News & Insights

Ideas, innovation and impact

The lead storyResearch1 min readYodiV3: NLP for Togolese Languages with the Eyaa-Tom Dataset and the Lom MetricPresents Eyaa-Tom, a multi-domain parallel corpus for ten Togolese languages and two lingua francas, and introduces the Lom metric to quantify each language's AI readiness.Read more

The editorial desk

Latest stories

Research1 min read

Eyaa-Tom: “Words of the people” in Kabiyè

A corpus collected in the field by a community of linguists and native annotators, designed to cover real domains: health, administration, everyday conversation.

By Community of linguists, researchers and Umbaji annotators

Research1 min read

Yodi: The Speech Recognition System for Ewè

Describes the architecture of the first speech recognition system for Ewè, built on word-to-waveform mapping and convolutional neural network classifiers.

By Umbaji research team