CPC H04L 63/0869 (2013.01) [H04L 63/08 (2013.01); H04L 63/0861 (2013.01); H04L 63/18 (2013.01)] | 15 Claims |
1. A processor implemented method, comprising:
receiving, via one or more hardware processors, an initial identity information corresponding to a valid user requesting a service;
obtaining, via the one or more hardware processors, a plurality of attributes and a plurality of reference voice features corresponding to the valid user that are pre-stored in a dynamically updated system database;
dynamically generating, via the one or more hardware processors, a question based on the plurality of attributes corresponding to the valid user that are pre-stored in the dynamically updated system database using a probabilistic template based approach;
generating, via the one or more processors, an audible recitation of the dynamically generated question in a first language from a set of languages identified from the plurality of attributes, wherein the audible recitation includes an added environmental noise;
requesting, via the one or more processors, the valid user to provide a spoken response to the audible recitation of the dynamically generated question in a specific language among the set of languages, wherein the spoken response comprises a plurality of incoming voice features corresponding to the valid user and a knowledge possessed by the valid user pertaining to the plurality of attributes that are pre-stored in the dynamically updated system database;
transforming, via the one or more hardware processors, the plurality of incoming voice features and the plurality of reference voice features to maximize a separation between contours and a subset of high energy points in a spectrogram of the spoken response and a synthetically generated response; and
determining, via the one or more hardware processors, one or more authentication metrics to verify a final identity of the valid user by comparing (i) the plurality of incoming transformed voice features with the plurality of transformed reference voice features and the knowledge possessed by the valid user comprised in the spoken response with the plurality of attributes corresponding to the valid user that are pre-stored in the dynamically updated system database.
|