Published Paper Details:

The rise of AI-generated voices has blurred the line between synthetic and human speech, raising concerns in areas such as voice authentication, media integrity, and human-AI interaction. This project aims to develop a deep learning-based system capable of distinguishing between AI- generated and real human voices from uploaded MP3 files. Leveraging advanced text-to speech models like Tortoise TTS to generate AI voice samples, these are compared to real human voice recordings collected from public datasets. Audio features such as Mel-frequency cepstral coefficients (MFCCs), pitch, and spectral contrasts are extracted using the Librosa library, and a Random Forest classifier is trained to recognize patterns that differentiate AI from human voices. The system will be deployed through a user-friendly web interface where users can upload audio files and receive a classification indicating whether the voice is AI-generated or human. Future expansions include real- time detection, support for multiple languages, and the integration of voice emotion analysis, offering practical applications in securing voice-based systems and maintaining the authenticity of audio content.