應用ESIF陣列技術來改善語音的品質

Abstract

本論文提出一種新的麥克風陣列技術運用聲學信號處理方法而實現在電信通訊系統中，此技術稱為聲源等值反濾波器設計演算法。 單進多出聲源等值反濾波器設計演算法(SIMO-ESIF)的目的在於在充滿迴響的環境裡能夠重建語音訊號，此系統能夠達到兩個重要的目標：抑止殘響和消除噪音。其適用的電信通訊系統如車內免持聽筒的系統，在密閉的車子環境裡所收到的語音通常夾雜著許多背景噪音且需要被改善，此演算法結合提出的GSC演算法是為了進一步在更嚴重迴響的環境裡改善噪音消除的效果。主觀測試的結果用變異數分析方法來做為分析的工具。進一步使用Fisher’s LSD 分析法來證明新提出的方法在改善含有噪音的語音訊號上效果有明顯的進步並且提供更棒的音質。

New microphone array techniques are proposed in this paper for acoustic signal processing in telecommunication application. These endeavors are based on the central idea of Equivalent Source Inverse Filtering (ESIF). The single input multiple output equivalence source imaging (SIMO-ESI) algorithms are suggested to reconstruct the speech signal in a reverberant environment. Specifically, the system serves two purposed: dereverberation and noise reduction. It has promise in telecommunication application such as the automotive hands-free system, where noise-corrupted speech signal often needs to be enhanced. In order to further improve the noise reduction performance in spatial filtering and robustness against system uncertainties, the SIMO-ESIF algorithm is combined with an adaptive Generalized Side-lobe Canceller (GSC). The system is implemented on an NI-PXI platform and evaluated experimentally in car environment. As indicated by several performance measures in noise reduction and speech distortion, the proposed microphone array algorithm proved effective in reducing noise in human speech without significantly compromising the speech quality. The results of subjective tests were processed by using analysis of variance (ANOVA) to justify the statistic significance. A post-hoc test Fisher’s LSD was conducted to further assess the pairwise difference between the NR algorithms.