Time-varying voiceprint authentication method and system based on online learning

Abstract

The invention discloses a time-varying voiceprint authentication method and system based on online learning, wherein the method comprises the steps: randomly generating a prompt text, prompting a userto read the prompt text, and collecting voiceprint features in an audio read by the user; and comparing and analyzing the acquired voiceprint features with previously registered user voiceprint features, judging that verification is passed when the voiceprint features are consistent, verifying the identity of the user through a password if the verification is not passed, and updating user voiceprint information in a registered voiceprint database through an online learning method if the password is successfully verified. The method has the advantages that compared with the prior art, the identification accuracy is improved, and the practicability and safety of the system are improved; the user voiceprint information in the registered voiceprint database can be updated through the online learning method, so that the user is allowed to update the voiceprint registration information through the password, and the usability of the system is improved.

Description

technical field [0001] The invention relates to a time-varying voiceprint authentication method based on online learning, which belongs to the technical field of biological information security. Background technique [0002] Voiceprint Recognition (Voiceprint Recognition), also known as speaker recognition, is divided into speaker identification and speaker confirmation. The former is used to judge which one of several people said a certain speech; while the latter is to judge whether a certain speech is spoken by a specific person. Different tasks require different techniques. [0003] Voiceprint password verification is an application of speaker confirmation, which requires the user to input voice and confirm the identity of the speaker accordingly. In this application, user registration and identity verification usually use voice input of fixed password text (text-dependent), which eliminates the difference in voice content, so the voiceprint is often more consistent, a...

AI Technical Summary

Problems solved by technology

[0004] The disadvantages of the existing voiceprint password verification technology include: the voice data templates required for registration and verification in the text-independent mode are many, the verification process is slow, it is not suitable for real-time, user voiceprint password verification, and the user experience is poor; The classic mode has the risk of being eavesdropped, and is more likely to be attacked by recording playback; the voiceprint of the user is easily affected by other factors, such as age and other reasons, which will cause subtle changes in the voiceprint and reduce the matching similarity, thus affecting Matching success rate, affecting user usage

Images