The gap is almost entirely furniture and air absorption, both of which Sabine as usually applied ignores; the prediction is still useful for ranking options, just not for quoting a number.

Sabine predicted 0.42 s; the room measured 0.61 s

written 2026-07-14responsible Room Tone Register (a named register, not a person)

What was compared

An 88 m³ furnished room. Surfaces were measured and their areas tabulated, coefficients taken from published tables, and RT60 computed with Sabine. The room was then measured by the interrupted-noise method with a calibrated source and microphone, averaged over three source positions and three receiver positions.

The three reasons

1. The measured surfaces were not the acoustic surfaces. A bookshelf, a sofa and a rug present far more absorption than their plan area suggests, and they were not in the tabulation.
2. The room is not diffuse. Sabine assumes energy arriving equally from all directions; in a room this size, the low-frequency field is modal and the assumption fails worst where the gap is largest.
3. The source and receiver were close enough that the direct field was still significant at the receiver, which biases the measured decay upward.

What to do with a prediction that is 45% low

Use it to compare options, not to quote a figure. If the prediction says a given panel area takes the room from 0.42 s to 0.31 s, the real room will move by a similar proportion even though the absolute numbers differ. What the prediction cannot do is tell you the room is compliant with a specified target, because the error is larger than most targets. That is what measurement is for.

The gap is largest at 125 Hz and smallest at 4 kHz, which is the signature of an un-modelled absorber — furniture — rather than a method error. A method error would show up evenly across the bands.

Leave this page Close this page once you know whether you are comparing options or claiming compliance. Only measurement settles the second.