【摘要】 为了提高两个说话人情况下电话语音双人检测的性能,提出了一种基于长时距离与短时距离之和的双人语音分段算法,并对分段过程中的聚类过程进行了改进。对于分段得到的语音采用基于GMM-UBM(Gaussian mixturemodel-universal background model,GMM-UBM)模型的说话人确认算法,构成一个电话语音双人检测系统。实验结果表明:在美国国家标准技术局1999年组织的年度说话人识别评测双人检测数据库上的检测等错误率为15.1%。该算法取得了良好的效果。更多还原

【Abstract】 An algorithm is given to detect and track speakers in two-person telephone conversations.The approach uses a Gaussian mixture model with a universal background model(GMM-UBM) of speaker detection system as the core speaker recognition engine.The segmentation algorithm is based on the sum of the long-term distance and short-term distance measures,with an improved clustering process.Experiments on the NIST’99 evaluation database show that the detection system based on the segmentation algorithm provides good performance with an EER of 15.1%.更多还原