Performance under non-isotropic noise profiles

Characterize the performance of multiscale generalized sparse plane wave decomposition under noise profiles other than isotropic diffuse noise.

Background

The numerical evaluation models additive noise directly in the spherical-harmonic domain with statistically flat spherical spectrum, corresponding to isotropic diffuse noise. The reported experiments suggest that the proposed approach is empirically robust under this particular noise condition, but they do not establish its behavior under spatially colored, directional, correlated, or otherwise non-isotropic noise. Assessing such profiles is important because practical spherical microphone-array recordings can contain structured environmental and sensor noise that differs from the isotropic model.

The unresolved problem is to analyze robustness and recovery accuracy across different noise distributions and spatial spectra, not merely to repeat the existing isotropic-noise experiments.

References

Similarly, the proposed approach is empirically shown to be robust to isotropic diffuse noise for $\textrm{SNR}\leq{}20$ dB, however analyses of its performance under different noise profiles is also left for future work.

— Multiscale Generalized Sparse Plane Wave Decomposition in the Spherical Harmonic Domain  (2610.21425 - Hacihabiboglu, 15 Oct 2026) in Section Conclusions