🌐 Language Translator App (English → French)

A many-to-many encoder–decoder sequence model built with LSTM to translate English to French in real-time. Developed following the DataFlair tutorial “Language Translation with Machine Learning”.

---

🧠 Overview

• Implements a sequence-to-sequence (seq2seq) model with LSTMs.
• Uses teacher forcing during training for more stable convergence.
• Provides both training and GUI modules for interactive translation.

---

🚀 Tech Stack

• Language: Python 3.8+
• Deep Learning: TensorFlow ≥2.2 (Keras API)
• Core Libraries: numpy, sklearn, pickle
• GUI: tkinter (via LangTransGui.py)
• Data: English-French parallel sentences (eng-french.txt)

---

📂 Project Structure

/
├── eng-french.txt          # Parallel corpus (training data)
├── langTraining.py         # Seq2seq model training
├── training_data.pkl       # Preprocessed training arrays
├── s2s/                    # Saved model weights, optimizer & metrics
├── LangTransGui.py         # GUI to load model and translate
└── README.md               # Project documentation

---

⚙️ Setup Instructions

1. Clone the repo

   git clone https://github.com/Saptha-Harsh/Language-Translation.git
   cd Language-Translation

2. Install dependencies

   pip install tensorflow numpy scikit-learn pickle5

3. Dataset
   Ensure eng-french.txt is present: each line contains English_sentence<TAB>French_sentence.

4. Train the model

   python langTraining.py

   • Trains on ~10,000 sentence pairs (adjustable).
   • Saves model and preprocessing in s2s/.

5. Launch the GUI translator

   python LangTransGui.py

   • Enter English text → click Translate → see French output.

---

🧩 How It Works

1. Preprocessing: Tokenize and vectorize input/output sentences into sequences.
2. Encoder: LSTM reads input English sentence.
3. Decoder: LSTM generates French translation, using teacher forcing during training.
4. Teacher Forcing: During training, the decoder gets true previous tokens for faster convergence.
5. Inference: GUI takes user input, tokenizes, and uses trained model to predict French output.

---

🛠️ Customization Tips

• Increase dataset size: Improve accuracy by using the full dataset.
• Tweak hyperparameters: Try different batch_size, epochs, or LSTM hidden dimensions.
• Add attention: Boost performance by integrating attention mechanisms.
• Expand to other languages: Substitute dataset, update tokenizers, retrain model.

---

📚 References

• DataFlair Project: https://data-flair.training/blogs/language-translation-machine-learning/
• TensorFlow Keras Seq2Seq Documentation
• Concepts of Teacher Forcing and LSTM-based Translation Models

---

🤝 Contributions & Contact

Contributions, suggestions, or improvements are welcome! Feel free to:

• ⭐ Star & fork the repo.
• 🐛 Report issues or suggest features.

---

Enjoy experimenting & translating!