论文部分内容阅读
语音信号是一维非平稳信号,在传播过程中容易以各种方式混入噪声,导致对其分辨造成影响。为了净化噪声信号,对语音信号降噪进行了研究,给出了小波语音信号降噪的基本方法,并在matlab环境下应用sym3小波对携带噪声的语音信号进行处理。分别采用了经典的硬阈值法、软阈值法以及集软硬阈值法优点而进行的软硬阈值结合的方法对不同噪声强度的信号进行处理,结果对比分析显示,软硬阈值结合的方法降噪效果最为显著,能得到更高的输出信噪比。
Speech signals are one-dimensional, non-stationary signals that are easily mixed in with noise in the propagation process, causing their resolution to be affected. In order to purify the noise signal, the noise reduction of the speech signal is studied. The basic method of noise reduction of the wavelet speech signal is given. The sym3 wavelet is used to deal with the speech signal with noise in the matlab environment. The methods of combining hard and soft thresholds, soft thresholds and the merits of hard and soft thresholds are used to deal with the signals with different noise intensities respectively. The comparative analysis shows that the combination of hard and soft thresholds reduces noise The effect is the most significant, can get a higher output signal to noise ratio.