
The two agree only when nothing varies — on a steady tone they are the same number to the last decimal, which is why the discrepancy survives casual testing. The energy average is never the smaller of the two.
The top quarter of the bar could never be reached
The reading is the signal level in dBFS — decibels relative to the loudest thing the microphone can record — plus a fixed offset of 90. Full scale is 0 dBFS, and nothing is louder than full scale, so the reading can never exceed 90.
| Signal level | dBFS | Reading | Old bar | Now |
|---|---|---|---|---|
| 1 | 0.0 | 90.0 | 75% | 100% |
| 0.5 | -6.0 | 84.0 | 70% | 93% |
| 0.1 | -20.0 | 70.0 | 58% | 78% |
| 0.01 | -40.0 | 50.0 | 42% | 56% |
| 0.001 | -60.0 | 30.0 | 25% | 33% |
The bar was drawn against a range of 120, so a full signal filled three quarters of it and the last quarter was decoration. It is scaled to the reachable range now, so a full bar means a full signal.
The displayed number is smoothed rather than raw, which is why it does not flicker. It covers 9 frames — about 150 ms — to show nine tenths of a change, and it only ever moves towards the new value.
And the offset is a guess, which is the honest part
Turning a recorded level into a real sound pressure level needs a calibration: a known source at a known level, measured through this microphone, in this room. There is no way for a web page to have that, so the 90 is a reasonable shift and nothing more.
What survives is the difference. A rise of 10 on this meter is a rise of 10 dB whatever the offset happens to be, because the offset appears in both readings and cancels. Relative readings are trustworthy; absolute ones are an estimate.
So it is a good tool for "is this louder than that", "did the fan help", "how much does closing the window take off" — and a poor one for "is this room over 85 decibels". That distinction is worth more than a more confident-looking number would be.
How to use
- Allow microphone access.
- Hold the device away from your body and any obstruction.
- Read the approximate level.
- Use a calibrated meter for anything that matters.
Frequently asked questions
How accurate is this?
Approximate at best. Phone microphones are not calibrated, they vary between models, they are optimised for speech rather than measurement, and most compress loud sounds automatically. Errors of several decibels are normal, and the very loud end is where it is least reliable.
Why does the decibel scale behave so strangely?
Because it is logarithmic. An increase of 3 decibels doubles the sound power, and around 10 decibels is needed before most people judge a sound to be twice as loud. This is why 90 decibels is not slightly worse than 85 — it is substantially worse.
At what level does hearing damage occur?
Occupational guidance commonly puts the action level around 85 decibels over an eight-hour day, with safe exposure time halving for roughly every 3 decibels above that. At the levels found at a concert, safe exposure is measured in minutes rather than hours.
Is hearing damage cumulative?
Yes, and permanent. Noise-induced hearing loss accumulates over a lifetime and does not recover — the hair cells involved do not regenerate. Temporary muffling or ringing after loud exposure is a warning that damage has occurred, not a sign that it has been avoided.
What is A-weighting?
A filter that adjusts the measurement to match how sensitive human hearing is at different frequencies, since we hear mid frequencies far better than very low ones. Almost all published noise limits are A-weighted, which is why an unweighted reading is not comparable to them.
Can I use this for a noise complaint?
No. Noise regulation requires a calibrated Type 1 or Type 2 meter, specified measurement procedures and often a qualified operator. A phone reading is fine for curiosity and useless as evidence, and nothing here is a substitute for a professional assessment.
🔒 This tool runs entirely in your browser. Nothing you enter is uploaded, logged, or stored.