【摘要】 语音气流中具有混沌特征，而分形可以定量地分析混沌现象，故分形可作为分析语音信号的数学工具．由于传统的Ｈａｕｓｄｏｒｆｆ－Ｂｅｓｉｃｏｖｉｔｃｈ 维数没有考虑关于集合中点的分布信息，本文引入多重分形维数来克服上述缺点．实验表明，多重分形维数语音分割方法明显好于单一Ｈａｕｓｄｏｒｆｆ－Ｂｅｓｉｃｏｖｉｔｃｈ 分形维数的语音分割方法更多还原

【Abstract】 There are chaotic characters in speech air flow. And fractal can be used to quantify the chaotic phenomenon. So we can use fractal as a mathematical vehicle to analyze speech signals. The traditional Hausdorff Besicovitch dimension does not consider the information about the distribution of the points in the set. Here we introduced multifractal dimension to overcome the drawback. The experiment result shows better segmentation is achieved by multifractal dimension than only by single Hausdorff Besicovitch fractal dimension.更多还原