Classroom Noise Meter
A large colour-coded volume display for a projector, with a sustained-noise sound alert and one-click fullscreen. Shows a relative level, not decibels.
🔒 This tool runs entirely in your browser. Your files are never uploaded to a server.
This needs your microphone
Audio is analysed in the page and never recorded, saved or sent anywhere. Nothing is written to disk and closing the tab ends it completely.
Silent work
0
relative level, 0–100 — not decibels
Stops a single dropped book setting it off.
Only useful if you have a real meter to match.
This is not a sound level meter. The number is a relative level from 0 to 100, useful for comparing "now" against "a minute ago" in one room with one device. It is not decibels and cannot be: a browser gives no access to the microphone's absolute sensitivity, and the operating system may be applying automatic gain that changes as the room changes. For anything involving hearing safety or a noise complaint you need a calibrated meter, not a laptop.
How to use it
- Turn on the meter and allow microphone access.
- Put it on the projector or the interactive board where the class can see it, and select Fullscreen on the display for maximum visibility.
- Set the alert level to whatever this activity should sound like.
- Give it a few seconds of patience so one dropped book does not trip it.
- Leave the sound alert on to hear a chime when the class stays too loud, or switch it off if the visual cue is enough.
Why there is no decibel reading
Nearly every browser-based noise meter displays a decibel figure, and nearly every one of them is fabricating it. A decibel of sound pressure is a ratio against a fixed reference — 20 micropascals — and to convert a microphone signal into that you need to know two things the browser will not tell you:
dB SPL = 20 · log₁₀(pressure / 20 µPa)
needs: the microphone's sensitivity unavailable
needs: the exact gain applied unavailable, and it moves
What a browser hands you is a normalised waveform between −1 and 1. You can measure that faithfully, and this does — as a relative level on a logarithmic scale, since hearing is logarithmic and a linear meter would spend most of its travel on sounds nobody would call loud. What you cannot do is claim it is 68 dB. Calling it a relative level is the only honest label, and it costs nothing in usefulness: for a classroom you care whether the room is louder than it was five minutes ago, which a relative scale answers perfectly well.
RMS, not peak
The level is the root mean square of each analysis window rather than its loudest sample. That matters in a room full of people: a chair scraping produces a single enormous spike that a peak meter would show as sustained noise, while thirty people talking quietly produces a lower peak and far more energy. RMS tracks the second one, which is what "the room is too loud" actually means.
On top of that the display uses asymmetric smoothing — rising 45% of the way to the new value each frame, falling only 6%. That is how analogue VU meters behaved, and for the same reason: a needle that tracked the signal exactly would be a blur. Fast attack catches the shout; slow release keeps the number readable from the back of the room.
Fighting the browser's helpfulness
Microphone streams are optimised for voice calls by default, which is precisely wrong here. Three features actively destroy what this tool measures:
echoCancellation off removes room reflections
noiseSuppression off removes background noise — the subject
autoGainControl off rescales as the room changes
Automatic gain is the worst of the three, because it makes a quiet room read the same as a loud one — it exists to keep a voice at a constant level, which is the exact opposite of measurement. All three are requested off, but not every device honours the request, so the tool reports what it actually got underneath the buttons. If gain control is still on, treat the numbers as indicative only.
The microphone is also deliberately not connected to the speakers. Routing a classroom microphone to the classroom speakers produces feedback, and a noise meter that generates its own noise would be a poor joke.
Where this must not be used
Not for hearing safety, not for a workplace noise assessment, and not for a noise complaint. Those need a calibrated meter with a known reference, usually A-weighted and often with a certificate. This is a visual cue for a room of people — a shared, obvious signal that saves the teacher saying "quieter, please" for the ninth time — and that is a genuinely useful thing for it to be.
FAQ
Why does it not show decibels?
Because it cannot, honestly. Decibels of sound pressure are measured against a fixed reference, which requires knowing the microphone's sensitivity and the exact gain applied to it. A browser gives you neither, and the operating system may be adjusting the gain as the room gets louder. Any web page printing "72 dB" from a laptop microphone is showing you a made-up number.
So what is the number?
A relative level from 0 to 100 on a logarithmic scale. It is meaningful for comparing this minute against the last one, in the same room, on the same device, without moving the laptop. It is not comparable between rooms or devices, and it is not a safety measurement.
What is the calibration offset for?
If you own a real sound level meter, you can shift this scale until the numbers roughly line up in your room, which makes the thresholds more meaningful to you. Without a reference meter it does nothing useful, which is why it defaults to zero.
Why does the meter drop slowly but rise instantly?
Deliberate ballistics, borrowed from analogue VU meters. A fast rise means a sudden noise registers immediately; a slow fall means the display stays readable instead of flickering. A meter that tracked the signal exactly would be unreadable at a glance, which defeats the purpose of putting it on a projector.
Is the classroom being recorded?
No. The audio is analysed frame by frame in the page and discarded — nothing is buffered to disk, nothing is uploaded, and there is no recording to recover. Closing the tab or pressing the off button releases the microphone entirely, and the browser will show its own indicator whenever it is live.
Why did the level jump when nothing changed?
Almost certainly automatic gain control. This asks the browser to turn it off, along with noise suppression and echo cancellation, but not every device honours the request — and those features exist precisely to remove the background noise you are trying to measure. If the device reports gain control still on, the meter says so under the buttons.
Does it make a sound when the class is too loud?
Yes, by default. Once the sustained-noise alert triggers it plays a short synthesised chime, then repeats every few seconds for as long as the room stays over the threshold, so you do not have to be watching the screen. It is a generated tone, never the microphone signal, so it cannot feed back through classroom speakers. Turn it off with the Sound checkbox next to the alert slider if the visual cue alone is enough.
Can I make it fullscreen for the projector?
Yes. The Fullscreen button on the coloured display expands just the level and the alert text to fill the screen, so it stays readable from the back of the room. Select it again, or press Escape, to leave fullscreen.
How we compare
| Feature | Online Tool Store | Other web noise meters | A calibrated SPL meter |
|---|---|---|---|
| Honest about what it can measure | ✓ | Usually shows fake dB | ✓ |
| Turns off gain control and noise suppression | ✓ | ✗ | Not applicable |
| Sustained-noise alert, not a single spike | ✓ | Rarely | ✓ |
| Audible alert, not just visual | ✓ | Some do (e.g. Classroomscreen, KiwiBee) | Varies by model |
| One-click fullscreen for the projector | ✓ | Some do (e.g. KiwiBee) | Not applicable |
| Nothing recorded or uploaded | ✓ | Varies | ✓ |
| Free, on any device with a browser | ✓ | ✓ | ✗ |
| Real decibels, A-weighted | ✗ | ✗ | ✓ |
| Valid for a safety or legal assessment | ✗ | ✗ | ✓ |
A clear visual signal a whole room can see, on the device already connected to the projector, with no recording and no invented numbers. It is not a measuring instrument and does not claim to be — for anything where the figure has consequences, use a calibrated meter.