It's Not as Bad as You Think: Detecting AI-Generated Voices

Advances in machine learning have opened up the world to a brand new frontier of fraudulent phone calls which the average person may not be in any way prepared for. From imitations of a loved one's voice to lifelike mimicry of human callers, telephone scams may become harder than ever to anticipate or prevent now that criminals have the help of AI on their side. This is why in my research paper, I aim to analyze and compare two existing methods of detecting the authenticity of human voice recordings in order to demonstrate and explain currently available technology that's capable of combating the potential threat of AI-generated scam calls. The paper aims to review previously published literature on the technologies used for voice authenticity detection, and describe how these technologies function in a way that novice students or even lay people can understand. The main goal of the paper would not be to condemn the use of AI technology all together but rather to demonstrate lesser-known tech in the machine learning field that's capable of assuaging fears common people may have about the future of AI. Currently, the paper compares the raw audio waveform analysis used by the RawNet2 neural network against a simple MFCC detection program meant for voice recognition, highlighting the principles behind their functionalities and their respective margins of error. The paper's analysis is aimed to highlight the conclusion that with these existing technologies, the future threat of AI-generated scam calls is not unavoidable.