What does a browser decibel reading mean?

Two microphones, two digital signals

Your microphone converts sound into an electrical signal, which the device turns into digital samples. Decibel Reader calculates the root mean square, or RMS, across a short block of samples rather than selecting its largest sample.

The Web Audio specification defines how the browser supplies those samples. It does not supply the microphone's acoustic sensitivity: the relationship between sound pressure and digital amplitude.

Two laptops on the same desk may therefore show different readings. Changing input gain can move the number while the room sounds much the same. Keep the device and its settings consistent.

Why dBFS is usually negative

Settings shows the raw digital level. dBFS means decibels relative to digital full scale. The calculation is 20 × log10(RMS): an RMS amplitude of 0.1 gives −20 dBFS. This is a mathematical example, not a room measurement.

A negative value means the signal is below the digital reference, not negative sound pressure. Halving the amplitude lowers the level by about 6.02 dB.

Perfect numerical silence has no finite logarithmic level. The tool uses −120 dBFS as a finite floor. Reaching it does not establish that a room is silent or reveal the microphone's acoustic noise floor.

The offset behind Estimated dB

The display uses Estimated dB = dBFS + base offset + manual adjustment. The default base offset is 94 dB, a display mapping rather than a measured sound-pressure reference. Manual adjustment can add or subtract up to 30 dB.

Reference alignment replaces the base offset and clears the manual adjustment. The label becomes Reference aligned; that match applies to the setup you checked. The quiet-room baseline leaves the offset alone.

Changing the mapping resets your statistics. Reset offset restores the uncalibrated 94 dB base and zero manual adjustment.

This tool has no A-weighting filter. Its readings are not dBA or certified SPL. The accuracy explanation covers limits an offset cannot fix.

Use the average for steady sounds

The live number follows the current sampled signal. Average summarizes the valid sampled period; Min and Max show its lowest and highest displayed sample levels. For the desk fan, Average is more useful than whichever number appears when you glance at the screen.

Decibels are logarithmic. The tool converts levels into linear energy, weights them by sample duration, then converts back to decibels. A loud interval contributes more than an equally long quiet interval. Directly averaging dB numbers would miss that relationship.

Check the history for events such as someone speaking near the microphone. The live chart shows the latest 30 seconds; statistics can cover longer sessions. Compare similar durations. A short sample only describes that sampled period.

Peak Hold keeps the highest displayed level

Peak Hold keeps the highest displayed RMS-window level in the session, helping you notice a brief increase you missed.

It is not an instantaneous acoustic peak measurement. A short impulse may fall between sampled windows, overload the microphone, or be altered by device processing. Do not use it to characterize gunfire, explosions, or other impulse noise.

Compare with the same setup

Set the device on a stable surface with its microphone opening clear. Measure for 30–60 seconds, then Stop and save the report before starting again. Change the fan setting—or close a window—without changing the device's position, orientation, or input settings. Repeat for the same duration, then return to the first condition and measure again to check that result.

A falling reading during a steady sound may indicate automatic gain or noise processing. A flat reading warrants checking the input; a sudden ceiling may indicate overload. Repeat before relying on a small difference.

Keep the page in the foreground. Suspension, an observation gap, or a disconnected input can interrupt the session; missing time is not filled with silence. Stopping, leaving the page, or losing the input releases the microphone and audio resources.

Local PNG and print/PDF reports include session time and duration, statistics, offset, accuracy status, and a full-session overview, condensed for longer sessions. They contain no audio. Keep separate notes about placement and conditions when sharing one. The room comparison guide gives a fuller sequence.