📝

Noise Comparison Tool

Capture sample A, change something, then capture sample B with the same microphone and gain — and compare them side by side: an overlaid octave-band spectrum, the broadband dB difference, and Leq / Lmax / L90 for each with deltas. Perfect for before/after checks: a fan on vs off, a window open vs closed, or acoustic treatment installed.

The A−B difference is the trustworthy number; the absolute levels are not. A browser microphone is uncalibrated, so each capture’s Leq is a relative dBFS estimate, not a certified dB SPL and not a substitute for a Type 1/2 sound-level meter or any legal/compliance evidence. But because A and B are captured with the same mic and gain, the unknown calibration offset cancels in A−B — so the difference (and the spectral shape) is genuinely meaningful. Auto-gain and noise suppression are requested off. Nothing is recorded or uploaded.

Idle — press Start, allow the microphone, then capture A and B.
Microphone idle.

Tip: keep the microphone in the same spot and don’t touch the gain between captures — that’s what makes the comparison valid.

Overlaid octave-band spectrum — A vs B (dBFS per band)
No captures yet. Press Start, then Capture A and Capture B.
Comparison of metrics for capture A, capture B, and the B minus A difference
Metric A (before) B (after) Δ (B−A)
Leq (average)
Lmax (loudest)
L90 (background)
Peak sample
Clipsn/a
Optional: calibrate for an SPL estimate (shared across all noise tools)

Calibration only affects the absolute SPL estimate — the A−B difference never needs it. Read the level on a real sound-level meter (or calibrated phone app) at the same time, enter both numbers, and the offset is stored once for every noise tool on this site.

Not calibrated.

How It Works

The tool records two short snapshots of your microphone’s level and spectrum — A (your “before”) and B (your “after”). For each capture it accumulates power over the whole window and reports Leq (the equivalent continuous level — the steady level carrying the same energy as the fluctuating sound), Lmax (the loudest moment) and L90 (the level exceeded 90% of the time, the classic statistic for the steady background under intermittent events). It also splits each capture into ten octave bands from 31.5 Hz to 16 kHz and overlays them so you can see where in the spectrum the noise changed.

Why the A−B difference is trustworthy even without calibration

A browser cannot know your microphone’s real-world sensitivity, so every absolute number it shows is in dBFS (decibels relative to digital full scale) and carries an unknown, fixed calibration offset. The key insight: when you measure A and B with the same microphone, in the same position, at the same gain, that unknown offset is identical in both — so when you subtract (B − A) it cancels exactly. The difference in decibels, and the difference in spectral shape, are therefore genuinely meaningful: if B reads 6 dB lower than A, your change really did cut the level by about 6 dB, regardless of the fact that neither absolute figure is a certified SPL.

What stays uncalibrated

The single-capture numbers themselves (“A is −38 dBFS”) are not a real environmental decibel reading. You can optionally enter a calibration offset — measured once against a true sound-level meter — to display a rough SPL estimate; that offset is saved under a shared key (fd-noise-cal) so every noise tool on the site uses the same calibration. Even then, treat the SPL figure as an estimate. And remember consumer mics roll off at the frequency extremes and generally cannot capture true infrasound (below ~20 Hz) or deep sub-bass, so the lowest band is approximate.

Frequently Asked Questions

Why is the A−B difference reliable if the mic isn’t calibrated?
Because both captures share the same unknown calibration offset. When you subtract B minus A, that offset cancels exactly, leaving a real difference in decibels. So even though “A is −38 dBFS” is not a certified SPL, “B is 6 dB quieter than A” is trustworthy — as long as you didn’t move the mic or change the gain between captures.
Can I use the absolute numbers as real decibel (dB SPL) readings?
No. A browser microphone is uncalibrated and the absolute values are relative dBFS, not certified dB SPL. They are not a substitute for a Type 1/2 sound-level meter and are not valid as legal, complaint, or compliance evidence. You can enter a calibration offset for a rough SPL estimate, but it remains an estimate. The comparison (A vs B) is the part you can rely on.
What do Leq, Lmax and L90 mean?
Leq is the equivalent continuous level — the steady level carrying the same total energy as the varying sound over the capture, the standard summary of overall loudness. Lmax is the loudest moment. L90 is the level exceeded 90% of the time, which represents the steady background underneath intermittent events. Here they are all in relative dBFS, so use the deltas between A and B.
How should I run a fair before/after test?
Place the microphone, press Start, and capture A. Then make exactly one change — turn the fan off, close the window, add a panel — without moving the mic or touching the input gain, and capture B. Keep the capture length the same. The whole point is that everything except the thing you’re testing stays identical, so the difference reflects only your change.
What does the exported report contain, and is it legal evidence?
It is a plain-text local download with each capture’s Leq, Lmax, L90 and peak, the octave-band levels, and the B−A deltas, clearly labelled as indicative only. It is not legal, complaint, or compliance evidence — for that you need a calibrated sound-level meter and a qualified measurement. Nothing leaves your device; the report is generated in your browser.
Is any audio recorded or uploaded?
No. The microphone signal is analyzed in real time to compute level and spectrum statistics only — it is never recorded, saved, or transmitted. Auto-gain, noise suppression and echo cancellation are requested off so the readings aren’t altered. The microphone is released when you press Stop or close the tab.