提出了一种基于小波多分辨分析的频带阈值去噪方法,并给出了此方法的主要算法步骤。用此方法与单一阈值方法分别对一个实例信号进行了去噪处理,并基于两个客观评价标准:信噪比(SNR)、均方根误差(RSME),评估了去噪效果。实验结果证明了提出的方法是一种更有效的语音去噪方法,大大提高了信噪比,改善了清晰度。
A wavelet frequency band threshold denoising method based on multi-resolution analysis is presented and the primary arithmetic steps is given. The soft threshold method and frequency band threshold method are used to denoise a speech signal respectively. The performances of wavelet threshold denoising method according to the objective evaluation standards: signal to noise ratio (SNR), root square mean error ( RSME ) are evaluated. Experimental results demonstrate that wavelet frequency band threshold method is more efficient for speech signal denoising. It Can greatly improve the SNR and definition.