Taleka
Preserving Semai through elder voices, community stories, and AI-guided language learning.
Problem Overview
Semai has limited digital learning resources and support in mainstream language tools. Preserving it means giving community elders a way to share their voices and helping learners practice with language the community can verify.
Taleka connects both sides in a mobile platform. Elder Studio records and transcribes spoken Semai, then lets elders review it against a curated lexicon. Verified recordings become illustrated stories, while learners practice vocabulary, play learning games, and ask a grounded AI Coach for guidance.
The project won Champion at BorNEO HackWknd 2026 among 103 teams from 7 ASEAN countries.
My Role
My work focused on connecting language preservation with practical learning tools. Key areas included:
- Building the mobile-first experience with Ionic, React, and Capacitor
- Connecting Elder Studio voice recordings to OmniASR transcription and lexicon-based review
- Grounding translation and AI coaching in verified Semai vocabulary and sentence examples
- Turning verified recordings into illustrated stories and supporting vocabulary practice and learning games
- Integrating Supabase for authentication, recording storage, and learner progress