Accuracy & Calibration

This page explains what our Online Sound Meter can and can’t claim, how to do reference-only calibration, and how to collect measurements that are reproducible and reviewable. We prioritize transparency over “perfect numbers.”

Important: Reference-only tool

This is not a certified Class 1 / Class 2 sound level meter. For legal enforcement, workplace compliance, or medical decisions, use certified instruments and qualified professionals.

For most everyday use, the most valuable output is often the repeatable difference between “before vs after” under the same setup.

What the meter measures

Outputs you see

Real-time level estimate derived from microphone input
Session stats: min, average, max for your current run
Frequency spectrum + history visualization for context
Saved report image generated locally on your device

What it is best for

Before/after comparisons (new fan, insulation, closed window)
Repeatable “is it getting worse?” tracking over days/weeks
Documentation workflow (logs + reports + context notes)
Education: spectrum patterns and burst behavior

Why accuracy varies (and how to reduce error)

Top error sources

Microphone sensitivity and device tuning differences
Operating system / browser audio processing (AGC, noise suppression)
Distance, orientation, and placement near reflective surfaces
Clipping/saturation at loud levels (phone mics often become nonlinear)

Most effective fixes

Measure a 60-second average, not a 1-second snapshot
Keep the same position/height/orientation for comparisons
Record context (room, windows, HVAC, distance markers)
Repeat 3 runs and compare averages first

Practical workflow: Measure decibels more reliably.

Reference-only calibration (offset) + verification

If you have access to a trusted reference (a professional sound level meter or a known calibrator), you can do an offset calibration at a realistic level (often 70–90 dB). Then verify that your device tracks changes reasonably across a few positions.

Align reading (in the app)

Start measuring and wait until the average stabilizes (30–60 seconds).
Open Align reading on the homepage and enter your reference device’s average.
Save. Your browser stores a constant offset for this device only (you can reset anytime).

This improves day-to-day usefulness on one device. It does not certify measurements for legal or workplace compliance.

Recommended steps

Pick a stable sound source (steady fan or tone/noise generator).
Measure 60 seconds at an anchor point and record avg on both devices.
Apply a single offset to align averages at the anchor point.
Verify tracking at 0.5 m and 2 m (and one context change such as door open).

Detailed guide: Phone Decibel Meter Calibration.

Algorithm summary (high level)

Core idea

Compute signal energy over time (similar to an average/Leq workflow).
Apply an A-weighting adjustment so results are closer to a dBA-style reading.
Apply a constant offset (optional) to align one device to a trusted reference.

Important limitation

Browser and OS audio processing can change the microphone signal.
Different microphones have different sensitivity and frequency response.
Without external calibration, absolute dB values may be shifted.

Verification evidence: protocols + public datasets

To make accuracy claims reviewable, we publish a validation protocol, a device/browser matrix, and data templates for runs that can be replicated by others.

See: Validation Protocol & Data.

Privacy and data flow

Audio processing runs locally in your browser. We do not record or upload raw microphone audio. “Save Report” generates a file on your device using summary statistics from your current session.