Mind the Microphone Gap: Benchmarking Array Upsampling Strategies for Latent Acoustic Mapping

2026-07-27

AI Summary

This study addresses the significant performance degradation of Latent Acoustic Mapping (LAM) under sparse four-channel microphone arrays, primarily caused by low-resolution cross-spectral matrices. The authors systematically evaluate various upsampling strategies—including lightweight convolutional networks, iterative back-projection, physics-informed neural networks, and generative adversarial approaches—and investigate their impact on spatial structure preservation when integrated with LAM either jointly or in a staged training paradigm. Their comprehensive comparison reveals, for the first time, that full-resolution LAM achieves optimal performance, that independently trained lightweight upsampling models yield the best results among alternatives, and that representation alignment between the upsampler and LAM is more critical than merely increasing model complexity. These findings offer new insights for self-supervised acoustic mapping in low-channel-count scenarios.

Abstract

Latent Acoustic Mapping (LAM) is a self-supervised learning method that generates high-resolution spherical acoustic maps from multichannel recordings without labelled data, matching supervised baselines on direction-of-arrival benchmarks. However, LAM degrades significantly with sparse 4-channel arrays, as the low-resolution cross-spectral matrix captures far less spatial information than the 32-channel inputs LAM was designed for. We benchmark a diverse set of upsampling architectures, spanning lightweight convolutional networks, iterative back-projection models, physics-informed networks, and generative adversarial approaches. We also study whether aligning these upsamplers with LAM by training them jointly or in different stages helps preserve the spatial structure that LAM depends on. Results show that the original full-resolution LAM is the strongest, that separately trained lightweight models are the most competitive learned approaches, and that representation alignment between the upsampler and LAM matters more than model complexity.

Problem

• Latent Acoustic Mapping
• microphone array sparsity
• cross-spectral matrix
• acoustic mapping
• spatial information

Innovation

• Latent Acoustic Mapping
• array upsampling
• self-supervised learning
• spatial representation alignment
• microphone array