Performance analysis of dual source transfer-function generalized sidelobe canceller

摘要

In this work, we evaluate the performance of a recently proposed adaptive beamformer, namely Dual source Transfer-Function Generalized Sidelobe Canceller (DTF-GSC). The DTF-GSC is useful for enhancing a speech signal received by an array of microphones in a noisy and reverberant environment. We demonstrate the applicability of the DTF-GSC in some representative reverberant and non-reverberant environments under various noise field conditions. The performance is evaluated based on the power spectral density (PSD) deviation imposed on the desired signal at the beamformer output, the achievable noise reduction, and the interference reduction. We show that the resulting expressions for the PSD deviation and noise reduction depend on the actual acoustical environment, the noise field, and the estimation accuracy of the relative transfer functions (RTFs), defined as the ratio between each acoustical transfer function (ATF) and a reference ATF. The achievable interference reduction is generally independent of the noise field. Experimental results demonstrate the sensitivity of the system's performance to array misalignments.