🎙️ TrueVoxAI – Speech Emotion Recognition

TrueVoxAI is a Speech Emotion Recognition (SER) project that analyzes voice recordings to detect emotions such as Happy, Sad, Angry, Neutral, Fear, etc.
It combines Machine Learning + Audio Signal Processing + Web Deployment into one complete project.

---

🚀 Features

• 🎤 Record or Upload Speech through the web interface
• 🧠 ML Model (MLP Classifier) trained on speech datasets (RAVDESS/CREMA-D)
• 🎼 Feature Extraction using MFCC, Chroma, Spectral Contrast
• 🌐 Flask-based Web App with a simple UI
• ☁️ Deployment-ready with Vercel + Hugging Face Spaces

---

🏗️ Tech Stack

• Python (librosa, scikit-learn, numpy, pandas, matplotlib)
• Flask (backend + UI rendering)
• MLPClassifier from scikit-learn for classification
• Vercel / Hugging Face Spaces for deployment

---

📂 Repository Structure

• app.py – Flask App
• requirements.txt – Python Dependencies
• vercel.json – Vercel Deployment Config
• models/ – Trained ML artifacts
  • mlp_emotion_model.pkl – Trained Model
  • scaler.pkl – Feature Normalizer
  • label_encoder.pkl – Label Mapping
• templates/ – HTML UI
• static/ – CSS/JS Assets
• Notebooks/TrueVox.ipynb – Training Notebook

---

⚙️ How It Works

1. User uploads/records speech 🎙️
2. Audio features extracted using librosa 🎼
3. Features normalized using scaler.pkl 📊
4. Model predicts emotion using mlp_emotion_model.pkl 🧠
5. Result displayed on UI 🎉

---

▶️ Running Locally

# clone repo
git clone https://github.com/tannuiscoding/TrueVoxAI.git
cd TrueVoxAI

# install dependencies
pip install -r requirements.txt

# run Flask app
python app.py

🧑‍🤝‍🧑 Contributors

This project was built with ❤️ by:

Mansi	Tannu Choudhary