Person Voice Recognition Methods

This article is devoted to the selection and evaluation of language features used in the tasks of automatic text-independent speaker verification. The task of automatic speaker identification is one of the most difficult tasks in the field of language processing. The methods used in current speaker identification systems are not ideal, that imposes some restrictions on such systems. Some systems work well in good acoustic conditions, with minimal noise, but noticeably lose recognition accuracy in low signal-to-noise conditions. The requirements for the accuracy of speaker identification for such systems set a certain bar, which increases every year. Improving authentication accuracy extends the scope of such systems, including biometric multi-factor authentication systems, remote banking, access control systems, and more. Thus, to meet the needs of these systems consumers, the high requirements for the speaker recognition accuracy are required.