A SYSTEM AND A METHOD FOR SECURE SPEAKER VERIFICATION

A method for secure speaker verification, the method comprising the steps of: generating (501) a random, textual prompt for a user to speak; awaiting (502) for a user to speak the generated, textual prompt and recording the corresponding speech; in parallel to said speech recording (502), recording...

Full description

Saved in:
Bibliographic Details
Main Authors Chochowski, Marcin, Gajewska-Dendek, Elzbieta, Tkaczuk, Jakub, Bronakowski, Lukasz
Format Patent
LanguageEnglish
French
German
Published 19.04.2017
Subjects
Online AccessGet full text

Cover

Loading…
More Information
Summary:A method for secure speaker verification, the method comprising the steps of: generating (501) a random, textual prompt for a user to speak; awaiting (502) for a user to speak the generated, textual prompt and recording the corresponding speech; in parallel to said speech recording (502), recording (503) ultrasound signals reflected from the user's face; subjecting (504) the recorded speech to automatic speech recognition in order to obtain a textual representation of the speech; detecting (505) user's liveness based on a correspondence of said recorded audio signal with said recorded ultrasound signal as well as based on the automatically recognized speech; executing (506) a text accuracy validation on said recorded speech subjected to the automatic speech recognition in comparison to the generated, textual prompt; and executing an authentication (507) of the user based on results of said detecting (505) user's liveness, and said text accuracy validation in order to verify whether the user was speaking the generated prompt, at the time of recording.
Bibliography:Application Number: EP20150461566