應用於個人電腦環境之即時語音純化系統設計

Abstract

本論文針對桌面環境的訊號的干擾源：如喇叭、迴響環境，使用適應性麥克風陣列訊號處理，抑制干擾源的影響，並使輸入輸出即時化。本論文亦提出一個符合實驗室規模、使用USB1.1介面、8通道之即時性麥克風陣列訊號處理實驗平臺，已實作完成。以此實驗平臺錄製麥克風陣列語音樣本供相關研究，並實作完成適應性空間濾波器。 論文主要分三大部分，第一部分介紹陣列式麥克風波束形成(Beamforming)演算法的概念、適應訊號處理理論。第二部分介紹實驗平臺的架構。第三部分則是演算法在實驗平臺上的實作，實際輸入輸出的結果與分析。

The objective of the thesis is to subpress the interference sources of the desktop environment like loudspeaker, echo, and environmental noise by using adaptive microphone array. The implementation also requires a real-time input and output of signals. A flexible real-time array-processing platform of 8 sensors using USB 1.1 interface is proposed and implemented. Using this platform, an adaptive spatial filter is also implemented. The thesis is divided into three parts. The first part is an introduction of the microphone array beamforming algorithm and adaptive filter used. The second part is the platform implementation. The third part is implementation of the adaptive microphone array algorithm on the platform and discussions of the experimental results.