Robust Room Impulse Response Measurement Using Perfect Sequences for Legendre Nonlinear Filters

Abstract

The paper proposes a novel approach for measuring the room impulse response that is robust toward the nonlinearities affecting the power amplifier or the loudspeaker. The approach is implemented by modeling the acoustic path as a Legendre nonlinear filter and by measuring the first-order kernel using perfect periodic sequences and the cross-correlation method. Perfect sequences for Legendre filters are periodic sequences that guarantee the orthogonality of the Legendre basis functions over a period. They ensure the robustness of the first kernel measurement toward nonlinear distortions. The paper also explains how perfect periodic sequences for Legendre filters that are suitable for room impulse response identification can be developed. Experimental results involving both simulated and real environments illustrate the effectiveness and the robustness of the proposed approach

A Novel Development of Acoustic SLAM

“…There are multiple methods for gathering a RIR: Impulse, Chirp, Linear/Logarithmic sweep, Maximum-Length Signal (MLS) and Inverse Repeated Signal (IRS). These have been compared and tested in many papers [6], [7] and [8]. However, the testing parameters from these works have never explored the situation where the receiver, source or environment is moving.…”

Robust Room Impulse Response Measurement Using Perfect Periodic Sequences for Wiener Nonlinear Filters

“…In [35,36], a novel approach for RIR measurement robust towards nonlinearities with memory was proposed. In this approach, the entire measurement system, composed of the power amplifier, the loudspeaker, the acoustic path, and the microphone, is directly modeled as a Legendre nonlinear (LN) filter.…”

“…Indeed, with a uniform distribution 42% of the samples have magnitude larger than the standard deviation, while with a Gaussian distribution the value reduces to 32% of the samples. In this paper, we extend the approach of [36] and consider the robust RIR measurement using PPSs for WN filters. This measurement was first proposed by the authors in the conference paper [39] and it is fully detailed in the current manuscript.…”

“…Thus, in this paper it is fully detailed how PPSs that are suitable for RIR identification with period proportionate to the filter memory length can be developed. Their derivation is nontrivial and is different from that of the PPSs for LN filters of [36]. A detailed analysis of RIR measurement under different non-ideal conditions, e.g., memory underestimation, order of nonlinearity underestimation, and noise effect, is also performed.…”

“…A detailed analysis of RIR measurement under different non-ideal conditions, e.g., memory underestimation, order of nonlinearity underestimation, and noise effect, is also performed. The analysis of the noise effect provides novel results that were not discussed in [36]. The paper also provides novel experimental results that compare the proposed approach with competing approaches, i.e., MLSs, exponential sweeps, PPSs for LN filters.…”