Synthesizing A Virtual Height Channel from Planar Microphone Arrays

In spatial audio processing, the ability to capture sound scenes in three dimensions is important for applying algorithms such as beamforming and direction-of-arrival (DoA) estimation. Conventional microphone arrays require at least four microphones that are not on the same plane to achieve three-dimensional processing. This work presents a method for synthesizing a virtual height channel from planar microphone arrays that could, for example, be placed on a flat surface such as a table, enabling full 3D Ambisonic processing without requiring physically elevated microphones. The approach relies on transforming array signals into the spherical harmonics (SH) domain, where only the W, X, and Y components are available for a planar array. The missing Z-component is synthesized using spectral subtraction, assuming that sources only originate from the upper half-space. The basic method strictly requires timefrequency sparsity; here, we also present an improved version incorporating diffuseness estimation to enhance robustness in reverberant environments. Evaluation results demonstrate the effectiveness of the synthesized height channel for DoA estimation and beamforming in simulated conditions, achieving comparable performance to recordings with a true height channel. This technique enables practical implementations of 3D sound field analysis with simpler, more flexible microphone configurations.