Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Unified SVD-Modal Solution for Sparse Sound Field Reconstruction with Hybrid Spherical-Linear Microphone Arrays

Shunxi Xu, Thushara Abhayapala, Craig T. Jin

TL;DR

The paper addresses sparse sound-field reconstruction with hybrid SMA-LMA arrays, where SMA-only SH resolution is limited and direct SMA-LMA concatenation is fragile in reverberant environments. It introduces a data-driven SVD-modal framework that diagonalizes the transfer operator into a reduced-rank dictionary via H(f)=U(f) Sigma(f) V(f)^H, yielding an orthogonal modal basis. The approach generalizes SH for SMA-only configurations and provides complementary, stable modes for hybrid arrays, with modal analysis showing a frequency-dependent divergence from SH. In reverberant tests, the method reduces energy-map mismatch and angular error, outperforming SMA-only and direct concatenation, and achieving competitive performance with residue refinement within a principled, unified framework.

Abstract

We propose a data-driven sparse recovery framework for hybrid spherical linear microphone arrays using singular value decomposition (SVD) of the transfer operator. The SVD yields orthogonal microphone and field modes, reducing to spherical harmonics (SH) in the SMA-only case, while incorporating LMAs introduces complementary modes beyond SH. Modal analysis reveals consistent divergence from SH across frequency, confirming the improved spatial selectivity. Experiments in reverberant conditions show reduced energy-map mismatch and angular error across frequency, distance, and source count, outperforming SMA-only and direct concatenation. The results demonstrate that SVD-modal processing provides a principled and unified treatment of hybrid arrays for robust sparse sound-field reconstruction.

A Unified SVD-Modal Solution for Sparse Sound Field Reconstruction with Hybrid Spherical-Linear Microphone Arrays

TL;DR

The paper addresses sparse sound-field reconstruction with hybrid SMA-LMA arrays, where SMA-only SH resolution is limited and direct SMA-LMA concatenation is fragile in reverberant environments. It introduces a data-driven SVD-modal framework that diagonalizes the transfer operator into a reduced-rank dictionary via H(f)=U(f) Sigma(f) V(f)^H, yielding an orthogonal modal basis. The approach generalizes SH for SMA-only configurations and provides complementary, stable modes for hybrid arrays, with modal analysis showing a frequency-dependent divergence from SH. In reverberant tests, the method reduces energy-map mismatch and angular error, outperforming SMA-only and direct concatenation, and achieving competitive performance with residue refinement within a principled, unified framework.

Abstract

We propose a data-driven sparse recovery framework for hybrid spherical linear microphone arrays using singular value decomposition (SVD) of the transfer operator. The SVD yields orthogonal microphone and field modes, reducing to spherical harmonics (SH) in the SMA-only case, while incorporating LMAs introduces complementary modes beyond SH. Modal analysis reveals consistent divergence from SH across frequency, confirming the improved spatial selectivity. Experiments in reverberant conditions show reduced energy-map mismatch and angular error across frequency, distance, and source count, outperforming SMA-only and direct concatenation. The results demonstrate that SVD-modal processing provides a principled and unified treatment of hybrid arrays for robust sparse sound-field reconstruction.
Paper Structure (12 sections, 13 equations, 5 figures)

This paper contains 12 sections, 13 equations, 5 figures.

Table of Contents

  1. Introduction
  2. Background
  3. Sparse Plane-Wave Decomposition
  4. Combined SMA–LMA Processing
  5. Proposed Method: Unified Modal Solution via SVD
  6. Transfer Operator and Modal Decomposition
  7. Modal Analysis
  8. Evaluation
  9. Experiment Setup
  10. Sparse Recovery Metrics
  11. Robustness Evaluation
  12. Conclusion

Figures (5)

  • Figure 1: (a) Hybrid array geometry: a 64-element open SMA of radius 10 cm (blue) surrounded by four 8-element LMAs (red) with 4 cm spacing, symmetrically placed 0.5 m from the SMA center along the $x$- and $y$-axes; (b) directivity index jin2013design comparison showing that the hybrid array achieves improved spatial selectivity over the SMA-only configuration.
  • Figure 2: Modal analysis of the hybrid SMA–LMA array. (a) Mean principal angle between SH and SVD subspaces, showing frequency-dependent divergence. (b)–(d) Representative SVD modes at 1.5, 2.0, and 3.5 kHz, illustrating the frequency-dependent spatial patterns of the first 16 modes.
  • Figure 3: Sparse recovery in a reverberant room (RT60 = 0.3 s) with 10 sources. (a) Energy map mismatch and (b) angular error versus frequency. Modal solutions consistently reduce mismatch and achieve lower angular error compared with SMA-only and joint SR.
  • Figure 4: Energy mismatch across different source distances (a) 1.5m (b) 2.5m (c) 3.5m. Legend indicates the processing methods.
  • Figure 5: Angular error across different source distances (a) 1.5m (b) 2.5m (c) 3.5m. Legend indicates the processing methods.