Using Recurrent and Convulation Neural Networks to Indentify the Fake Audio Messages
Артем Ховрат, Volodymyr Kobziev · 2023
With the improvement of deep fake technology, the problem of detecting fabricated audio has gained significant proportions. And if in peacetime, people, in the majority, can detect a fake by themselves, due to the imperfection of the generation algorithms, during crisis events, it becomes much more difficult to do this. The article is devoted to determining the effectiveness of using recurrent and convolutional neural networks to classify the content of audio information to detect fakes. The results of the research conducted on a self-created data set dedicated to Russia's invasion of Ukraine allow us to assert the high efficiency of using the selected models.