Abstract: This paper introduces a method based on the combination of spectral subtraction and adaline linear neural networks in the enhancement of Chinese whispered speech.First,enhance the whispered speech with spectral subtraction to get a good envelope;then applie AD networks to the enhanced signal to have a high quality whispered speech.The result shows that the SNR output can be improved by 20 dB even if the SNR input is low.The algorithm can give better SNR improvement and better aural effect.

摘要： 论文针对带噪的耳语音提出了一种利用ADALINE神经网络消除背景噪声的耳语音增强算法。首先利用传统的谱减法来取得较好的谱包络,在此基础上使用AD线性神经网络进行自适应预测以达到提高耳语音质量的目的。结果表明,即使在低信噪比的情况下,信噪比也能提高20 dB左右,而且取得了良好的听觉效果。