From microphone to digital level
After you press Start and grant permission, the browser provides a live microphone stream. An analyser reads short blocks of time-domain samples. For samples x₁ through xₙ, we calculate RMS = √(Σx² / n) and then dBFS = 20 × log10(RMS).
The Web Audio API provides these digital samples. They describe the signal reaching the browser. The API does not establish an acoustic calibration for the device microphone.
The raw digital silence floor is −120 dBFS. It prevents non-finite logarithms and keeps the interface usable. It is not a measured acoustic noise floor. The instrument has no A-weighting filter, and its readings must not be labeled dBA.
How Estimated dB is displayed
The main display uses Estimated dB = dBFS + base offset + manual adjustment. The default base offset is 94 dB. This default is a display mapping, not a calibrated sound-pressure reference.
Manual adjustment can add or subtract up to 30 dB from the current base offset. The number itself is not clamped to the visual ruler: only the ruler position is limited to its 0–130 range. A number outside that range still needs the same accuracy caution.
Statistics and recent history
Min and Max are the lowest and highest valid sampled display levels in the current session. Peak Hold retains the highest displayed RMS-window level; it is not an instantaneous acoustic peak metric.
Average is calculated in the energy domain, with valid sample durations as weights: Lavg = 10 × log10(Σ(t × 10^(L/10)) / Σt). Directly averaging dB numbers would underestimate the contribution of louder intervals.
The chart keeps the latest 30 seconds of valid history. Session statistics can cover a longer period than the chart. Browser suspension and other observation gaps are not inserted as silent samples. Calibration changes reset statistics so readings on different scales cannot be mixed.
Reference alignment and relative baseline
Reference alignment takes a 10-second sample under a steady condition. You enter the reading from a reference instrument observing the same sound field. The new base offset is reference level − sampled raw energy-average level; the manual adjustment returns to zero.
The status is Reference aligned. This describes the action performed, not verification of the reference meter or an acoustic accuracy tolerance. Recheck after changing the microphone, gain, browser, or device settings. Follow the reference alignment guide.
A relative baseline observes a room's digital level without assigning an acoustic value. The quietness of an unknown room cannot supply a known sound-pressure reference, so a baseline does not create SPL calibration.
Browser processing and microphone release
We request capture with echo cancellation, noise suppression, and automatic gain control disabled. The actual settings depend on browser and hardware support. Where available, the tool reports whether processing remains enabled; an unsupported setting is not proof that the signal is unprocessed.
Stopping, leaving the page, or losing the input releases the microphone tracks and audio processing resources. Each fresh start uses one capture pipeline. The signal is not connected to speaker playback.
Local reports
A report contains the session time, valid sampled duration, Min/Average/Max, full-session overview, offset, and accuracy status. PNG creation and the print view run locally. Printing to PDF uses your browser or operating system's print facility. Reports contain no audio.
These calculations establish consistent digital arithmetic, not acoustic calibration. Read what can affect accuracy before using a result.