Google Translate has added 15 more African languages on Voice Search, typing with voice on Gboard and voice input on Translate enabling approximately 300 million more people across Africa to use voice input across these products to communicate with friends and family.
Google already supports typing input with custom keyboards in Gboard for around 200 African languages and varieties, with automatic translation with Translate for over 60 languages spoken in Africa.
In a statement, Alex Okosi, Managing Director of Google Africa said: “This technology will make a difference to over 300 million more people across the continent – enabling them to interact with the web with just their voice. With teams in Google Accra working on this, it’s one example of how Google in Africa is building technology for Africans – and for the world”.
Today, along with the 13 African languages which already enjoy talk-to-text support on Gboard and Voice Search, Google is excited to extend voice input to 12 more languages on these platforms. Google is also extending voice input to 13 more African languages on Google Translate – along with the 9 already supported. Voice input on Translate enables people to dictate to Translate and get translations in other languages – while talk-to-text support on Gboard and Voice Search allows people to type with voice anywhere where Gboard is enabled, or Search on Google using their voice.
These languages are spoken by an estimated 300 million people across West, East and Southern Africa. In West Africa, Google now supports Twi, one of the most widely spoken languages in Ghana, as well as 4 major languages of Nigeria, a country with over 500 languages and 218 million people. Hausa, Yoruba, Igbo and Nigerian Pidgin are spoken by an estimated 129 million people, or around 60% of Nigeria’s population. Google is also expanding its offering on Voice Search and Gboard in Kenya with the addition of Kikuyu, alongside Swahili – as well as for wider East and Southern Africa.
These additions will bring Google’s total to 25 languages with voice support in Africa – and 94 languages total across the world.
This language extension has been made possible by advances in AI, specifically multilingual speech recognition – which converts speech into text. The AI model learns languages in the way a child would – learning to associate certain speech sounds with the specific sequences of character in the written form. Multilingual speech recognition models are trained on data from multiple languages, and then are able to transcribe speech into text in any of those languages.