What a sweep does
The system plays a sine that glides continuously from low to high while recording. Because only one frequency plays at any instant, the excitation can later be separated from the room’s response with precision: deconvolving the recording reconstructs the room’s impulse response (IR) — and frequency response, RT60 and early reflections are all computed from that single IR.
Why not pink noise and an FFT
A pink-noise FFT averages all frequencies over the measurement window, folding the room’s noise floor — voices, air conditioning, traffic — into the result. A sweep treats time as a resource: each frequency owns its own slice, and matched filtering converts that into roughly 10·log₁₀(T·B) of processing gain — about +41 dB for a 3-second, 180 Hz–4 kHz sweep. A truck passing outside ruins a clap measurement; a sweep usually survives it.
ESS’s party trick: separating distortion
With ordinary sweeps, nonlinear gear — and every loudspeaker is a little nonlinear — folds harmonics into the fundamental response. The exponential sweep lines the harmonics up on the impulse response’s time axis, arriving before the fundamental, where they can be cleanly windowed out. That is how Farina’s convention paper became the industry standard.
Choosing sweep parameters
Longer and wider sweeps buy SNR — at the cost of waiting longer and demanding more stability from the environment. Acouvero ships four presets: a 3-second phone quick check (180 Hz–4 kHz), full-band, a low/sub preset (15–300 Hz, which needs external output), and custom. At home, 3–6 seconds is plenty; subwoofer work needs a patient low-frequency sweep because the wavelengths are long.
How Acouvero implements it
A short chirp precedes the sweep as an acoustic time anchor, re-detected on every measurement. Recording is lossless Linear PCM — perceptual codecs are unfit for impulse work. Clipped sweeps are rejected outright, and unexpected sample rates trigger an explicit confidence warning instead of silent numbers. The greatest virtue of a measurement tool is preferring an error message over a confident wrong answer.