Derin Sahte
Production of artificial intelligence that places a person’s face or voice in a realistic way to another content.
Deepfake is the production of artificial intelligence, which is derived from the combination of words "deep learning" (deep learning) and "fake" (sahte), using deep learning techniques. Technically often used manufacturer traction networks (GAN) or autocoders: a model learns the facial characteristics of the target person from a large number of photos or video frames, then the properties learned from another video are "bined by preserving light, angle and mimics." Sound cloning techniques can produce artificial sounds, similarly, for example, which can mimic the person’s speech tone, accent and rhythm.
Although Deepfake technology has creative uses in the entertainment industry and education, the most identity scammers, dezenformation and non- consent content production concerns; a fake politician video can be misleaded or a fake CEO voice can be used in corporate fraud. These risks lead to the rapid development of deepfake detection technologies (models looking for inconsistencies in video and audio or digital traces), which creates a continuous "magic race" among production and detection technologies.
Although Deepfake technology has creative uses in the entertainment industry and education, the most identity scammers, dezenformation and non- consent content production concerns; a fake politician video can be misleaded or a fake CEO voice can be used in corporate fraud. These risks lead to the rapid development of deepfake detection technologies (models looking for inconsistencies in video and audio or digital traces), which creates a continuous "magic race" among production and detection technologies.