雙麥克風可適性抗噪系統於通訊環境改善之應用

Application of Dual Microphone Adaptive Echo Cancellation System in the Communication EnvironmentImprovement

葉庭宏(TING-HONG YEH)

指導教授 ： 王昭男

國立臺灣大學/工學院/工程科學及海洋工程學研究所/碩士(2018年)

https://doi.org/10.6342/NTU201702707

摘要

本文提出一種雙麥克風方法，以麥克風陣列(Microphone Array)理論為基礎，發展出一種可套用在通訊用麥克風設備上的系統，主要目的為消除環境噪音以及喇叭因為時間延遲而傳來的回音(來自近端或遠端)，進而創造一個更舒適的通訊環境，並使用最小均方誤差濾波器(Least Mean Squares filter)、正規化最小均方誤差濾波器(Normalize Least Mean Squares filter)、回歸最小平方法濾波器(Recursive Least Square filter)三種常見的濾波方式，以找到最適合的參數與演算方式讓抗噪效果最佳化。為了解本文方法之優劣,文中將與單麥克風方法進行結果的比較。
在抗噪系統中，濾波器的運作方式主要是透過兩段有相關性的訊號進行比對，找出並消除其中的雜訊部分，而訊號間的相關性越大，則越能準確的找出其中的雜訊部分。相較於單麥克風方法，雙麥克風方法用以比較的兩段訊號由於處在同樣的空間且時間相差極短，因此相關性較大，能更有效的找出雜訊並消除。
本研究預設的情境是遠端會議進行的情況下，在溝通時，有時會出現雙方同時講話的情況，此時麥克風會收到除了近端聲音之外，還有遠端自通訊設備的喇叭傳來的對話聲音，傳遞之後遠端會接收到近端的訊號以及自己的回音，夾雜在一起使得訊號模糊，進而使溝通無法順利進行，本文將這種狀況稱為"雙向溝通(Double Talk)造成的通訊干擾"，研究的雙麥克風裝置就是用以解決這類問題。
由實驗結果可以證實雙麥克風方法在處理通話方面有較好的結果，不論是單向溝通或是雙向溝通，在經由雙麥克風系統處理後都可以有效的抗噪。

關鍵字

遠端會議 ； 通訊 ； 抗噪處理 ； 濾波器 ； 麥克風陣列 ； 訊號處理

並列摘要

This paper presents a dual-microphone method based on the Microphone Array theorem, which develops a system that can be applied to a microphone device for communication purposes. The main purpose is to eliminate the ambient noise and the echo of the speaker due to time delay (Come from one's own way or each other) ,in order to create a more comfortable communication environment. Least Mean Squares filter(LMS filter), Normalized Least Square filter(NLMS filter) and Recursive Least Square filter(RLS filter) are used with deep learning to find the most suitable parameters and algorithm to optimize the anti-noise effect. To compare with a method which is the more intuitive way to eliminate noise, we call it's ingle microphone method" in this paper.
In the anti-noise system, the filter works mainly through two pairs of relevant signals to match, find and eliminate the noise part of the noise, and the greater the correlation between the more accurate, the more accurate to find Out of the noise section. Compared to the single-microphone method, the dual-microphone method is used to compare the two signals because of the same space and the time difference is extremely short, so the correlation is large, can more effectively find the noise and eliminate.
The default of the study is the case of remote meetings, during communication, sometimes the both sides speak at the same time. At this point the microphone will receive our voice, as well as the sound of dialogue that comes from the speaker of communication equipment. After the transmission the other party will receive our message and their own echo, both sounds mixed together makes the signal blurred, so the communication cannot proceed smoothly. This article calls this situation as "Double Talk" which cause the communication interference. The dual-microphone device is used to solve such problems.
The results of the experiment can be proved that the dual-microphone method has good results in dealing with the call, whether it is single talk or double talk, after the treatment by the dual-microphone system can effectively anti-noise.

並列關鍵字

remote conference ； communication ； anti-noise processing ； filter ； microphone array ； signal processing.

參考文獻

[2]E. Soria, J. Calpe, J. Chambers, M. Martinez, G. Camps, and J.D.M. Guerrero, “A novel approach to introducing adaptive filters based on the LMS algorithm and its variants”, IEEE Transactions on Education, vol. 47, pp. 127-133, 2008.

[3] J.Benesty,F.Amanf,AGilloire,"Adaptive filtering algorithms for stereophonic acoustic echo cancellation", IEEE Xplore: pp.3099 - 3102 06 August 2002,1995

[4] P.L.Feintuch," An adaptive recursive LMS filter",Proceedings of the IEEE Volume: 64, Issue: 11, pp.1622 - 1624 , Nov. 1976

[5] C.P.Kwong,“Dual Sign Algorithm for Adaptive Filtering”,IEEE Trans. Commun., vol. Com-34, pp.1272-1275, Dec. 1986

[6] J. R. Triechler, et. Al., Theroy and Design of Adaptive Filters, John Wiley and Sons, 1987