针对生物特征模板涉及用户隐私而易受到各种攻击的问题，在定义随机映射形式化表示的基础上，结合主流文本无关说话人识别技术，提出了一种基于随机映射的声纹模板保护方法．在注册阶段，将声纹特征映射至随机空间再训练高斯混合模型( GMM) ，并存储模型参数作为模板; 在认证阶段，待认证的声纹特征在相同的变换域与模型库进行匹配．文中还给出了该方法在认证性能保持和安全性方面的理论分析．实验结果表明，适当降维能在提高安全性的同时近似保持GMM 的身份认证性能，而目前基于矢量量化的认证性能下降超过8%，说明随机映射更适用于基于GMM 的声纹认证系统的模板保护．

Biometric templates are vulnerable to various attacks because they involve user’s privacy.In this paper,by combining the defined formalized representation of random projection with the mainstream text-independentspeaker recognition,a template protection method based on random projection is proposed for voiceprint-biometricsystems.In the enrollment stage,voiceprint characteristic data are projected onto a random space,a Gaussian mixturemodel ( GMM) is then constructed,and the corresponding model parameters are stored as a template.In theverification stage,the voiceprint characteristics to be verified are matched with the model base data in the same randomspace.Moreover,the performance and security of the proposed method are theoretically analyzed.Experimentalresults show that suitable dimensionality reduction helps to improve the system security and approximately maintainthe performance of GMM,while the existing vector quantization method may result in a performance degradationof more than 8%,which means that the random projection is more suitable for the template protection of GMMvoiceprint-biometric systems.