multilingual-emotion-detection

Multilingual Emotion Detection in Voice

This project focuses on detecting emotions from voice recordings across multiple languages using machine learning and deep learning techniques.

📌 Project Overview

The goal is to classify emotions such as happy, sad, angry, neutral, etc., from audio speech signals, leveraging multilingual datasets and applying feature extraction techniques and neural networks.

🔍 Features

Preprocessing of audio data (e.g., MFCC extraction)
Multilingual dataset support
Deep learning models (e.g., CNN, LSTM)
Emotion classification
Model evaluation and visualization

🧰 Technologies Used

Python
Jupyter Notebook
Librosa (audio processing)
Scikit-learn
TensorFlow / Keras or PyTorch (depending on the notebook)
Pandas, NumPy, Matplotlib

📁 Project Structure

multilingual_emotion_detection_in_voice.ipynb
data/                    # Directory for audio datasets
models/                  # Saved model files (optional)
README.md
requirements.txt         # Python dependencies (optional)

🚀 How to Run

Clone the repository or download the notebook.
Install required dependencies:
```
pip install -r requirements.txt
```
Ensure datasets are available in the correct directory structure.

Open the notebook:

jupyter notebook multilingual_emotion_detection_in_voice.ipynb

Run all cells in sequence to train and evaluate the model.

📊 Datasets

This project can be used with publicly available datasets such as:

RAVDESS (English)
Emo-DB (German)
CREMA-D (American English)
TESS (Canadian English)

Make sure to adjust preprocessing steps if your dataset differs.

🧠 Model Training

Models are trained using extracted MFCCs and other relevant features from audio. Performance is evaluated using accuracy, confusion matrix, and other classification metrics.

📈 Evaluation

Confusion Matrix
Accuracy/Precision/Recall
Visualization of training history

📬 Contact

Author: Ashika M Email: ashikasjcetcse@gmail.com ✉️ GitHub: https://github.com/ashika67 👨💻