Ghana NLP Notebooks

A curated collection of Jupyter notebooks for Natural Language Processing tasks focused on Ghanaian languages. This repository provides ready-to-use examples, tutorials, and research implementations for the Ghana NLP community.

---

Available Notebooks

Machine Translation (MT)

Notebooks for translating between Ghanaian languages and English/other languages.

Notebook	Description	Open in Colab
twi_english_translation.ipynb	Baseline Twi ↔ English translation using transformers
multilingual_ghanaian_mt.ipynb	Multilingual translation for 5+ Ghanaian languages
mt_evaluation_metrics.ipynb	BLEU, chrF++, and custom evaluation for low-resource MT

Text-to-Speech (TTS)

Notebooks for synthesizing speech from text in Ghanaian languages.

Notebook	Description	Open in Colab
StyleTTS2_Twi_Finetune_working_final.ipynb	Fine-tuning Style TTS English Base with a Twi Dataset
ga_tts_data_preparation.ipynb	Data preprocessing for Ga TTS dataset creation
tts_evaluation.ipynb	MOS evaluation and audio quality assessment

Automatic Speech Recognition (ASR)

Notebooks for transcribing Ghanaian language audio to text.

Notebook	Description	Open in Colab
twi_asr_whisper.ipynb	Fine-tuning Whisper for Twi ASR with code-switching
multilingual_ghanaian_asr.ipynb	Wav2Vec 2.0 for multiple Ghanaian languages
asr_data_augmentation.ipynb	Audio augmentation techniques for low-resource ASR

🛠️ Data Generation

Notebooks for synthetic data generation in Ghanaian languages.

Notebook	Description	Open in Colab
prestine_twi_dataset_generation.ipynb	Generating high quality multifaceted Twi text for training models
dataset_statistics.ipynb	Analyzing dataset distributions and quality

---

Quick Start

Running on Google Colab (Recommended)

Click any "Open in Colab" badge above to run the notebook immediately with free GPU/TPU access. No local setup required!

Running Locally

# Clone the repository
git clone https://github.com/GhanaNLP/notebooks.git
cd notebooks

# Create virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Start Jupyter
jupyter lab

---

Contributing

We welcome contributions! To add a new notebook:

1. Create your notebook in the appropriate task folder (mt/, tts/, asr/, or utils/)
2. Ensure it has a clear description and runs end-to-end on Colab
3. Add the Open in Colab badge (see template below)
4. Update this README with the new entry
5. Submit a Pull Request

Colab Badge Template

[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/GhanaNLP/notebooks/blob/main/FOLDERNAME/FILENAME.ipynb)

---

License

This project is licensed under the MIT License - see the LICENSE file for details.

Made with ❤️ for Ghanaian languages