About
Gold-quality language data, built with the people who speak the language.
Sabdakunja began with the languages of Nepal and the Himalaya. The method generalises: wherever a language is spoken but not written into technology, the same platform applies.
Our mission
Documentation
Comprehensive digital records of languages and cultural heritage.
Preservation
Sustainable methods that protect linguistic diversity long term.
Empowerment
Communities keep ownership of their language and their data.
Principles
Consent before capture
No recording enters a dataset without informed, documented, revocable consent from the speaker and community.
Quality over volume
Every hour is transcribed and reviewed twice. A small verified corpus outperforms a large noisy one.
Dialects are not errors
Variation is tagged and preserved rather than normalised away into a single prestige form.
Give the archive back
Communities receive their recordings, transcripts and learning material — not just a model they cannot inspect.
Let's document a language together.
Reach out with the language, the community and the outcome you want — we will tell you honestly what is possible.
Start a conversation