A different-from-conventional voice recognition approach for person recognition by identifying their voice is presented in this work. The main mathematical tool applied in the study is based on Töeplitz matrices. The recorded voice signal is filtered and processed for power spectrum estimation. The feature vector is then derived from the power spectrum and Töeplitz forms. This vector furnishes a unique voice-print differing from person to another. The new idea of this study lies in applying Toeplitz matrix minimal eigenvalues algorithm to Burgs estimating model for signal-image description and hence classification. This graphical approach of voice signal processing for human identification has introduced a new biometric approach to find its place among the non-conventional classical methods of human identification by verifying their voice. In this article related recent patents are also discussed.