Deepfake Audio Detection Using Hybrid Random Forest and Convolutional Neural Network Architecture

Abstract: The rapid evolution of speech synthesis and voice conversion technologies has enabled the generation of highly realistic synthetic speech, commonly referred to as deepfake audio. While such technologies offer innovative applications in media and accessibility, they also introduce serious threats to security, privacy, and information authenticity. This paper presents a hybrid deepfake audio detection system that combines classical machine learning and deep learning techniques to identify spoofed speech. The proposed framework integrates a Random Forest classifier trained on Mel- Frequency Cepstral Coefficients (MFCCs) and a Convolutional Neural Network (CNN) trained on Log-Mel Spectrogram representations. The system is implemented as a standalone desktop application with real-time visualization support. Experimental results demonstrate that the hybrid approach achieves high classification accuracy while maintaining computational efficiency suitable for consumer-grade hardware. The proposed solution aims to provide an accessible and reliable tool for combating synthetic audio misuse.

Keywords: Deepfake Audio, Audio Spoofing Detection, MFCC, CNN, Random Forest, Spectrogram Analysis, Anti- Spoofing