About Tapra
Tapra is a Tunisian AI lab building speech and voice technology for Tunisian Derja, the language Tunisians actually speak every day: Arabic mixed with French and English, written in Arabic script or in Arabizi.
What we build
Tapra-Scribe transcribes Tunisian audio and keeps the code-switching intact, with French and English words written in Latin script exactly where the speaker said them. Tapra-Scribe-Realtime does the same live, while someone is still talking. Tapra-Speak gives Derja a natural Tunisian voice for voiceovers, videos, and call flows. Tapra-Flip converts Arabizi to Arabic script and back.
Why Tunisian Derja
Most speech technology treats Arabic as one language and pushes everything toward Modern Standard Arabic. Tunisians do not speak like that. A meeting, a voice note, or a customer call moves between Derja, French, and English inside one sentence. Tapra is trained on real Tunisian speech and Arabizi so that transcripts read the way people talk, and synthetic voices sound like they grew up here.
Services and developer platform
Creators and teams use Tapra Speak Studio, Tapra Subtitles, Tapra Live Dictation, Tapra Audio Search, and Tapra Flip directly on the website. Developers use the same models through the Tapra REST and WebSocket API, the TypeScript and Python SDKs, and the Tapra MCP server for AI agents. Everything is billed from one prepaid balance in Tunisian dinars, with no subscription.
Privacy
Content sent to Tapra is processed to return the result you asked for. It is not used to train our models unless you explicitly agree, and it is never sold.