Decibel Meter
Measure the room with your microphone: live dB estimate, a 60-second history curve, and a Leq average — all local, all approximate.
Press Start to grant microphone access. Readings are uncalibrated estimates shown only on this page — never recorded or uploaded.
🔒 This tool runs locally in your browser — no audio is uploaded.·Keyboard: Space play / Esc stop
How to use
Start and allow access
Press Start and allow microphone access. The browser asks only after your click, and the stream is analysed on this page — it is never recorded or uploaded.
Read the level and band
The big number is the estimated level, colour-coded by band: quiet, normal conversation, loud, or potentially harmful (above the 85 dB(A) line NIOSH uses for the 8-hour limit).
Calibrate against a known source
Consumer mics are not measuring instruments. If you have a trusted meter (or a phone app calibrated against one), match this page to it with the offset slider; the setting is remembered locally.
How it works & limits
The microphone stream is opened with getUserMedia and analysed locally by the Web Audio API. The page computes the RMS of the samples, expresses it in dBFS, and converts it to an estimated sound level using a typical mobile-microphone mapping (0 dBFS ≈ 94 dB SPL) plus your calibration offset. The Leq average is the energy mean of the estimates since start, alongside the minimum and maximum, and the graph keeps a rolling 60-second history sampled ten times per second. Stopping releases the microphone track; nothing is recorded or uploaded.
This is an estimate, not a measurement. Browser microphones have unknown sensitivity and frequency response, phone voice processing reshapes the signal (this page disables echo cancellation and automatic gain where the browser allows), and the 0 dBFS ≈ 94 dB SPL mapping is an assumption, not a calibration. Treat the numbers as order-of-magnitude references that are good for comparing rooms and times on the same device — not for noise compliance, which requires a calibrated Class 2 (or better) sound level meter. The optional A-weighting applies the standard A-weight curve per FFT bin to approximate how the ear weighs frequencies.
Common noise levels (with sources)
| Scene | Typical level | Source |
|---|---|---|
| Whispering, quiet library | ≈ 30 dB(A) | CDC |
| Normal conversation | 60–70 dB(A) | CDC |
| City traffic heard from inside a car | 80–85 dB(A) | CDC |
| Motorcycle | ≈ 95 dB(A) | CDC |
| Approaching subway train | ≈ 100 dB(A) | CDC |
| NIOSH recommended exposure limit (8-hour average) | 85 dB(A) | NIOSH REL |
| Safe-listening weekly sound dose for personal audio | 80 dB(A) weekly average | WHO / ITU-T H.870 |
Values are quoted from the public pages named in the last column; this tool only estimates where the room you are measuring falls among them.
FAQ
Not accurate in an absolute sense. It depends on your microphone’s sensitivity, its frequency response, and assumptions about the signal chain. Expect errors of several dB or more between devices. It is built for relative comparisons — quieter or louder than before, room A vs room B — on the same device.
It shifts every reading up or down by the set amount (±20 dB) so you can match this page to a source you trust, such as a calibrated meter running beside it. The offset is stored in your browser (st.db.cal) and applies only on this device.
A-weighting filters the level the way human hearing does: less sensitive to very low and very high frequencies. Standards such as the NIOSH 85 dB(A) exposure limit are expressed in dB(A). This page approximates A-weighting per FFT bin, so it too is an estimate.
Microphones differ. Phone mics are tuned for voice calls, laptop mics vary widely, and cases or covers can block sound. Keep the device position constant and compare one device against itself; across devices, expect a constant offset that the calibration slider can remove.
From public health guidance: the CDC “What Noises Cause Hearing Loss?” page, the NIOSH recommended exposure limit (85 dB(A) as an 8-hour time-weighted average), and the WHO/ITU safe-listening standard (80 dB(A) weekly dose). Each row cites its source — nothing on this page is invented.
No. The analysis runs in this tab with the Web Audio API, no recording is ever created, and the only thing persisted is your calibration offset preference. Closing the page releases the microphone.
Related tools
Mic Test
Live level meter, device picker, and local record & playback. Nothing uploaded.
Open toolSpectrum Analyzer
Live FFT bars, waveform, and spectrogram from your mic — all computed locally.
Open toolWhite & Pink Noise Generator
White, pink, and brown noise with a 15–120 minute timer for focus or sleep.
Open tool