Chakma Language Technology
The Institute of Policy Dynamics (IPD) is building Chakma (𑄌𑄋𑄴𑄟) language technology one working instrument at a time — keyboard input first, then OCR, corpus, model, speech, and dictionary.
Six workstreams
Linux Keyboard & Transliteration Engine
A native Chakma keyboard layout for Linux, paired with a bidirectional engine converting text to and from Bengali Eastern Nagari.
Open the Chakma keyboard and transliterator →Chakma OCR
Optical character recognition for the Chakma script (Ojhapath), trained across both synthetic renders and real scanned documents.
Corpus Creation Drive
A community collection effort assembling the written Chakma corpus that every downstream model in this program depends on.
Chakma LLM & Evaluation
A language model for Chakma with a benchmark suite, built on a growing Chakma–Bengali parallel corpus.
Speech Corpus — 200 Hours
200 hours of annotated spoken Chakma audio, laying the ground for the script's first speech systems.
Structured JSON Dictionary
A machine-readable Chakma dictionary, structured for direct use in NLP pipelines rather than static lookup.
Six workstreams, one script. Keyboard, OCR, corpus, model, speech, and dictionary — built in that order, in the open, as the template for every indigenous language IPD takes on next.