Noise Comparison: Which Room Is Quieter?

Which bedroom, which desk, which side of the building. These are comparison questions, and they are the ones a browser sound meter is actually good at — better than it is at the question it looks like it answers, which is how loud something is in absolute terms.

That is not modesty. It follows from where the error in an uncalibrated reading comes from, and it is worth setting out before you record anything, because it also tells you how to record.

Why a comparison beats a reading

A reading from a device microphone is an estimate carrying a wide band. When Kardous and Shaw compared ten sound measurement apps against a reference instrument, the closest sat about half a decibel from it and the two furthest were roughly thirteen decibels underKardous & Shaw. This site publishes a ±10 dB band for its own number and explains where that comes from under how accurate the reading is.

Now notice what kind of error that is. Your microphone's response, its housing, the browser's audio path, the gain the operating system applies — none of these change between two o'clock and five past. They are a fixed offset sitting under every reading that one device takes. When you subtract one of those readings from another, the offset goes with it.

So the difference between two places measured on one phone is a far better figure than either place measured on its own. Not perfect: what does not cancel is anything that actually changed. Where you stood. How you held it. Whether a lorry went past one of the two times. Those are real, and most of the advice below is about keeping them small.

Recording two places so the comparison means something

One device, both places. The whole argument above depends on it. Two phones give you two unknown offsets and no way to separate them from the answer.

The same height and the same orientation. Hold it the same way in both rooms — arm out, roughly chest height, microphone not covered by a hand or a case. A phone pressed against your body reads differently from one held clear of it, and that difference will land in your result looking like a property of the room.

Close together in time. Record both within a few minutes. Traffic, plant machinery and neighbours all follow the clock, and a comparison between eleven in the morning and seven in the evening is a comparison between two times as much as two places.

Ten seconds, or twenty where the noise moves. Ten settles a steady room. A corridor, a bar, a street-facing window — anywhere the sound arrives in bursts — wants twenty, so the average has enough to average.

Retake when something one-off happens. A door slam during a ten-second capture is not the room; it is an event that happened to be in your sample. Record it again. This is not cherry-picking as long as you apply the same rule to both places.

What the difference actually says

Sound energy doubles every 3 dB. That is the equal-energy principle NIOSH's own exchange rate rests on: three decibels more, half as long to reach the same totalNIOSH. Run it the other way and it tells you what a gap between two rooms amounts to.

The tool prints that multiple beside the decibel gap because the decibel gap alone under-sells itself. Ten decibels does not sound like much written down. Ten times the sound energy does.

This is energy, and only energy. Every competing comparison tool takes one more step and tells you a 10 dB gap "sounds about twice as loud". That step is a real result from psychoacoustics, and it is not one this site can cite: the sources it publishes from are health agencies writing about exposure, not about loudness perception, and the decibel chart already says that the scale describes energy rather than how loud something seems. So this page stops where its sources stop. If you want the honest short version: a 3 dB gap is small but real, a 10 dB gap is large, and the multiple tells you how large in the one currency that is actually being measured.

What the maxima say that the averages do not

The tool reports Leq — the energy average over the capture — and also the loudest moment in it. Read both.

A place can average two decibels lower and still be the one you would not sleep in, because its noise arrives in spikes while the other's sits at a steady hum. The average will tell you the first room is quieter. The maxima will tell you the first room is the one with the slammed door in it. Neither figure is wrong; they answer different questions, and for a bedroom the second question is usually the one that matters.

If the two averages are within a decibel, the tool declines to name a winner. That gap is smaller than what changes while you walk from one room to the other, and calling it would be dressing up a rounding difference as a result.

A and B do not have to be two places

The stronger use of this tool is one place, twice, either side of a change.

Close the window. Put the rug down. Turn the extractor fan off. Move the desk away from the wall. Record before, make the change, record after, and read the gap. Everything the previous section asks you to hold still is held still for free, because the microphone has not moved and neither have you — the only thing that differs between the two captures is the thing you changed.

That is as close as a browser page gets to a controlled measurement, and it answers the question people usually have underneath which room is quieter, which is will this actually help. A secondary-glazing quote, a door seal, a mat under a washing machine: each of them is a claim about a number of decibels, and you can now check the claim in your own room instead of taking it.

Two cautions. Make the change quickly, so the outside world has not moved on while you were doing it; and if the change is one you cannot undo, record the before twice and check the two agree, because a single before-capture with a lorry in it will make any change look like an improvement.

Expect small numbers. An improvement you can measure is real, and it is usually not the transformation the product page implied. The tool will tell you which of the two you bought, which is worth knowing before the second window rather than after it.

When this is not the tool you want

For how much noise costs you over a day, a comparison says nothing. A level is not a dose, and the noise dose calculator is the page that turns levels and durations into one figure.

For a neighbour, a landlord or a council, one comparison is not a record. What makes a noise complaint credible is a pattern across many nights, with times written down as they happen — set out under documenting neighbour noise. Use this tool to establish that the front bedroom is far worse than the back one; use a log to establish that it happens every Friday.

For anything with a compliance consequence, neither this nor any browser page is the instrument. The figures here are uncalibrated estimates, the same kind of number the online decibel meter on this site reports, and they are not IEC 61672 measurements.

What this page does not claim

It does not claim either room's absolute level is right. It claims the difference between them is worth more than either number — which is a smaller claim, and one that survives the microphone being unknown.