2026-04-29T13:33:05Z https://www.peertechzpublications.org/oai-pmh

oai:www.peertechzpublications.org:10.17352/ara.000020 2025-08-22 PTZ.ARA:VOL9

Multi-Method System for Robust Text-Independent Speaker Verification Stefan Hadjitodorov<p>A system for robust speaker verification based on four recognition approaches and methods (classifiers) is proposed, in order to use different statistical characteristics of the speech parameters. These methods are: 1) Prototype Distribution Maps (PDM); 2) AR-vector models (ARVM); 3) Two-level approach: the first level uses several PDMs for preprocessing, and the second employs multilayer perceptron (MLP) networks; 4) Gaussian speaker’s models combined with the arithmetic-harmonic sphericity measure (GMAHSM).</p><p>These classifiers generate four preliminary classification decisions. The reliability and confidence of these preliminary decisions are evaluated by means of a weighting algorithm. The weights are assigned using the relative measures to the most similar speakers (or cohorts), i.e. a Cohort Normalization technique is implemented. The final classification is then performed using simple logical and threshold rules.</p><p>The speech signals of 92 speakers have been analyzed. The speaker verification accuracy was over 98%. Robust impostors’ detection was observed, because the classifiers never fail simultaneously.</p> Annals of Robotics and Automation - Peertechz Publications 2025-08-22 Review Article https://doi.org/10.17352/ara.000020 en Copyright © Stefan Hadjitodorov et al.