基于维纳滤波的语音增强研究

Spectral subtraction speech enhancement technology is the most widely used technique, but its residual musical noise is too strong; while improved spectral subtraction to eliminate residual musical noise despite the strengthening aspects, it is difficult to determine the appropriate algorithm coefficients must be in order to obtain a large number of experiments, this paper presents a Wiener filtering and spectral subtraction speech enhancement algorithm combined. The collected a set of real speech data respectively improved spectral subtraction and Wiener filtering with spectral subtraction processing, simulation results show that the spectral subtraction with a Wiener filter can improve SNR and speech intelligibility, good reduces the musical noise, greatly enhanced to meet the requirements of the speech signal.