Background Noise Test

Make the comparison repeatable

Place the phone or laptop on a stable surface with its microphone uncovered. Keep the same position, selected input, weighting and response for each comparison. Avoid touching the device while measuring.

Why phones disagree

Microphone sensitivity, internal gain, automatic gain control, noise suppression and case design vary between devices. The uncalibrated dBA value therefore uses a nominal +100 dB offset and is an orientation estimate, not a formal instrument reading.

Improve the estimate

A compatible local calibration profile replaces the nominal offset with a device-specific reference comparison. Even calibrated browser estimates are not suitable for workplace, legal or hearing-safety decisions. Review how to calibrate, accuracy and limitations, and dBFS vs dB SPL.

Run a 60-second room check that means something

Close the door, set the phone face-up on a table at ear height, and leave it alone. Press Start, let Fast smoothing settle for about five seconds, then watch a full 30 to 60 seconds without typing or walking past the microphone. Keep A-weighting and the same response for every comparison, and note the Leq energy average plus the duration rather than chasing the jumping live number. The on-page graph holds 240 points so you can see whether the room was steady or just quiet between gusts from a vent.

Read the four numbers together, with an example

Say an evening bedroom shows current 38, Leq 39, minimum 34 and maximum 53. The minimum tells you how quiet the room gets when the fridge cycles off. The maximum catches the one door slam that skewed your impression. The Leq sits near 39 because it averages energy over time instead of copying the loudest second. Treat the uncalibrated figure as an orientation estimate built from weighted dBFS plus a nominal +100 dB offset, round away the decimals, and compare only readings taken in the same spot with the same input. Raw dBFS stays visible under Digital Diagnostics when you want the unadjusted digital level.

Mistakes that inflate a room reading

• Holding the phone: handling rumble and breath add several decibels. Use a stable surface with the microphone uncovered.
• Measuring near vents, windows or keyboards: move a metre away and pause typing while the sample runs.
• Swapping phones mid-comparison: sensitivity and gain differ, so never carry an offset across devices. A local calibration profile is device-specific, done against a known reference in the same spot, and worth redoing whenever the setup changes.
• Mixing weightings or Fast and Slow response between runs: Fast follows peaks every 125 ms while Slow smooths over 1 s, so lock one choice per comparison.

Access asks for permission once per origin over HTTPS, captures in mono with echo cancellation, noise suppression and automatic gain requested off where the browser allows it, and never routes input to speakers. For the digital side of the same room see the microphone noise-floor test, for weighting choice see dB vs dBA, and for the signal path see how the meter works.

Run a 30-second microphone noise-floor test → · Use the full meter →