Synthesising Obama

SYNTHESISING OBAMA – is the title of the project in which a group of computer scientists from the University of Washington were able to train a neural network to 'learn' Barack Obama's voice and facial movements in order to accurately and convincingly create a video of a 'fake' Obama reading a fake speech. The neural network required weeks of training in order to synthesise Obama but the technology (along with that of other similar projects such as face-swapping neural nets) indicates a future in which such a process could be performed in minutes and would put the credibility and reliability of any video footage at stake. Video is perceived as our most trust-worthy evidentiary media, analysed and poured over by conspiracy theorists, security and police services equally in the hunt for truth in the information. In an age of 'Fake News' where we're asking questions of the objectivity and reliability of all media, the potential for quick, computationally cheap fakery raises concerns today for the reliability of our media.

• https://tobiasrevell.com/Synthesising-Ob...