How accurate is a browser decibel meter?
Browser-based dB SPL measurement is uncalibrated. The Web Audio API exposes signal level in dBFS (digital scale, 0 = clipping) — we add an offset to estimate dB SPL based on a typical phone microphone, but the real number depends on your mic's sensitivity, the OS gain stage, and any automatic gain control. Expect ±5 to ±10 dB error versus a calibrated SLM. Trends and relative comparisons are reliable; absolute values are approximate.
Should I use a phone app instead of this web tool?
Use this tool for quick spot checks, comparisons between rooms, or to demonstrate how loud something is. Use a calibrated app (or a real SLM) for OSHA/legal noise readings, hearing protection decisions, or any data you'll publish. Native apps can access lower-level mic streams and ship per-device calibration tables, which a browser cannot. For most everyday curiosity — Is this restaurant too loud? Is my keyboard louder than my fan? — the browser is fine.
What do typical decibel levels sound like?
30 dB = whisper or a quiet bedroom at night. 40 dB = library. 60 dB = normal conversation a meter away. 70 dB = busy office or vacuum cleaner. 85 dB = lawn mower or heavy traffic — extended exposure causes hearing damage. 100 dB = motorcycle or jackhammer at distance. 110 dB = rock concert front row. 120 dB = jet engine at 50m. Each +10 dB is roughly 'twice as loud' perceptually and 10x the acoustic energy.
Why does iOS Safari show very low readings?
iOS aggressively applies automatic gain control and noise suppression on browser audio streams, even when the page requests them off. The signal arrives pre-compressed, so dB SPL estimates skew quiet — especially in noisy environments where AGC turns down the gain. Android Chrome respects the `autoGainControl:false` flag more faithfully. For iOS readings, use a native SLM app. On Android, our readings are within a few dB of a calibrated meter at moderate levels.
Can I calibrate this tool against a real SLM?
Not in the browser today — Web Audio doesn't expose a constant analog reference. The workaround used by serious tools is to measure a known source (e.g. play pink noise at 80 dB from a calibrated speaker), note our reading, and apply the offset by hand. For most users this is overkill; if you need true SPL, buy a $30 USB SLM or use a phone app with mic profiles.
What is the 60-second log for?
Press 'Start log' and we record one sample per ~100ms for 60 seconds, then export a CSV with timestamp + dB SPL columns. Useful for measuring how loud a fan, AC, or street is over a minute, then graphing it in Excel. Combine with our microphone noise floor at /microphone/noise for the digital-side equivalent.