The embodiment of the invention provides a voiceprint recognition method and device. The method comprises the steps that a deep neural network primary model is built on the basis of label-free voice data, the deep neural network primary model is trained on the basis of label voice data, and therefore a deep neural network secondary model is obtained; the speaker high-level voiceprint characteristic and speaker identification are registered by means of the deep neural network secondary model; test voice data are received, the test voice high-level voiceprint characteristic is extracted from the test voice data on the basis of the deep neural network secondary model, and the speaker identification is determined on the basis of the test voice high-level voiceprint characteristic. Due to the fact that deep neural network automatic learn voice signals comprise high-level speaker information capable of resisting noise interference, the noise interference can be effectively resisted.