← Back to Portfolio

Speech Emotion Recognition

Project Screenshot

Speech Emotion Recognition

This project introduces a real-time Speech Emotion Recognition (SER) system to assist cybercrime police in analyzing live emergency calls. It uses a CNN + NLP ensemble model to detect emotions such as fear, distress, or anger, helping prioritize urgent cases and enhance decision-making during investigations.

Key Features:

• Real-time emotion detection: On incoming emergency calls.

• CNN for audio-based analysis: And NLP for text-based analysis.

• Live dashboard: For displaying detected emotional state.

• Audio preprocessing: Using FFmpeg.

• Multilingual support: For emotion detection in multiple languages.

Tech Stack:
Python, TensorFlow/Keras, FFmpeg, Asterisk VoIP Server, Web Dashboard (HTML/CSS/JS)