Speech Enhancement Using Cascaded LMS Algorithm

In this paper, we investigate the enhancement of speech by applying Cascaded LMS Algorithm. Noise removal is very important in many applications like telephone conversation, speech recognition, etc. Least Mean Square (LMS) adaptive noise cancellers are widely used to recover signal corrupted by additive noise due to its simplicity in implementation. But it has limitation when the desired signal is strong, that the excess mean-square errors increase linearly with the desired signal power. This results in poor performance when the desired signal exhibits large power fluctuations. Here several algorithms were proposed to achieve maximum SNR value with minimum distortion, such as normalized least mean square algorithm (NLMS) and variable step size least mean square algorithm (VSSLMS). But in NLMS algorithm, selection of step size and filter length of adaptive filter for different type of noise with different noise level (dB) that gives maximum SNR is difficult. This needs various trials of step size and filter length to get optimum solution. The aim of this paper is to implement various adaptive noise cancellers for speech enhancement. In this paper, we can say that the signal to noise improvement in the input signal after Cascaded LMS filtering is much higher and it is also simple to implementation compared to that of LMS filter algorithm. Therefore we conclude that the Cascaded LMS algorithm is an efficient adaptive filtering algorithm than least mean square (LMS) algorithm. The maximum signal to noise ratio (SNR) with minimum mean square error (MSE) in simulations which were carried out using MATLAB software with different noise signals.