Search papers, labs, and topics across Lattice.
This paper introduces AmbiDrop, a novel framework for speech enhancement that achieves array independence by utilizing ideal Ambisonics as input for deep neural networks. By incorporating a channel-wise dropout layer during training to mimic Ambisonics encoding errors, AmbiDrop effectively decouples the learning process from specific microphone configurations, enhancing generalization to unseen layouts. Extensive evaluations reveal that AmbiDrop not only excels in robustness across diverse array setups but also maintains effectiveness under sensor failures and reduced network sizes, making it ideal for deployment on resource-constrained devices.
AmbiDrop's innovative use of Ambisonics allows for superior speech enhancement across arbitrary microphone configurations, outperforming traditional methods constrained by fixed geometries.
Multichannel Deep Neural Networks (DNNs) have significantly improved speech enhancement performance; however, they typically remain constrained by reliance on fixed microphone array geometries, leading to poor generalization on unseen or irregular configurations. Current array-agnostic approaches often rely on high-complexity architectures or massive, diverse datasets, yet they still struggle to generalize to out-of-distribution layouts. In this paper, we present an in-depth analysis of AmbiDrop, a recently proposed framework that achieves geometry independence by leveraging ideal Ambisonics as the DNN input. By employing a channel-wise dropout layer during training to simulate Ambisonics encoding errors, AmbiDrop decouples the learning process from the physical sensor arrangement. During inference, microphone signals from arbitrary array configurations are transformed into the Ambisonics domain via Ambisonics Signal Matching (ASM) before processing. Extensive experiments demonstrate that AmbiDrop maintains high robustness across a diverse suite of unseen simulated arrays and real-world recordings. Furthermore, our results show that the framework is resilient to sensor failures and remains effective even with reduced network scales, making it highly suitable for deployment on resource-constrained edge devices and versatile wearable hardware.
