基于相空间重构的语音增强

Speech Enhancement Based on Phase Space Reconstruction

将相空间降噪方法应用于语音增强之中。由于语音信号集中在有限空间,而随机噪声则分散在各个分量中,通过找到信号能量集中的信号空间,去除噪声能量集中的冗余空间,达到减少噪声的目的。针对语音信号的特点,本文对相空间降噪基本算法中邻点的选取方法进行了改进,在没有增加太多运算量的前提下,提高了对不同信噪比信号的适应性。另外,应用改进后的算法,本文分别对汉语单个音素和连续语音进行了相空间语音增强测试。实验结果显示,改进邻域选择后的相空间语音增强方法可以显著提高信噪比。

The phase space denoising is applied to speech enhancement.The speech signal concentrates on a subspace of finite dimensions,but the noise spreads to all directions.So It is possible to reduce the noise by finding the signal subspace and removing the noise subspace.The method for selecting neighbor points in the phase space is improved to increase the adaptability of different noise levels.Chinese phonemes and continuous speech are taken as an example to evaluate the new algorithm.Experimental results show ...