Tether Unveils Offline AI Translation Tools for 19 African Languages

Tether has made significant strides in expanding its operations beyond its stablecoin business. The company has released open-source AI translation models that support 19 African languages, allowing for offline translation on phones, laptops, and edge devices.

The TranslatePsy-AfriSLM family of models includes Hausa, Swahili, Yoruba, Amharic, Zulu, Somali, and Wolof, among others. The smallest model has been shown to outperform larger systems in translation benchmarks, achieving a 96% reduction in low-quality training data through a new filtering method.

The release of these models marks an important step for Tether's QVAC strategy, which aims to connect private computing with its existing digital payment operations. The open-source nature of the models allows developers to inspect and adapt them, paving the way for independent field testing in various dialects and communication settings.

The potential applications of these tools are vast, from education to healthcare and humanitarian aid. Schools could use them to translate lessons and materials into local languages, while farmers could receive crop guidance and weather instructions without relying on continuous connectivity.