GhanaNLP/ghanaian-nlp-datasets-models

This repository curates NLP datasets and models for languages spoken in Ghana. # Ghanaian NLP Datasets & Models This repository curates **NLP datasets and models** for **languages spoken in Ghana**. It is maintained by Ghana NLP, with the goal of supporting research and development of natural language processing for all Ghanaian languages. Join Ghana NLP to stay involved. --- ## Table of Contents Click to jump to any language: ### Languages with Data - Akan (aka) - Dagbani (dag) - Ga (gaa) --- ### Akan (aka) #### 📁 Datasets | Name | Description | Link | |------|-------------|------| | Twi-English Parallel Sentences | Twi and English Aligned translation pairs | View | | Fante Speech Transcribed | Transcribed multi-speaker speech dataset | View | | Twi Transcribed | Asante Twi Bible single speaker transcribed dataset (split at verse level) | View | | Twi Transcribed | Asante Twi Bible single speaker transcribed dataset (split at utterance level) | View | #### 🤖 Models | Name | Description | Link | |------|------|------| | ABENA | BERT model for Asante Twi - cased 1 , uncased 2 and distilled uncased 3 , and Akuapem Twi - cased 4 . | 1 \| 2 \| 3 \| 4 | | Akan Whisper | Speech recognition model for Akan | View | | Asante Twi Speech Recognition | Speech recognition and transcription model for Asante Twi | View | --- ### Dagbani (dag) #### 📁 Datasets | Name | Description | Link | |------|-------------|------| | Dagbani Orthography | Spelling guide corpus | View | #### 🤖 Models | Name | Task | Framework | Link | |------|------|-----------|------| | DagBERT | Language modeling | Transformers | GitHub | --- ### Ga (gaa) #### 📁 Datasets _(No entries yet — contribute!)_ #### 🤖 Models _(No entries yet — contribute!)_ --- ### Languages without Data (Awaiting Contributions) - Abron (abr) (No data) - Adamorobe Sign Language (ads) (No data) - Adangbe (adq) (No data) - Adele (ade) (No data) - Ahanta (aha) (No data) - Akposo (kpo) (No data) - Animere (anf) (No data) - Anufo (cko) (No data) - Anyin (any) (No data) - Avatime (avn) (No data) …