The invention discloses a voice and lip language fused identity authentication method, and relates to the field of information safety and mode authentication. The voice and lip language fused identityauthentication method comprises the following steps of: 1, extracting to-be-authenticated voice cepstrum characteristics and lip language characteristics; 2, fusing the voice cepstrum characteristicsand the lip language characteristics by adopting a parallel strategy to obtain complex vector characteristics; and 3, training a multiple input multiple output model of a complex extreme learning machine by using the complex vector characteristics as training samples, judging whether authentication is successful according to distribution and values of outputs, if a maximum value in output resultvectors is consistent with a prompt and greater than a set threshold, determining that authentication is successful, or determining that authentication is failed. The voice and lip language fused identity authentication method adopts voice and lip language combined multi-modal biological characteristics, so that robustness of an identity authentication system is improved.