Speech Enhancemwnt In High Noise Environment

The trade-off among the amount of noise reduction, the level of musical residual noise and the speech distortion, which is the key problem of spectral subtraction speech enhancement, is need to be solved. Although so many methods have been advanced recent years, this problem is not resolved until now. And also, because of so many kind of noise that degrade speech signal, different method is needed to enhancement speech degraded by different noise.This paper focuses on the trade-off among the amount of noise reduction, the level of musical residual noise and the speech distortion, which is the key problem of spectral subtraction speech enhancement. The approaches of speech enhancement are numerous. The first part of this paper is origin and history of speech enhancement, several enhancement approaches and some knowledge related to speech enhancement are given too, for example, the models speech and noise, and the estimation of noise. This article has emphasisly carried on speech enhancement in noise environment that which has a low signal to noise rate, and advances two improved arithmetic to suit noise environment of the signal. One is A Speech Enhancement Algorithm Based on Masking Property of Human Auditory System and Speech Absence Probability, in this approach, the masking model of human auditory is modified, the masking thresholds of the key frequency segments of each speech frame are determined dynamically through introducing the speech absence probability(SAP); Another is An Improved Spectral Subtraction Speech Enhancement Method Based on Adaptive Smoothing, in this approach the smoothing-parameters in noise estimation are adaptively adjusted to the Speech Absence Probability (SAP) parameter which reflects stationary degree of the measured signal. The change of smoothing-parameters allows for automatic adaptation with various noisy environments and obtains the best trade-off. Experimental results demonstrate that the proposed algorithm has better performance of speech articulation and less musical noise. Moreover, this superiority is more significant at low SNR.