Dr Michael Wagner
This PhD project by Dr Girija Chetty investigated multi-modal approaches, specifically auditory-visual approaches, for the purpose of speaker identity verification, a problem in the area of biometrics.
In her PhD project, Dr Girija Chetty under the supervision of Prof. Michael Wagner investigated various multi-modal approaches in the area of speaker identity verification, which is a sub area of biometrics involved in the verification of a person through voice analysis. In particular, this project investigated the combination of auditory (voice) data and visual (lip) data. This allowed research in the area of liveness detection, designed to counter spoofing attacks on biometric identification systems where impostors try to gain access to the system by showing a picture of and playing the voice of an enrolled user. As part of this project, audio-video data of Australian TV news broadcasts were recorded in the University of Canberra Broadcast News dataset.