Google has added support for 30 more language variants to its Cloud Speech API, with a focus on emerging markets in Africa and India.
It has added support for Swahili and Amharic, as well as English variants for Ghana, Kenya, Nigeria, and Tanzania.
TechCrunch reported that people in those regions will be able to type and search the web by voice using Gboard, Google’s keyboard app.
Its Cloud Speech API has also gained several new features, including timestamps for every word in a transcript and support for files up to three hours long.
The latest improvements follow Google adding Zulu and Xhosa to the languages it translates using an artificial neural network.
Previous translations were based on programming an understanding of the grammar of a language into the translator.
The switch to neural translation for Zulu and Xhosa means that translations will be closer to spoken language, said Google.