利用基礎函數法、等效聲源法以及聲源揚起技術實現聽覺聲場重建

Abstract

本論文著重於利用雙聲道以及多聲道喇叭實現空間聲場的重建。對於如何重建聲源揚起的聽覺，頭部轉移函數 (HRTF) 是非常重要的技術，經由頻域和時域兩種解摺積的方法可計算出所需的逆濾波器，其中正規化以及逆濾波器實現架構的設計也是很重要的考量。此外我們將介紹波場合成 (WFS) 的基本概念以及提出以平面波為原則的基礎函數法 (BFM) 和已被廣用於噪音源辨識 (NSI) 的等效聲源法 (ESM) 來實現聲場重建的問題。其中我們利用截斷奇異值分解 (TSVD)來解決矩陣逆轉換的問題。為了避免重建聲場的失真，本論文提出逆風區判定法 (HD) 適當得使部分喇叭不作用，而麥克風與喇叭距離 (RD) 的最佳化是利用黃金比例法 (GSS-PI)。所有以上提出的技術將在本文中進行數值模擬以及主觀性評估，最後經由實驗結果可提出最佳的聲場重建設計方法。

This thesis focuses on the spatial audio reconstruction for two-channel stereo loudspeakers and multi-channel loudspeakers. To create spatial impression during audio reproduction, the head-related transfer function (HRTF) is very important in many audio spatializer such as source lifting. Two deconvolution approaches, the frequency-domain method and the time-domain method, are employed to design the required inverse filters. Issues in the implementation phase such as regularization, complex smoothing and structures of inverse filters are also addressed. In addition, we introduce the basic idea of Wave Field Synthesis (WFS) and propose the Basis Function Model (BFM) which is based on the concept of plane wave, and the Equivalent Source Model (ESM) which has been generally used in Noise Source Identification (NSI). Truncated Singular Value Decomposition (TSVD) had been employed to deal with the inverse problems. Then we use the Headwind Detection (HD) to disable a part of loudspeakers to avoid the serious spatial aliasing. For the optimization of retraction distance (RD), an algorithm based on the golden section search and parabolic interpolation (GSS-PI) will be revealed. All proposed methods were examined by numerical simulations and subjective experiments, and had been proven to be effective in spatial audio reconstruction.