论文部分内容阅读
文中针对加性白噪声的环境下,通过在传统的Wiener滤波算法中引入人耳听觉感知特性及谱减算法,提出了一种新的基于感知滤波的语音去噪算法。该算法的关键是采用LPC分析得出的感知加权函数修正维纳滤波方程,使噪声谱分布随语音而变。不仅保持Wiener滤波算法的优点,而且降低噪声对纯净语音的影响。实验表明,该方法能更有效地抑制背景噪声,提高语音质量,且具有较低的计算复杂度。
In the context of additive white noise, a novel speech denoising algorithm based on perceptual filtering is proposed by introducing the auditory perception of human ear and spectral subtraction into the traditional Wiener filtering algorithm. The key of this algorithm is to modify the Wiener filter equation by using the perceptual weighting function derived from LPC analysis, so that the noise spectrum distribution changes with the voice. Not only to maintain the advantages of Wiener filtering algorithm, but also reduce the impact of noise on pure speech. Experiments show that this method can suppress background noise more effectively and improve speech quality with lower computational complexity.