Geometry Design Based on Microphone Density Analysis

For a fixed number of microphones, array geometry is the dominant factor to affect beamforming performance for speech signals in immersive environment. Our work, therefore, focuses on the relationship between microphone distribution properties and spatial filtering performance. Array geometries for immersive environments are statistically analyzed to specify the important mic density parameters related to array performance. Feasible optimization algorithms were proposed with the objective function rules using relationship functions and probabilistic descriptions of acoustic scenes to incorporate various levels of prior knowledge of the source distribution. General guidelines in real scenarios were introduced to effectively control the mic distribution parameters related to microphone density, and directly build arrays with superior SNR performance. In addition, arrays mutated from the regular configurations were also introduced to overcome the limitations of regular arrays, and provide good SNR results for speech signals with easily installation.