Is a 51 dB unit twice as loud as a 25 dB one?

No, it is much more than twice. Decibels are logarithmic. Every ten decibels is roughly a doubling of how loud something seems, so twenty six decibels is somewhere between five and six perceived doublings. That is the difference between a machine you can sleep through and one you cannot.

Why does this site refuse to sort some units by noise?

Because a figure published as a minimum cannot be ranked against one published as a maximum without inventing a comparison. Sorting by noise here uses only figures the maker published as a maximum. Everything else still shows, with the setting named in words, and simply does not take part in that ordering.

What is the difference between sound power and sound pressure?

Sound power is what the machine emits, a property of the machine alone. Sound pressure is what a microphone reads at a stated distance in a stated room, so it depends on where you stand and what the walls are made of. They are different quantities with different values, both written in decibels, and a specification that does not say which one it is quoting cannot be compared with one that does.

Does a quiet unit cool less?

Not necessarily, but a given unit is quieter on a lower fan speed and moves less air at that speed. That is why the setting attached to the figure matters so much. A maker quoting its lowest speed is telling you about a mode you may never actually use on a hot night.

The decibel figure on the box is not a number you can compare

Look at any air conditioner listing and there is a decibel figure. It is usually the only number on the page about how the thing will feel to live with, and it is presented exactly like the capacity and the price, as though it were the same kind of fact.

It is not. Two of those figures side by side are very often not measuring the same thing, and nothing on the page tells you.

What the makers actually publish

This site holds eleven units, every figure copied from the manufacturer’s own specification page. All eleven publish a decibel number. Here is what those numbers turn out to be.

Eight are a maximum: the unit at its loudest. Three portable air conditioners quote 51, one quotes 54, and three wall mounted split units quote 42, 42 and 44.

Two are a minimum. Both are pedestal fans, and both are published as “from” a figure: 25 and 29. That is the machine on its lowest speed, which is the quietest it is capable of being.

One is neither. It is a bare 65 with no setting attached at all.

Now put those in one column and sort them, which is what a comparison table normally does. The fans come top. A reader concludes that the fan is half as loud as the air conditioner, when what they have actually compared is one machine at its quietest against another at full tilt. The fan may well be quieter in practice. The published figures do not demonstrate it.

Why the gap is bigger than it looks

Decibels are logarithmic, which means the arithmetic instinct is wrong in a direction that matters.

Roughly every ten decibels is a doubling of perceived loudness. So 25 and 51 are not “about twice”, they are five or six doublings apart. On the other hand 51 and 54 look close and are close, near the edge of what most people would confidently pick out in a blind comparison.

This is why the ranking matters more than the differences. Sorting a list by noise is useful. Reading a three decibel gap as meaningful is not.

The second problem nobody mentions

Even among the honest maximums, there is a question underneath: sound power or sound pressure.

Sound power is a property of the machine, the total acoustic energy it puts out. Sound pressure is what you would measure standing somewhere specific, and it falls away with distance and changes with the room.

Both are quoted in decibels. For the same machine they are different numbers. A specification that says “51 dB” without saying which quantity it is, and at what distance, has given you a number you cannot line up against a rival’s number with any confidence.

Most domestic listings do not say. This site records what the maker published and does not guess at the rest, which means some of these figures are less comparable than they look even after the maximum and minimum problem is dealt with.

What this site does about it

Every record carries the basis alongside the figure, and the schema rejects a decibel number that arrives without one. The unit database shows it in words rather than a footnote: “51 dB at its loudest”, “25 dB at its quietest”, “65 dB, setting not stated”.

Sorting by noise uses only the stated maximums. The rest are still listed and still visible, they simply do not join an ordering they cannot honestly join. The same rule already applied to a missing figure: an unpublished number does not pass a filter or take part in a sort, because absent is not zero.

That does mean the sort here returns fewer units than a table that just ranks everything. That is the point of it.

What to do when you are buying

Ask for the maximum. If the listing quotes “from”, the number you care about is at the other end of the range and the maker has chosen not to lead with it.

Treat gaps under about three decibels as noise in the data rather than a difference. Treat ten as real.

And be careful with the sleep or night mode figure. It is usually genuine, and it is usually measured at a fan speed that moves considerably less air. On the night you actually need the machine, you may not be running it in the mode the brochure was describing.

The figures in this piece were computed by this site from daily weather readings it stores and processes. Underlying readings from the Open-Meteo historical archive, licensed CC BY 4.0.