"[W]e are trying to reduce this gap both for synthetic content detection and also the detection of replicated content”, describes Luca Cuccovillo his and his teams' motivation and goal within the vera.ai project.
The gap Luca is referring to can be derived from the facts that there was “… for several years a lot of research in the detection of synthetic images but the detection of synthetic speech was lacking” and that “the speech was not yet that high quality, so it wasn't really interesting.”
What does that mean for the work of Fraunhofer IDMT? How exactly is the team tackling these challenges and narrowing this gap? And what can journalists, factcheckers and other potential users expect from this work? – Watch the full video below to get answers and further interesting insights!
Author: Anna Schild (DW)
Editor: Jochen Spangenberg (DW)