超指向性麥克風陣列應用於近場聲源全像及高感度麥克風陣列

Abstract

本論文探討兩項主題：一為高感度麥克風，另一為超指向性麥克風陣列應用於近場聲源全像技術；前者技術中可分為號角(horn)及曲型麥克風陣列(curved array)，在此論文一一探討其理論、設計方法及成效，最後將此技術應用於遠距離收音應用。超指向性麥克風陣列應用於近場聲源全像技術主要有別於傳統近場聲源全像是採用雙層麥克風陣列，其各個通道包含兩麥克風且可視為一超指向性麥克風，根據超指向性麥克風之設計方法，可設計出指向性為全向(omni)、雙極(dipole)、心型(cardioid)指向性麥克風，或者利用最佳化方法評估三種客觀參數：指向性因子(directivity index)、前後比(front-to-back ratio)以及不變的的波束寬(constant beam-width)，將上述三種客觀參數最佳化，便能得到出超指向性麥克風陣列的濾波器，以上演算法都是估算出聲壓，另外可利用有限差分(finite difference)估算出粒子速度，進而利用等效聲源法(ESM)完成近場聲源全像；此論文中探討各方法之優缺點，並透過模擬及實驗驗證。

There are two topics in this thesis: one is supersensitive sound pickup and the other is super directive microphone array with application in nearfield acoustic holography. The former contains horn and curved array, and we will discuss the theory, design method and the result. Final, the device is applied to the application in a distant recording. Conventional nearfield acoustical holography (NAH) is generally based on the free-field assumption, which can cause errors when interfering sources are present in practical environment. Although the measurement of particle velocity as the input to NAH provides certain advantage, the noise problem of finite difference estimation of particle velocity can nullify the velocity-based reconstruction that is better conditioned than the pressure-based process. Alternatively, this paper examines the feasibility of using directional sensors in each channel of the microphone array such that the robustness of inverse reconstruction is enhanced against reflections from boundaries. With two microphones in each channel, the directivity of each array element is tailored according to various design criteria of first-order differential microphones. Directivity index, front-to-back ratio and constant beam-width are employed as the objective functions for optimizing array filters. The proposed methods are utilized in an Equivalent Source Model (ESM)-based NAH. The proposed techniques are verified by numerical simulations and experiments, with interfering source positioned at various directions. Sound field is reconstructed using the pressure input and the particle velocity estimated by the finite difference method.