Independent Reproduction of a Mixed-Directivity Linear Superarray Beamformer: Evidence of a Microphone-Configuration Inconsistency

Compact linear microphone arrays provide a practical approach to beamforming, but their performance depends strongly on array geometry, microphone directivity, and beamformer design. Independent reproduction of published beamforming methods is therefore important for verifying reported performance and identifying ambiguities in implementation.

This paper presents an independent implementation of a published mixed-directivity linear superarray (LSA) beamformer. The implementation follows the reported array geometry, microphone directivity model, null-constrained beamformer formulation, and evaluation metrics. Performance is assessed using beampatterns, Directivity Factor (DF), and White Noise Gain (WNG), and compared quantitatively with reference values digitized from the published performance curves.

The documented microphone configuration reproduced the steering response, prescribed nulls, and overall beamforming behaviour, but did not reproduce the published DF and WNG quantitatively. Under the documented assignment, the mean absolute error (MAE) was 8.07 dB for WNG and 0.12 dB for DF. A controlled comparison in which only the microphone assignment was changed showed that assigning omnidirectional microphones to the odd-indexed array positions reduced the WNG MAE to 0.64 dB and the DF MAE to 0.02 dB. These reductions are substantially larger than the estimated digitization uncertainties of 0.016 dB for WNG and 0.011 dB for DF.

The results therefore provide evidence of an apparent inconsistency between the documented microphone configuration and the configuration associated with the published performance curves. More broadly, the study demonstrates how independent implementation and controlled verification can identify configuration-level ambiguities that materially affect the reproducibility of compact microphone-array beamforming results.