低階三度音訊之實現

Abstract

由於虛擬實境應用的蓬勃發展，三度音訊的研究顯得日益重要。這裡，我 們將以 ＨＲＴＦ作為３Ｄ音訊處理的基礎。然而，許多問題諸如記 憶體與計算量之需求， 未量測點ＨＲＴＦ之合成等依然存在。有鑑於 此，我們使用時域主軸分析法減少記憶體需求。重新設計低階濾波器以減 少計算量，並保持頻譜特性。最後我們使用主軸權數上之線性內插，雲線 內插以及函數化以達到合成未量測ＨＲＴＦ的目的。當這些方法適當地組 合，記憶體的需求可以減少到原本的五十分之一，而乘法器則可以減少到 二十分之一。最後我們以實做來驗證３Ｄ移動音源的效果。 The study of 3D sound becomes more and more important because of the application of virtual reality grows vigorously. Here, the HRTFs are introduced and applied to display 3D sound. However, there are many problems such as memory requirement, computational complexity, and unmeasured HRTFs. We use T- PCA algorithm to reduce the memory requirement, redesign the filter to a shorter one to reduce the multiplication but keep the spectrum cue, and finally synthesize unmeasured HRTFs by linear, spline, and functional interpolation in principal weights. The memory requirement become one 50th and multiplication of FIR become one 20th if the methods in the thesis are combined. Finally, we establish a 3D moving sound demo system to verify the performance of the suggested algorithm.