To the existing problems of the flexible segmentation algorithm proposed by Wang Yongzhong,et al.,in this paper,we make some improvements,and apply the new improved algorithm to automatic speech voiced-unvoiced segmentation.A great deal of numerical simulations and practical data processing show that this new algorithm assuredly solves several problems of the binary-based segment algorithm and the segment algorithm proposed by Wang Yongzhong,and achieves adapted effective segmentation for speech signal.The criterion of distinguishing voiced from unvoiced speech segments proposed by Wesfreid,et al.is adopted,and the new improved algorithm is also applied to automatic speech voiced-unvoiced segmentation.It not only obtains ideal segmentation result,but also there are not much more redundancy segmentations.