How to Calibrate a Decibel Meter on Your Phone or Laptop

A consumer phone or laptop microphone, used straight from the factory, can report sound levels that are off by 5 to 15 dB from a calibrated Class 2 sound‑level meter. That is not a defect — it is the unavoidable consequence of using a device that was engineered for voice calls and ambient noise suppression as if it were a measurement instrument. The good news is that calibrating away most of this error is a 5‑minute job, requires no special equipment, and brings the browser meter on this site into agreement with a reference within ±2 dB for typical environmental and occupational ranges.

This page covers three calibration methods (in order of accuracy and practicality), explains how to set the calibration offset in our settings panel, and is honest about the limits of what consumer hardware can do.

Why calibration matters

The decibel scale is calibrated to a fixed physical reference: 0 dB SPL = 20 micropascals (µPa) of pressure variation. A correctly calibrated sound‑level meter, given a 94 dB SPL reference signal, reads exactly 94 dB. Anything less is a measurement bias — a constant offset between the level of the actual sound and the number on the screen.

Three sources of bias dominate consumer hardware:

• Microphone sensitivity tolerances. MEMS microphones in phones and laptops are typically specified at −38 dBV/Pa ± 3 dB. That ±3 dB spread translates directly into ±3 dB of measurement bias before any other factor.
• Pre‑amp gain and ADC scaling. The signal chain from the microphone through the analog front end to the digital sample value adds a few decibels of variability and is not standardized across devices.
• Audio pipeline processing. Most modern operating systems and browsers apply automatic gain control (AGC), echo cancellation, noise suppression, and equalization to microphone input. Each of these changes the relationship between the room SPL and the digital sample value, often non‑linearly with level.

The browser meter on this site already requests the microphone with processing disabled where the platform supports it (echoCancellation: false, noiseSuppression: false, autoGainControl: false in the getUserMedia constraints). Even so, residual platform processing and microphone tolerances mean every device benefits from a one‑time calibration.

Method 1 — Reference SLM cross‑check (the practical default)

If you have access to a calibrated Class 2 sound‑level meter (yours, a co‑worker's, an EHS department's), this is the easiest and most reliable calibration. It captures the offset under a realistic room signal and takes about five minutes.

1. Place both meters together. Set your phone or laptop running our meter side‑by‑side with the reference SLM, microphones at the same height and orientation, ideally within 5 cm of each other.
2. Pick a steady source at moderate level. A constant white‑noise source at 60 – 80 dB is ideal — running fan, ventilation duct, dial tone on speakerphone. Avoid music with loud transients and avoid the quietest environments where the noise floor of either meter dominates.
3. Configure both for matching settings. A‑weighted, Slow time weighting on both. Slow averages out short fluctuations that would otherwise make the comparison noisy.
4. Average for 30 – 60 seconds. Watch both readings. Most reference SLMs show a running Leq; for our meter, watch the Avg statistic.
5. Compute the offset. Reference reading minus our reading, in dB. If the reference reads 70 dBA and ours reads 65 dBA, the offset is +5 dB.
6. Enter the offset in settings. Open the gear menu, scroll to "Calibration offset," enter the computed value (signed). Hit save.

A second 30‑second comparison after entering the offset should show the two meters within 1 dB of each other. If they differ by more than 2 dB on a steady source, your device's frequency response is non‑flat at the test source's spectrum — try a broadband source like a fan or HVAC duct, not a tonal source like a TV.

Method 2 — Acoustic calibrator (the gold standard)

Industrial hygienists and acoustic researchers calibrate with a pistonphone or acoustic calibrator — a device that fits over the microphone and produces a precise reference tone, typically 94 dB SPL or 114 dB SPL at 1 kHz, traceable to national metrology standards (IEC 60942 Class 1 or Class 2).

If you have access to one (e.g., from an EHS or audio engineering team):

1. Fit the calibrator over the microphone. This is the awkward step on a phone or laptop, because most consumer mic ports are too small, recessed, or off‑axis for the calibrator coupler. A small piece of adapter foam or tape can help, but the seal will not be perfect.
2. Activate the calibrator (94 dB SPL, 1 kHz). Run our meter for 30 seconds with A‑weighting, Fast.
3. Compute the offset. 94 minus our reading. Enter in settings.

The calibrator method is the most accurate by design — but consumer microphone form factors limit how reliably you can couple the calibrator to the actual microphone diaphragm. For most users, Method 1 (reference SLM in the same room) gives indistinguishable results with much less hassle.

Method 3 — Quiet‑baseline (last resort)

If you have neither a reference SLM nor a calibrator, you can anchor the meter against a known quiet level. This is the least accurate of the three methods but takes nothing more than your device and a quiet room.

1. Find a known quiet space. A quiet bedroom at night is typically 30 – 35 dBA; a sound‑treated home studio can be 25 dBA; a library reading room is usually 35 – 40 dBA.
2. Wait until the room is steady. Stop walking, hold the device still, ask anyone present to be quiet. Run our meter with Slow time weighting, A‑weighting, for 60 seconds.
3. Compare to the expected value. A typical quiet bedroom reads 30 – 35 dBA on a calibrated meter. If your meter reads 25 dBA in a space you'd describe as "quiet bedroom," your offset is approximately +5 to +8 dB. Conversely, if it reads 50 dBA, your device is over‑reading by 15 to 20 dB and the calibration offset is negative.

This method is rough — it assumes you correctly classify the room acoustically — but for someone with no reference equipment, it gets the calibration into the right ballpark and prevents readings from being silly. It is much better than no calibration at all.

Setting the calibration offset

Once you have a number, enter it in the meter:

1. Click the gear icon (top‑right of the meter).
2. Find the Calibration offset field.
3. Enter the offset, including the sign. Examples: +5 if your meter reads 5 dB lower than the reference; -3 if it reads 3 dB higher.
4. Save.

The offset persists in your browser's local storage for that origin and device, so it is remembered across sessions. It does not sync across devices — phones and laptops typically need different offsets, and even two phones of the same model can differ by a few decibels.

If you want to test multiple devices, store each device's offset separately. We have a roadmap item to add named per‑device profiles; for now you'll need to remember each device's number.

When to recalibrate

A few situations warrant re‑running the calibration:

• New device or new browser. The audio pipeline differs across Chrome / Firefox / Safari and across OS versions. Re‑calibrate after a major OS update if readings drift.
• New microphone. External USB or Bluetooth microphones have their own calibration. Bluetooth in particular is tricky because the encoded codec changes the perceived level non‑linearly.
• Significant temperature or humidity change. MEMS microphones drift with environment, but the effect is small (sub‑decibel) for typical indoor conditions.
• Disagreement with a known reference. If a measurement disagrees with a published reference by more than 5 dB on a steady source, recalibrate before treating the disagreement as real.

A monthly check against the same reference space (same fan in the same room) is a good habit if you use the meter for any decision that matters.

What you cannot calibrate around

Three limitations of consumer microphones are not fixable by setting a calibration offset.

Maximum SPL clipping

Most phone and laptop microphones clip — that is, their analog front end saturates and the digital signal is no longer proportional to the input — somewhere between 95 and 110 dB SPL. Above the clipping point, the meter under‑reports the actual level. There is no offset that fixes this; the information is gone before the meter ever sees it.

If you need to measure 100 dBA or louder reliably, you need a Class 2 sound‑level meter (IEC 61672‑1) or a dedicated dosimeter. Our browser meter is an excellent screening tool below the clipping range.

Frequency response

A perfect measurement microphone has a flat frequency response from 20 Hz to 20 kHz. Phone and laptop mics have peaks and dips of several decibels across that range — engineered for voice (200 Hz – 4 kHz) at the expense of low‑frequency rumble and high‑frequency air. The calibration offset is a single scalar; it cannot correct a non‑flat frequency response.

In practice this means our meter will be quite accurate on broadband mid‑range sources (traffic, conversation, HVAC) and somewhat less accurate on heavily low‑frequency or high‑frequency sources. C‑weighting amplifies low‑frequency error; A‑weighting suppresses it. When in doubt, A‑weight.

Time integration

The meter reads time‑weighted instantaneous values; calibration of a single value at a steady level does not guarantee accurate Leq integration over an hour, especially with variable noise. The tips page covers when this matters.

Verifying the calibration

After setting the offset, do one repeatability check:

1. Take three 30‑second readings of a steady source (same fan, same distance). Each should land within 1 dB of the others.
2. If readings drift more than 1 dB, the source isn't actually steady or your device's AGC is still active despite the getUserMedia flags. Try a different source (an HVAC duct from 1 m is usually the most reliable).
3. Cross‑check with a different reference if you have one. If two independent references agree within 2 dB, your calibration is trustworthy.

A calibrated browser meter, used within its limitations, is good enough for almost everything except formal compliance documentation. The guide, tips, and comparison chart all assume you've done this once — without calibration, every reading on this site is off by an unknown but probably non‑trivial amount.