Предварительная обработка голосового сигнала в системах аутентификации

Abstract

The scientific problem of increasing the efficiency of voice authentication systems is considered. As the direction of research, the procedures for preprocessing the voice signal are chosen, which in modern systems are reduced to normalizing the signal in amplitude and duration. Based on the phase information of the voice signal, it is proposed to implement procedures for eliminating distortions in the recording materials of the user's acoustic wave and the signal-tonoise ratio.