Open, maintained inventory of NLP resources for Tunisian Arabic (aeb). Contributions welcome.
-
Updated
Oct 28, 2026 - Python
Open, maintained inventory of NLP resources for Tunisian Arabic (aeb). Contributions welcome.
The largest open-source dataset for Tunisian Arabic (Derja) NLP, featuring social media text, transcripts, and e-commerce data for LLM training and fine-tuning.
Whisper-small fine-tuned for Tunisian Arabic (Derja) speech recognition with French/English code-switching. WER 30.4% / CER 12.0%. Two models: full-precision Transformers, and a fast int8 version for real-time voice conversations with LLMs.
Tapra MCP server: Tunisian Derja transcription, subtitles, voices, translation and Arabizi for Claude, Cursor, VS Code and Codex
To associate your repository with the derja topic, visit your repo's landing page and select "manage topics."