提出了一种二维空间的语音分形增强算法。算法对已加窗、模块化的语音数据进行二维傅里叶变换。以分形维数构成自适应滤波器,降低噪声,增强语音,得到清晰的信号。研究表明,分形算法可以弥补传统线性分析方法的不足,二维自适应滤波前后语音的包络形状较好地保持了原始语音信号的特征,算法减小了由于非平稳噪声的影响,并有效抑制了音乐噪声。 A two-dimensional speech fractal enhancement algorithm is proposed. Algorithm for the windowed, modular speech data for two-dimensional Fourier transform. The fractal dimension constitutes an adaptive filter to reduce noise, enhance speech, get a clear signal. The results show that the fractal algorithm can make up for the lack of traditional linear analysis methods. The envelope shape of the speech before and after two-dimensional adaptive filtering preserves the characteristics of the original speech signal well. The algorithm reduces the influence of non-stationary noise and effectively suppresses Music noise.