摘要
针对传统单一音频分割算法中存在的冗余分割点过多问题,研究了一种基于一般似然比(GLR)和贝叶斯信息准则(BIC)相结合的广播音频顺序分割算法,提出了候选跳变点潜在区域的判断准则,并给出跳变点在潜在区域的检测方法,最后对检测到的跳变点进行校验。实验结果表明,与传统的音频分割算法相比,该算法的综合性能大大提高,达到较好的分割效果。
Due to traditional single audio segmentation algorithm suffers from a large amount of redundancy change points, a hybrid approach for audio sequential segmentation in broadcasting based on generalized likelihood ratio （GLR） and Bayesian Information Criterion （BIC） is proposed. The criterion of potential region of candidate change point and the detection method of change point is presented, and the validation of true change points is given. Compared with the algorithms of traditional audio segmentation, the results show that this approach is effective and feasible.
出处
《计算机工程与设计》
CSCD
北大核心
2009年第13期3120-3123,共4页
Computer Engineering and Design
基金
重庆市教育委员会科学技术研究基金项目(KJ080524)
关键词
广播音频分割
一般似然比
贝叶斯信息准则
声学特征跳变点
校验
broadcasting segmentation
generalized likelihood ratio （GLR）
Bayesian information criterion （BIC）
acoustic change points
validation
