Audio + Video Deepfakes
Creating visual deepfakes with free-form audio input
In the last blogpost, we saw how an attacker can synthesize any voice they wish, cloning for example the German changellor Angela Merkel.
This technology can be augmented with video frames, adapted to match the audio. Watch the following video to see the system in action.
It is important to inform people about this technology and raise awareness, such that malicious activity (i.e. fake news) does not meet an uneducated public. It is to be expected that this technology will become more prominent and more widely spread in the near future.