A speaker recognition system using power spectrum density and similarity measurements

Most of speaker identification systems are based on the computation of distance or likelihood between the feature vectors of the unknown speaker and the models in the database. The identification process depends on the number of feature vectors, their dimensionality and the number of speakers. This research aims to develop a system able to identify a person from a sample of his speech. Our system is based on English words. Recognition relies on a text-dependent system using the English digit “one” as a single password. Speech features are extracted using power spectral density estimation. The similarity measurements based on the distance is used in the matching phase. The system is designed for a set of speakers. The results obtained by the proposed algorithm give a good accuracy rate.