dBA vs dBC vs dBZ: Which Weighting to Use

The letter after dB is not a unit. It names a filter applied to the signal before the level is worked out, and choosing a different letter can change the number for the same sound by a lot — most of all for sounds with a lot of low frequency in them.

That is the whole subject. Everything else is which filter answers which question.

What each letter does

A discounts the very low and very high frequencies, keeping the middle of the range where human hearing is most sensitive. It is a rough model of the ear, which is why it tracks the risk to hearing better than the raw energy in the air does.

C is much flatter. It keeps far more of the low end, so it reports more of what a large speaker or an engine is actually producing and less of what your ear is making of it.

Z is no weighting at all — the level as recorded, flat across the range. Research reports it as "un-weighted", which is the clearer name for itKardous & Shaw.

On a sound with most of its energy in the middle of the range, the three land close together. On a bass-heavy sound they do not, and the gap can run to double figures. That is why quoting a decibel figure without saying which weighting it uses is close to quoting no figure at all.

Why every health figure is in dBA

Because the question those figures answer is about hearing, not about air.

The exposure limit NIOSH publishes is 85 decibels, A-weighted, as an eight-hour time-weighted averageNIOSH. When NIOSH names the settings it recommends for a sound level meter, the frequency weighting is A, alongside Slow response and a 3 dB exchange rateNIOSH.

Every occupational noise limit, every hearing-safety table, and every "how long is this safe" figure on this site or anywhere else is downstream of that choice. A number in dBC cannot be compared against them without a conversion that depends on the spectrum of the specific sound — which is to say, without information you do not have.

So the rule is simple, and it is the one worth taking away: if the question is about hearing, the answer is in dBA, and a figure in anything else has to be converted before it means anything.

When C or Z is the right one

C, for low-frequency complaints. The classic case is a neighbour's subwoofer, a nightclub through a wall, or a generator. A-weighting deliberately discounts the frequencies those produce most of, so an A-weighted figure can look unremarkable while the room is unlivable. A C-weighted figure keeps more of that energy, and the difference between a C and an A reading of the same sound is itself informative: a large gap means the sound is dominated by low frequencies.

C, for peaks. Instruments generally offer a C-weighted peak measure for impulsive sounds, on the grounds that A-weighting is a model of how sustained sound damages hearing rather than how a single impulse does.

Z, when you want the signal rather than a model. Anything where the question is what the equipment or the room is producing — troubleshooting a machine, checking a speaker, comparing two recordings — wants the un-weighted level, because a weighting curve is a deliberate distortion added for a purpose you do not have.

None of those is the question most people arrive with, which is why A is the default nearly everywhere.

Converting between them, and why you mostly cannot

There is no fixed offset between dBA and dBC. The difference depends on what frequencies the sound actually contains, so the same conversion that is right for a hairdryer is wrong for a bass bin, and any single number offered as "the" conversion is an average over sounds you are not listening to.

The one thing the pair does give you for free is a diagnostic. Take both readings of the same steady sound and subtract: a small difference means the energy sits in the middle of the range where the two curves broadly agree, and a large one means it is concentrated at the low end that A discards. You do not need to know the exact spectrum to act on that — it tells you whether you have a low-frequency problem, which is usually the question behind the question.

That trick needs an instrument that offers both weightings on the same sound at the same moment. It is not something two separate readings from two separate devices can give you, and it is not something this site can do at all.

Weighting is not only a display setting

Here is a detail that is easy to miss and worth knowing before trusting any app or page in a weighting other than the one it was built for.

When Kardous and Shaw compared ten iOS sound apps against a reference, they reported A-weighted and un-weighted results separately — and the apps that did best were not the same set. Three came within ±2 dBA on A-weighted levels; a partly different three came within ±2 dB un-weightedKardous & Shaw.

Whatever is happening inside those apps, the weighting is not a cosmetic filter applied at the end of a correct measurement. It interacts with everything upstream of it. An app that is good in one weighting has not thereby earned your trust in another.

Why this site defaults to dBA, and what the other two are for

The meter reads on all three curves. It starts on A, and A is the only one it will reason about.

Because every figure it points at is in dBA. The exposure limits, the safe durations, the dose arithmetic and the everyday-sound tables on this site are all A-weighted. A switch that let the headline number silently become something else would let a reader compare a C-weighted figure against an A-weighted limit, and that comparison is simply wrong. So the switch is not silent: on C or Z the meter changes the unit beside every number, and the dose, the zone bands and the NIOSH lines withdraw until you switch back. A reading with no threshold attached is the honest form of a number no threshold describes.

Because the comparison is the useful part. Holding one signal and moving between A and C tells you something a single number cannot: how much of what you are hearing is bass the A curve is discarding. A neighbour's music through a wall and a fan on a desk can read the same in dBA and separate by twenty decibels or more in dBC. That gap is the answer to why does this feel louder than it measures, and it is the reason to offer the curves at all.

Because neither of the other two earns a threshold here. The low-frequency case is exactly where a phone microphone is least trustworthy, and the impulse case C is used for needs headroom this hardware does not have. So the curves are offered as a comparison, not as a second opinion on whether a room is safe. How wide the band around this site's number actually is is set out under online meter accuracy.

If you need a weighting this page does not offer

You need an instrument. A hardware sound level meter offers the weightings as first-class settings, with a microphone whose response is known across the range each of them cares about — which is the part that cannot be added in software.

If the situation is a low-frequency complaint, that is worth knowing early. An A-weighted reading of a bass problem understates it by design, and taking a series of A-weighted browser readings to a landlord is bringing the wrong evidence to an argument you might otherwise win.

The short version

Use dBA for anything about hearing, health, or occupational noise limits, and expect every published figure to be in it. Use C when the complaint is low frequency or the sound is impulsive. Use Z when you want the signal itself rather than a model of an ear. And whichever you use, write the letter down — a decibel figure with no weighting attached is a number that cannot be checked.

The A-weighted online decibel meter on this site starts on dBA and offers the other two beside it, labelling every number with the curve it came from. What it will not do is carry an A-weighted limit over to a reading that is not A-weighted: choose C or Z and the safe-exposure figures withdraw rather than being restated for a curve nobody defined them on.