助聽器回授路徑測量及適應性的噪音消除演算法

Abstract

本篇論文主要探討助聽器回授路徑(feedback path)的測量及使用適應性濾波器(adaptive filter)消除噪音的演算法模擬。

我們首先用一個耳內型(ITC)助聽器，裡面無放大電路僅有接收器(receiver)及麥克風(microphone)，接上脈衝產生器(pulse generator)使用掃頻(sweep stimulus)的方式驅動助聽器的接收器並接收麥克風的聲音得到回授路徑的頻率響應(frequency response)，再由程式將之轉回時域的脈衝響應(impulse response)，我們的測量結果可供消除回授演算法的參考及實現。

卡爾曼濾波器(Kalman filter)是一種有效率的適應性濾波器且可應用在時變系統上。尤其是更新估記狀態時僅需計算前一個狀態估記值及新得到的資料，所以只有前個狀態需要儲存。因此我們考慮將卡爾曼濾波器列入助聽器消除噪音演算法的可能性。我們做了一些卡爾曼濾波器在單一頻帶消除白雜訊的學習，至於在分頻濾波上加上卡爾曼濾波器因為可能增加大量運算故先暫時予以保留。

In this thesis, we focus on the measurement of feedback path of hearing aids and the simulation of adaptive filter algorithm for noise cancellation.

First of all we put an ITC hearing aid embodying only a receiver and a microphone in the artificial ear in the anechoic chamber. We use the pulse generator to inject the sweep signal to the receiver and receive the sound from the microphone to get the frequency response. Then make use of Matlab to transform it into impulse response. Our measurement result may supply to the realization of the feedback cancellation algorithm.

Kalman filter is an efficient adaptive filter and can be used for time-varying system. It only needs the estimated state from the previous time step and the current measurement to compute the estimate for the current state, so only the previous estimate requires storage. Therefore we consider the possibility of using the Kalman filter in hearing aids for noise cancellation. The single-band Kalman filter for white noise cancellation is studied while the multi-band Kalman filter is kept aside due to the possible surge in computational cost.