🧠 ML-Based Self-Identification of Mental Health

This academic mini project aims to detect and classify a person’s emotional state using voice and facial expressions. By combining speech and image data, the system attempts to provide an early indication of mental health status using machine learning techniques.

---

📌 Project Overview

The system is divided into three core modules:

1. Voice Emotion Recognition – Classifies emotions based on audio signals
2. Facial Emotion Recognition – Identifies emotions from facial images
3. Integrated Model – Combines predictions from both modules for better accuracy

This project was developed as a team effort during our final year of B.Tech in Information Technology.

---

🧩 Modules

🎙️ Voice Emotion Recognition

• Dataset: RAVDESS (not included in repo due to size)
• Features Used: MFCC, Chroma, Mel Spectrogram
• Model: CNN-LSTM
• Output: Emotion label (e.g., happy, sad, angry)

😀 Facial Emotion Recognition

• Dataset: FER2013 (publicly available on Kaggle)
• Model: CNN
• Output: Facial emotion classification into predefined categories

🔁 Integrated Model

• Merges predictions from both modules
• Gives a more reliable assessment of emotional state

---

🛠️ Technologies Used

• Python
• TensorFlow / Keras
• NumPy, Pandas
• Librosa (audio feature extraction)
• OpenCV (image processing)
• Scikit-learn
• Matplotlib (visualization)

---