About the project
Fake and synthetic media are becoming a well-known issue in society, from AI regulation and AI safety to protecting democracy from election interference.
This exciting PhD project investigates two technical aspects of voice deepfakes:
- What is it about a particular speaker’s voice that makes them easy or difficult to deepfake?
- How can hidden information be embedded into the speech signal to make it impossible to create deepfakes of a person’s voice?
You will explore a variety of deep learning solutions, including voice synthesis, audio watermarking, and voice spoofing detection.
This project is funded by the Centre for Doctoral Training in Complex Integrated Systems for Defence & Security (CISDnS). They are recruiting candidates interested in digital, physical, and biological systems to create a diverse and interconnected training environment.
Besides training with a world-leading research group, a CISDnS membership will give you the skills and knowledge to handle real-world interdisciplinary challenges through a Systems Thinking approach.
You will learn about the wider challenges of research and innovation within the Defence & Security sector.