Google Translate Expands to 24 New Languages
Google Translate has absorbed another 24 languages, giving it fluency in 133 of the world’s tongues. The tech giant’s new language options encompass more than 300 million speakers, and extend to many languages that few if any other voice assistants can converse in at the moment.
Lingua Google
The new collection of languages is focused on non-European languages. The list includes eight languages spoken largely in India and surrounding countries, three languages spoken by the indigenous people of Central and South America, two South Asian languages native to the Philippines and the Maldives, and 10 languages spoken across Africa. Notably, the list includes Kurdish, which is largely spoken by people in Iraq, as well as Krio, a creole technically based on English and spoken mostly in Sierra Leone.
Google leveraged a new technical approach to teach Google Translate the new languages called Zero-Shot Machine Translation. Where traditional AI models would need to see the new language translated into one it already understands, Zero-Shot relies just on text in the new language. So, instead of translating a collection of books from English to Kurdish, the AI simply took in Kurdish texts and used neural language identification models to teach itself the languages. Native speakers then helped refine the AI’s understanding through editing and correction.
“For years, Google Translate has helped break down language barriers and connect communities all over the world. And we want to make this possible for even more people — especially those whose languages aren’t represented in most technology,” Google Translate senior software engineer Isaac Caswell explained in a blog post. While this technology is impressive, it isn’t perfect. And we’ll keep improving these models to deliver the same experience you’re used to with a Spanish or German translation, for example.”
Upping Google Translate’s language count opens a lot of potential doors for Google as it tries to reach new potential customers. It also helps boost many of its related products. For instance, Google Translate’s real-time transcription feature and the instant translation feature for Google Assistant on Android are suddenly useful in a lot more contexts. The extra languages and the way Google added them may also help with Google’s Translatotron AI translation model project that recreates a speaker’s voice in a different language.
Follow @voicebotai Follow @erichschwartz
Google’s Translatotron 2 Improves Linguistic Shifts Without the Deepfake Potential
Amazon Unveils Speech Datasets for Alexa Skill Development in 51 Languages