Final Year Project

Project Title

Yaran-Burushaski Translator

Description

This project develops an end-to-end speech-to-English translation system for Burushaski, a critically endangered and low-resource language spoken in the mountainous regions of Pakistan. The core aim is to address the severe lack of digital linguistic resources for Burushaski by building an annotated speech dataset, fine-tuning state-of-the-art automatic speech recognition models, and deploying a functional translation pipeline accessible to researchers and communities alike. The project contributes to language preservation efforts by creating tools that document and digitize Burushaski speech across dialects, including Yasin and Hunza. Beyond preservation, it benefits the broader NLP research community by establishing a foundation for future low-resource language work. Ultimately, this research bridges the gap between endangered linguistic heritage and modern artificial intelligence.