摘要
声纹识别是一种重要的生物认证方式,同指纹一样,声纹也是不可复制的。由于语音获取的便利性和声纹识别技术的提高,其在司法身份鉴定中日渐被重视。本文采用高斯混合模型(Gaussian Mixture Model,简称GMM)并结合统一背景模型(Universal Background Model,UBM)进行身份鉴定,取得了良好的效果。
Voiceprint recognition is one of the important techniques for biology authentication. Like fingerprint, voiceprint cannot be cloned. Due to the convenience to obtain the speech utterances and the development of the technique, the application of voiceprint recognition in judicature is becoming more and more popular. In this paper, Gaussian Mixture Model (GMM) and Universal Background Model (UBM) are adopted to conduct the speaker verification, which has demonstrated good performance.
出处
《心智与计算》
2008年第1期1-7,共7页
Mind and Computation
