Deep learning has proven to be a successful tool for many complex problems in academia, industry, and healthcare. Some of the applications of deep learning include computer vision, huge data analytics, thyroid detection, lung nodule recognition, and human-level control. Nevertheless, software that endangers democracy, national security, and confidentiality has been created using advancements in digital technology. Among the deep learning, powered apps that have been making an appearance lately is deepfake. Therefore, the main way that deepfake systems can produce fake visuals is by substituting scenes or images, videos, and sounds that are indistinguishable from real ones by people. Our fingertips can now manipulate artificial speech, images, and videos thanks to a variety of technologies. Moreover, photo and video scams have become so realistic that it is difficult to tell the difference between real and fake content with the unaided eye. It might lead to many problems, from using fabricated evidence in court to swaying public opinion. It is essential to have tools that can help us distinguish reality considering such factors. This paper provides a thorough analysis of the literature on DL-based algorithm-based deepfake detection techniques. In this work, we classify deepfake detection methods according to the applications they are used for: hybrid multimedia detection, image, audio, and video detection.