为了减少语音数据量 ,提高处理速度和识别的准确性 ,提出了一种采用公共码本、个人隐 Markov模型 (HMM)和个人拒识阈值进行两级决策来实现开集说话人辨认的新方法。在系统实现时 ,采用了一种改进的语音切分算法来提高输入数据的有效性 ,并将说话人识别和人脸识别融合在一起进行身份验证。实验证明这种融合方法能够有效地降低识别的相等错误率至 1%。 In order to reduce the amount of voice data and improve the processing speed and accuracy of recognition, a new method of public speaker recognition based on public codebook, Hidden Markov Model (HMM) and personal rejection threshold is proposed . In the implementation of the system, an improved speech segmentation algorithm is adopted to improve the validity of input data, and speaker recognition and face recognition are combined for identity verification. Experiments show that this fusion method can effectively reduce the recognition error rate to 1%.