How loud is it?

The decibel is a ratio, not a unit of loudness

A decibel expresses a logarithmic ratio between two quantities. For sound pressure level, the reference is 20 µPa — roughly the quietest pressure variation a healthy young ear can detect. The formula is:

L_p = 20 · log₁₀(p / p_ref)

That logarithm is the source of every weird thing about dB arithmetic. Adding two equal sources adds only 3 dB. Doubling the perceived loudness takes about 10 dB. Halving the distance from a point source adds 6 dB. None of these are intuitive, none of them follow the linear math you use for everything else, and most online “dB calculators” quietly fudge these to look simpler.

The other awkward thing: dB ratios are dimensionless. A “120 dB SPL” tells you the pressure ratio. A “120 dB attenuation” tells you a different ratio about an amplifier or a wall. These look identical and mean different things. Always read the suffix: dB SPL, dBA, dBu, dBV, dB FS are all different scales that share a unit name.

The 3 dB rule, the 6 dB rule, and the 10 dB rule

Three rules cover ~95% of practical dB arithmetic:

• +3 dB — doubled power. Two equal incoherent sources sum to 3 dB above either one. This is why a 60 dB AC unit plus a 60 dB fridge equals 63 dB, not 120. Equivalent to: power has doubled.
• +6 dB — doubled pressure / halved distance. Halving the distance from a point source raises the level 6 dB. A speaker at 1 m measured 80 dB will read 86 dB at 0.5 m and 74 dB at 2 m. This is the inverse-square law in dB form.
• +10 dB — doubled perceived loudness. The human auditory system maps roughly 10 dB onto a doubling of subjective volume. So a 70 dB conversation sounds about twice as loud as 60 dB, even though the actual sound pressure is only ~3× greater.

If two sources differ by more than ~10 dB, the louder one wins almost completely — the quieter source contributes less than 0.5 dB to the total. This is why an HVAC unit at 60 dB swamps a refrigerator at 45 dB; you might as well not have the fridge in the room as far as the noise floor is concerned.

A useful check: when this calculator combines two equal sources, the answer should be exactly 3.01 dB more than either input. If it’s not, something has gone wrong (or you used a non-logarithmic calculator).

A-weighting, C-weighting, and the bass-hides-here problem

An unweighted dB SPL measurement treats every frequency equally. Human hearing does not. We’re much less sensitive to low frequencies (especially below 100 Hz) and to very high frequencies (above ~10 kHz) than to the speech band around 1–4 kHz.

Weighting curves correct for this:

• A-weighting (dBA) applies a filter that approximates the equal-loudness contour at moderate listening levels. This is what regulatory limits, OSHA, and most product specs use. A 60 dBA reading understates the bass content of a measurement.
• C-weighting (dBC) is much flatter and includes most of the audible range. C−A (the difference between dBC and dBA on the same source) tells you how much low-frequency energy is present. A train rumble might read 70 dBA but 90 dBC. A flute might read 70 dBA and 71 dBC.
• Z-weighting is no weighting at all (Zero-curve). Used in research, rarely on product spec sheets.

The honest summary: dBA hides bass. A 60 dBA HVAC unit and a 60 dBA conversation are not the same sleep-disturber. The HVAC has 30–40 dB more energy at 50 Hz than the conversation does, but A-weighting filters that out before the meter computes the “single number.”

If you’re measuring something low-frequency-dominated — subwoofers, traffic rumble, refrigerator compressors, footstep noise — use dBC or unweighted dB, or look at the spectrum directly. Reporting only dBA in those situations gives you a number that’s technically correct and practically misleading.

About the Measure mode — and why we won’t pretend it’s an SPL meter

The Measure mode reads your device’s microphone in real time and reports the level using the standard SPL-meter integrations: instantaneous, 1 s Fast, and 10 s L_eq. With a flick of the toggle, it applies an IEC 61672 A-weighting filter (6-biquad analog approximation, accurate to within Class 2 tolerance from 100 Hz to 10 kHz). It runs entirely in your browser; no audio leaves your device.

What it cannot do, by design, is tell you the absolute sound-pressure level without your help. Browsers expose only the digital level of the audio stream (dBFS, 0 to −∞) — not pascals. Every microphone has a different sensitivity, every operating system applies a different gain stage, and there is no public API to query either. Until you tell the calculator what your current reading actually is in dB SPL (read from a Class 2 SLM, the NIOSH iOS app, or a reference of known level), it can only show relative values.

That is the honest shape of the trade. Uncalibrated, the readings are dBFS — useful for "is this getting louder?" or "is the AC unit actually 5 dB quieter on Eco?" Once calibrated against a reference you trust, the same readings become dB SPL (or dBA, if you switched to A-weighting first). The offset persists locally so each visit doesn’t start from scratch. Re-calibrate any time the reference changes.

If you want a number a regulator will accept, you still need an actual Class 1 or Class 2 SLM. The Measure mode is for sense-making, A/B comparisons, and feeding plausible numbers into the rest of this calculator (combine, distance, reference). It is not a substitute for measurement that has legal weight.

Distance attenuation and the limits of the inverse-square law

The free-field inverse-square law (6 dB drop per doubling) holds for a point source radiating into open air with no reflections. Real environments rarely meet that condition. The corrections:

• Indoor reverberant fields: beyond a critical distance, the level barely changes — reflections from walls and ceiling create a roughly uniform sound field. In a typical untreated bedroom, you reach the reverberant field around 1–2 m from a speaker. Walking further away doesn’t help.
• Line sources (roads, trains, conveyors): level drops only 3 dB per doubling of distance, because as you move away, the visible “length” of the line decreases by half but each surviving segment is now further away — the geometry partially compensates. Same reasoning for area sources (a noisy roof) where falloff is even slower.
• Atmospheric absorption: high frequencies (>2 kHz) lose extra energy to air over hundreds of metres. A train horn 500 m away is geometrically attenuated and spectrally darker than the same horn at 5 m.
• Ground & foliage absorption: grass, soil, and dense vegetation can add 3–10 dB of attenuation at distance, but only for the higher frequencies. Bass passes through a tree line almost unchanged.

This calculator’s distance mode covers only the geometric falloff. For anything beyond ~50 m outdoors, the real attenuation is greater at high frequencies and identical at low frequencies. That’s why a distant freeway sounds like a low rumble, not a hiss — the highs got absorbed by the air; the lows didn’t.

For accurate environmental noise modelling, use ISO 9613-2 (outdoor sound propagation) or commercial tools like SoundPLAN. The inverse-square approximation is good enough for the “will my neighbour hear this?” class of question; not good enough for a noise study a regulator will read.

References & standards

• IEC 61672-1:2013 — Electroacoustics — Sound level meters. Defines A, C, and Z weighting curves and meter performance classes. The reference standard for any weighted dB measurement.
• ISO 1996-1:2016 — Description, measurement and assessment of environmental noise. Methodology for outdoor noise measurement and the time-averaged levels (L_Aeq, L_night) that environmental regulators use.
• ISO 9613-2:1996 — Attenuation of sound during propagation outdoors. The full method for outdoor distance modelling including atmospheric, ground, and barrier effects.
• OSHA 1910.95 — US occupational noise exposure: 8 hour PEL of 90 dBA, action level 85 dBA. NIOSH recommends 85 dBA as the actual hearing-damage threshold.
• EU Directive 2003/10/EC — European workplace noise directive: 8 hour exposure limit 87 dBA, action threshold 80 dBA, hearing protection mandatory above 85 dBA.
• WHO Environmental Noise Guidelines (2018) — recommends night-time outdoor levels below 40 dB L_night to prevent sleep disturbance from road traffic.