"Lezgi Ghed" Award — AI for Language Preservation
Honored with the Lezgi Ghed award for building the first Lezgin language AI translator and text-to-speech system — preserving a minority language through technology.
Lezgi Star Award
Our team was honored with the “Lezgi Ghed” (Lezgi Star) award by FLNK for preserving the Lezgin language through AI.
What we built
- The first Lezgin translator — a neural machine translation model between Lezgin and Russian, trained on community-collected data
- Text-to-speech system — 30 hours of studio-recorded speech data powering the first Lezgin TTS model
Why this matters
Lezgin is spoken by roughly 800,000 people, primarily in Dagestan and Azerbaijan. It has limited digital presence and is classified as a vulnerable language by UNESCO. Before our project, there were no AI tools for Lezgin — no translator, no TTS, no digital corpus.
We changed that. With 1,000+ volunteers contributing data, we built tools that make the language accessible to a new generation. Everything is open-source on HuggingFace.
This project proves that modern AI isn’t just for major languages — with the right community and architecture, even a language with a small speaker base can get state-of-the-art tools.
The 2.0 release of the translator — trained on a 200K synthetic corpus — is covered in detail: First Lezgin Translator 2.0.