LipVerify uses analysis of lip movements (a.k.a. viseme analysis) during speech to uniquely identify a user. The technology can be used on any device with a standard forward facing camera. During the initial enrolment phase the system ‘learns’ the user’s unique viseme profile by asking them to speak a phrase/digit sequence. During a subsequent user verification the user is asked to speak/mime a randomly generated phrase/digit sequence. An algorithmic comparison is then performed between the verification and enrolment videos and the result used to determine whether or not the user is who they claim to be.
Our product incorporates a so-called ‘liveness’ test by ensuring that the user responds correctly to a randomly generated challenge phrase during each authentication attempt. This test prevents what are known as ‘replay attacks’ which involve pre-recorded video data designed to fool the biometric system. LipVerify’s inherent liveness test is a significant advantage when compared to other biometric systems especially when the authentication process is carried out remotely in an unsupervised environment.
Note that LipVerify does not use the audio input and thus the user can mime the challenge phrase if they wish to authenticate themselves silently, e.g. when in a public space.