Take three positions, log for at least twelve minutes in each so a compressor cycle is captured, and repeat the whole thing with the air handling off; the pair of numbers is the measurement.
Measure a room's noise floor in about twenty minutes
checked 2026-09-16responsible Room Tone Register (a named register, not a person)
What you need
- A meter with A-weighting and slow averaging, calibrated at the start and the end of the session with a 94 dB, 1 kHz calibrator. Uncalibrated readings are relative, not absolute.
- Or a phone with a sound level app, understood as indicative: the microphone and its housing are not flat, and the app cannot be more accurate than the hardware. Use it for comparing two rooms, not for quoting a figure.
- A note of the HVAC state, the time of day, and whether the building is occupied. A floor without those three is not reproducible.
The three positions
- Where the microphone will actually be, at the height it will actually be. This is the number that matters.
- The centre of the room. This is the number other people will quote back at you, and it is usually the quietest.
- Near the noisiest boundary — the wall shared with the plant room, the window on the street. This is the number that explains why position one failed.
The twelve-minute log
Twelve minutes is chosen to exceed the duty cycle of a domestic refrigerator and a small air handler. Take a reading every thirty seconds and write them down, or use the meter's Leq function over the whole window. What you are looking for is the shape: a flat trace means a steady floor, a stepped trace means a cycling source, and a stepped trace is the more common result.
Do not report the minimum. The minimum is the floor with the compressor off, and you cannot record in a room with the compressor off unless you can actually switch it off for the take.
Three errors that ruin the measurement
- Holding the meter. Your hand reflects and your body absorbs; hold it at arm's length or put it on a stand. This alone can move the reading by 2 dB.
- Measuring near a wall. Within about half a metre of a large surface the reflected energy adds to the direct, and the reading is no longer the free-field level.
- Quoting the meter's own accuracy figure as the uncertainty of the result. The meter's tolerance is one term; positioning, calibration drift and the room's own variation are usually larger.
Source log
- 01Class 1 sound calibrator, 94 dB at 1 kHz (NTi Audio)supports the calibration step and the 94 dB at 1 kHz reference used at the start and end of a session
- 02Hearing, UNSW Music Acousticssupports the relationship between frequency content and what a listener perceives as noise
- 031910.95 Occupational noise exposure (OSHA)supports the exposure framework that makes an A-weighted level the meaningful one to log
Each line points at one specific document. No line is a home page.
Leave this page Close this page when the numbers are written down. The decision they feed belongs to the noise floor bands.