WHO - Voice Identification System
CNN-based voice identification system using spectrograms to identify speakers from 3-second audio clips with data augmentation.
Project Overview
WHO is a sophisticated voice identification system built using Convolutional Neural Networks (CNN) that can accurately identify speakers from just 3-second audio clips using spectrogram analysis.
The system employs advanced data augmentation techniques on over 500 audio samples to improve model robustness and accuracy in speaker identification tasks.
This project strengthened expertise in CNN-based audio processing and demonstrated the practical application of deep learning in biometric identification systems.
Key Features
- 3-second audio clip identification
- CNN-based architecture
- Spectrogram analysis
- Data augmentation pipeline
- High accuracy speaker recognition
- Real-time processing capabilities
Technologies Used
Project Details
Client
Personal Project
Timeline
February 2024
Role
ML Engineer & Audio Processing Specialist
© 2026 Samarth Borade. All rights reserved.