How to Calibrate a Browser Decibel Meter

Why calibration matters

Every consumer microphone has different gain, so the same sound produces different raw readings on different devices. Until you measure that difference against a known reference, the meter can only show microphone input strength — never an environmental sound level in dBA.

What you need

• A stable reference of known level: a calibrated sound level meter side by side, or an acoustic calibrator tone at a stated level (commonly 94 dB).
• The same device, microphone, position and browser settings you will measure with later.
• Matching frequency weighting (A, C or Z) and time response (Fast or Slow) on both sides.

How to calibrate

1. Place the device where you will measure — stable surface, ear height, mic port uncovered, away from wind.
2. Expose it to the stable reference and let the display settle for several seconds.
3. Read the uncalibrated value on this site and enter the difference (reference − shown) as the offset in Customize → Calibration.
4. Confirm the status badge changes from “Uncalibrated” to “Calibrated” and re-check against the reference.

Saving device-specific profiles

The offset is stored in this browser for the current microphone and weighting setup. It auto-applies only when the same setup is detected; change the input device and the profile stops applying rather than silently producing wrong numbers. Reset any time to return to the raw estimate.

When profiles become incompatible

• You switch microphones, headsets or USB inputs.
• You change weighting (A/C/Z) or response (Fast/Slow).
• The browser or OS changes gain, enables AGC, or updates audio processing.
• You measure in very different acoustics (near-field vs room) — re-check against a reference.

Common calibration mistakes

• Calibrating with music or speech instead of a stable tone — levels jump too much to read.
• Holding the phone in your hand while calibrating, then measuring on a table.
• Mixing weightings: calibrating in dBA, then comparing in dBC.
• Assuming one phone's offset transfers to another device.
• Treating a calibrated browser as a certified meter — it remains an estimate (see browser measurement accuracy).

Worked offset example with real numbers

Borrow a reference meter and set both devices on small tripods, capsules at the same height, one metre from a loudspeaker playing steady pink noise. Let Slow response settle. The reference holds 74.0 dBA while this site holds 66.3 as an uncalibrated estimate. The difference is +7.7 dB, so you open Customize → Calibration and save +7.7 for this microphone and A-weighting. Play the same noise again: the panel should now hover near 74, drifting perhaps between 73.6 and 74.4 across a 60-second Leq. Write down both averages plus duration before you trust the profile. One steady tone beats ten minutes of fluctuating speech, because speech swings 10 dB with every sentence and you end up chasing peaks.

What the panel shows you at each step

Before you save anything the badge reads Uncalibrated and the number carries an estimate label — weighted dBFS plus the nominal +100 dB offset, with raw dBFS still inspectable in Digital Diagnostics. After you save, the badge flips to Calibrated and the engine swaps the nominal offset for your reference-derived one, shown in your chosen dBA, dBC or dBZ unit on the 20–120 visual scale. Swap in earbuds with a built-in mic and watch the profile stop applying rather than invent numbers: the stored offset belongs to one microphone plus one weighting and response pairing. Session stats keep showing current, minimum, Leq, maximum and duration through all of this, the graph keeps the last 240 points, and CSV or JSON export includes dB values only while a compatible calibration is active. Reset any time and you fall back to the raw estimate.

When to redo the offset instead of trusting it

• You changed anything physical: new phone, different headset, USB mic, fresh case covering the port, or table mount versus hand-held.
• You changed anything acoustic in settings: A to C or Z, or Fast 125 ms to Slow 1 s — each pairing keeps its own offset.
• The browser updated, the OS toggled gain, or the processing flags now report echo cancellation or gain control active when they were off before.
• You moved from room sound to near-field sound, like miking a speaker at 20 cm after calibrating at 1 m — re-check against the reference at the new distance.
• You are tempted to copy the number to another device — never do this. Measure the second device against the same reference in the same spot and save its own profile, then confirm with guidance in how to use the meter.

Sources and technical references

For context on validating phone measurements against reference instruments, see the CDC/NIOSH Sound Level Meter app guidance. Browser input constraints are documented by MDN.

Open the meter → · How the meter works → · Decibel chart →