Method for performing voiceprint verification of Chinese through single arithmetic

Abstract

The invention relates to a method for performing voiceprint verification of Chinese through single arithmetic. The method comprises the following steps that: when the voiceprint enrollment is performed, a user is required to speak from zero to nine, randomly speak one of arithmetic operation symbols, namely plus, minus, multiplication, division, and finally speak equal; and when the voiceprint verification is performed, the user is required to read after a randomly generated simple arithmetic problem for voiceprint verification. By the technical scheme, compared with a text-dependent random mode (Text-Prompted), the method has the advantages that: the arithmetic operation symbol is added during voiceprint enrollment, the content and character strings which are read after by the user during voiceprint verification are increased, and the accuracy rate of voiceprint verification of Chinese is improved; and the user reads after the simple arithmetic problem, and the definite correlation exists among figures so as to facilitate the memorization of the user, so that a user experience of the voiceprint verification can be improved.

Description

technical field [0001] The invention relates to the field of biological feature identification, in particular to a method for verifying Chinese voiceprints by using simple arithmetic. Background technique [0002] Voiceprint verification is a way to verify identity based on the uniqueness of personal voice biometrics. Voiceprint verification has two links: registration (Voice Enrollment) and verification (Voice Verification). According to the correlation between the content of verification and registration, voiceprint verification technology is divided into two categories: Text-Dependent and Text-Independent; Text-Dependent is based on Whether the speech content during verification is completely consistent with that during registration is divided into Classic mode and Text-Prompted mode. [0003] The text-independent mode (Text-Independent) needs to collect more voice data during registration, and requires more user voice data during verification. At the same time, the ver...

AI Technical Summary

Problems solved by technology

However, because Chinese syllables are shorter than English syllables, the application of this method in Chinese voiceprint verification will lead to an increase in the voiceprint verification error rate FAR / FRR (False Accept Rate / False Reject Rate), which cannot meet the accuracy requirements of voiceprint verification

If the random number string is increased, the user needs to speak more words during registration and verification. Although the error rate during voiceprint verification can be improved, it is not easy for the user to memorize and read a string of irregular random number strings. When the user follows the reading, it will cause memory impairment to the user, thereby reducing the user experience of voiceprint verification

Images