Optimal 3D Beamforming Using Measured Microphone Directivity Patterns
Mark Roland Thomas, Jens Ahrens, Ivan Tashev · 2012
The design of time-invariant beamformers is often posed as an optimization problem using practical design constraints. In many scenarios it is sufficient to assume that the micro-phones have an omnidirectional directivity pattern, a flat fre-quency response in the range of interest, and a 2D environ-ment in which wavefronts propagate as a function of azimuth angle only. In this paper we consider a generalized solu-tion for those cases in which one or more of these assump-tions do not hold, yielding a beamformer that is optimized on measured directivity patterns as a function of azimuth, ele-vation and frequency. A comparative study is made with the 4-element cardioid microphone array employed in Microsoft Kinect for Windows, whose beamformer weights are calcu-lated with directivity patterns using (a) 2D cardioid models, (b) 3D cardioid models and (c) 3D measurements. Results on a recorded noisy speech corpus show similar PESQ and speech recognition accuracy comparing (a) and (b), but a 50% relative improvement in word error rate using measured direc-tivity patterns. Index Terms — Microphone array, beamformer, superdi-rective beamformer, MVDR