FIELD: medicine. SUBSTANCE: method involves selecting at least one sign from the first speech fragment pronounced by the speaker. The sign is classified by applying a set of classifiers for building classification results data set. The sets of classification results are united for building a set of generalized classification results. The generalized classification results are recognized by determining similarity between the generalized classification results and the second speech fragment pronounced by the speaker before speaker verification. The decision to accept or reject the speaker is taken on the basis of recognized generalized classification results. The system has all means for carrying out the mentioned operations. EFFECT: enhanced accuracy in person identification. 24 cl, 20 dwg
