Voice Biometric Authentication System

Overview

Developed a voice biometric authentication system that verifies user identity using speech characteristics. The system extracts Mel-Frequency Cepstral Coefficients (MFCCs) from audio samples and applies machine learning-based speaker verification to distinguish authorized users from imposters.

---

Features

• Voice-based user authentication
• MFCC feature extraction from speech
• Speaker verification using K-Nearest Neighbors (KNN)
• Web interface for audio verification
• JSON-based voiceprint storage
• Real-time authentication workflow

---

Tech Stack

• Python
• Flask
• Librosa
• NumPy
• Scikit-learn
• HTML

---

Project Structure

CyberSecurity/
│── app.py
│── index.html
│── voiceprints.json
│── fraudster1_1.wav
│── fraudster1_2.wav
│── fraudster2_1.wav
│── fraudster2_2.wav
│── test.wav

---

Workflow

1. Record voice samples.
2. Extract MFCC features.
3. Store voiceprints.
4. Compare incoming audio with stored voiceprints.
5. Authenticate or reject the speaker.

---

Applications

• User authentication
• Secure login systems
• Voice identity verification
• Cybersecurity demonstrations

---

Future Enhancements

• Deep learning-based speaker verification
• Noise reduction preprocessing
• Multi-user enrollment
• Real-time microphone support
• Cloud deployment

---