MLS (Maximum Length Sequence)

MLS was the dominant impulse response measurement technique before log sine sweeps gained popularity in the 2000s. An MLS of order n consists of 2ⁿ - 1 binary values (+1 or -1) that appear random but are deterministic, repeating exactly when the LFSR cycles. Common orders are 15 (32,767 samples = 0.68s at 48 kHz) and 16 (65,535 samples = 1.37s). The defining mathematical property of MLS is its circular autocorrelation: when cross-correlated with itself, the result is a near-perfect impulse (a Kronecker delta). This means that if you play an MLS signal through a system and circularly cross-correlate the output with the original MLS, you recover the impulse response. The processing gain equals approximately 10 × log₁₀(L) dB, where L is the sequence length. MLS advantages include deterministic behavior (exactly repeatable), binary values (insensitive to DAC nonlinearity when using +1/-1), and fast computation via the Hadamard transform. MLS measurements can achieve high SNR: an order-16 MLS provides about 48 dB of processing gain. However, MLS has significant disadvantages compared to log sweeps. It is sensitive to time variance: any change in the system during measurement (temperature drift, audience movement, mechanical changes) corrupts the result. MLS cannot separate harmonic distortion from the linear response — nonlinear distortion folds into the measured IR as noise, potentially obscuring late decay and biasing RT60 measurements. MLS is also sensitive to clock synchronization between playback and capture devices. For these reasons, modern best practice prefers log sine sweeps for most acoustic measurements. MLS remains relevant in some embedded systems, for very fast repeat measurements, and in educational contexts. SonaVyx supports both MLS and log sweep measurement methods, with MLS deconvolution via circular cross-correlation.