Lingo turns spoken contributions into open voice data. Researchers launch a campaign, speakers record short phrases, the community verifies them — and everyone earns rewards. Together we build the corpora that train tomorrow's translation models.
A simple loop, designed for low bandwidth and people who speak a language but may never have written it.
See a prompt in a language you read, then hold to speak it in your mother tongue. Re-record until it feels right.
Listen to others' recordings and rate whether they match the prompt. Consensus decides what enters the corpus.
Quality contributions earn points from the campaign budget — redeemable for cash, mobile money, or goods.
Pick a target language, set a points budget, and import your prompts from a CSV — or generate culturally-adapted ones with AI. Invite speakers and verifiers by role with a single link, then watch verified recordings flow into a clean, exportable dataset.
Lingo began as lingo.cm — a cost-efficient, French-pivot machine-translation service for Cameroonian languages. We compiled its corpus by hand from every written source we could find — books, pamphlets, scanned booklets, and our open cameroon_bibles dataset — unified under a single alphabet (AGLC). Those models are open and live. The voice project is the next chapter — but the foundation is yours to use and download.