DeepFake Audio Detection Details 

Detects whether an audio is real  or deepfake

Project Description 

This project focuses on building a web application for detecting deepfake audio (voice) using deep learning techniques. The system analyzes voice samples and identifies whether the audio is genuine or synthetically generated. A deep learning model based on LSTM (Long Short-Term Memory) networks is trained to learn temporal patterns in speech signals, making the solution effective for audio-based forgery detection. The application provides an easy-to-use interface where users can upload audio files and view prediction results. 

Features 

• Detection of deepfake or spoofed voice recordings

• LSTM-based deep learning model for sequential audio analysis

• Audio feature extraction using Librosa

• Web-based interface for easy audio upload and testing

• Visualizations for audio features and model performance

• Suitable for final-year and research-oriented projects

Tech Stack

• Python

• TensorFlow

• Librosa 

• LSTM (Deep Learning Model) 

• Flask (Backend)

• HTML, JavaScript (Frontend)

• other supporting libraries

System Requirements

• Python 3.10+

• Anaconda

• Minimum 8 GB RAM 

• GPU is optional

Deliverables 

• Jupyter Notebook

• Dataset

• Setup over video call 

• WhatsApp support