This project is titled as “Deep Fake Audio Detection Using Deep Learning”. The rapid advancement of artificial intelligence and deep learning technologies, deepfake audio has emerged as a significant threat in today’s digital world. It enables the generation of highly realistic synthetic voices that can closely imitate real individuals. Such audio can be misused for malicious purposes such as fraud, impersonation, spreading misinformation, and unauthorized access to voice-based systems. Therefore, detecting deepfake audio has become essential to ensure security, authenticity, and trust in digital communication.

This project proposes a deep learning-based approach for detecting deepfake audio using a hybrid model that combines Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks. The system begins with preprocessing the audio data, followed by feature extraction using Mel-Frequency Cepstral Coefficients (MFCC), which effectively capture the important characteristics of speech signals. The CNN model is used to extract spatial features from the audio representation, while the LSTM model analyzes temporal patterns and sequential dependencies in speech.The proposed model is trained and tested on a dataset consisting of both real and fake audio samples. The system is evaluated using performance metrics such as accuracy, precision, recall, and F1-score to ensure its effectiveness and reliability. The system is designed to provide an efficient, scalable, and robust solution for deepfake audio detection.