The 3 dB rule and decibel averages

Why 60 and 70 average to about 67.4 dB

Suppose a sound stays at 60 dB for one interval and 70 dB for an equally long interval. These are hypothetical levels for a calculation, not readings collected by Decibel Reader.

The arithmetic mean is 65 dB. But the 70 dB interval carries ten times as much energy as the 60 dB interval, so an energy average falls closer to the higher level.

Convert both levels into relative energy, average them, then convert back: 10 × log10((10⁶ + 10⁷) / 2) ≈ 67.4 dB.

Duration matters too. One second at a higher level contributes less energy than a minute at that same level. Decibel Reader weights its valid samples by their sampled durations. A paused or interrupted browser leaves a gap; it does not add silent samples.

That is why a brief louder event can raise Average even if most of the chart looks quiet. Max and Peak Hold tell you about the highest displayed level instead. The guide to browser readings explains when each is useful.

What doubles at 3 dB?

A decibel difference expresses a ratio. For energy, the calculation is 10 × log10(energy ratio). Doubling energy gives about 3.01 dB. Four times the energy gives about 6.02 dB.

Amplitude uses 20 × log10(amplitude ratio). Doubling amplitude gives about 6.02 dB; halving it gives a change of −6.02 dB, under the same conditions. Amplitude and energy are different quantities, which is why both formulas are needed.

These calculations describe a change, not its reference. dBFS measures a digital signal relative to full scale. Sound-pressure level uses an acoustic reference. A weighting applies a frequency filter. The letters beside a reading tell you which kind of measurement you have.

Estimated dB is a device reading

Decibel Reader calculates the microphone signal’s RMS level as 20 × log10(RMS), in dBFS, then adds a display offset. The default offset is +94 dB and is explicitly uncalibrated. It makes a convenient display scale; it does not establish the microphone’s sensitivity to sound pressure.

The browser also has no A-weighting filter. Entering a reading from an A-weighted reference meter can align one condition, but it cannot make the browser respond identically to a different mix of frequencies.

A display of 85 Estimated dB therefore cannot be inserted as 85 dBA in the table below. The colored ruler and an exported report do not change the measurement’s status. The accuracy page describes the device and processing limits.

The NIOSH workplace reference uses dBA

NIOSH’s occupational recommendation starts at 85 dBA over eight hours. Its 3 dB exchange rate halves the corresponding daily duration for each 3 dBA increase. It requires A-weighted exposure measurements.

Those durations describe a workplace exposure reference, not a personal safe-listening countdown. If a day includes several levels, the time spent at each contributes to the total exposure. Looking up only the loudest reading misses that information; the NIOSH source explains daily noise dose and personal dosimeters.

For a room comparison, look for a repeatable change

When comparing fan settings or a window open and closed, keep the same microphone, position, and gain settings. Measure each condition for a similar period and repeat the comparison. A consistent difference is more informative than one isolated maximum.

Use the difference to describe what changed at that position. An occupational table cannot answer whether a bedroom suits sleep, whether an infant’s surroundings are appropriate, or what caused a hearing symptom. A short browser reading cannot establish a legal noise limit either.

Save the setup details

Save the estimated average alongside the duration, device position, and condition you compared. Keep the report’s offset and accuracy status attached. That gives you enough information to repeat the check.

For workplace exposure or compliance, use an appropriate calibrated instrument and measurement procedure. The 3 dB arithmetic remains useful, but it cannot turn an uncertain browser estimate into the A-weighted measurement those decisions need.