Swept Sine (Log Sweep)

The logarithmic sine sweep, popularized by Angelo Farina in 2000, has become the gold standard for impulse response measurement. It works by exciting the system under test with a signal that starts at a low frequency (typically 20 Hz) and sweeps upward to a high frequency (20 kHz) over a configurable duration (2-20 seconds). The logarithmic rate means each octave receives equal sweep time. After capturing the system output, the impulse response is extracted by convolving with the inverse filter of the sweep signal. This inverse filter is computed in the frequency domain as the time-reversed sweep with amplitude compensation. The process is mathematically equivalent to matched filtering, which maximizes the SNR of the extracted IR. The key advantage of log sweeps over other excitation signals is harmonic distortion rejection. Because harmonic distortion products have higher frequencies than the fundamental, they appear at earlier times in the deconvolution result (before the main IR peak). This temporal separation allows clean extraction of the linear impulse response, uncontaminated by system nonlinearity. MLS sequences, by contrast, fold distortion into the measured IR. Sweep duration controls the trade-off between measurement time and SNR. Longer sweeps inject more total energy into the system, improving SNR by approximately 3 dB per doubling of sweep length. A 5-second sweep typically provides 30-40 dB of processing gain over a simple impulse, while a 20-second sweep can achieve 50+ dB. For RT60 measurements requiring high INR, longer sweeps are essential. Practical considerations: the sweep must be played at a level that exercises the system without clipping. The measurement environment must remain stationary during the sweep — any movement (doors, people) will contaminate the IR. Multiple sweeps can be averaged to further improve SNR. SonaVyx generates log sine sweeps using the Farina method and performs deconvolution via WASM for in-browser impulse response extraction.