Google launches new language features for Indian users

Google also unveiled its new multilingual model ‘MuRIL’ that supports 16 Indian languages.

Google India 

Google today announced four new language features in India at its L10n event. The new features add support for more Indian languages across different Google products. Google also announced its new multilingual model ‘MuRIL' developed in India.

Google Search in India shows results in both English and Hindi. Users can toggle between the two languages for search queries with just a tap. Google said this has seen a 10x increase in Hindi queries in India. It is now adding support for four more Indian languages - Tamil, Telugu, Bangla and Marathi.

Google Search gets support for more Indian languages. (Google)

Google Search gets support for more Indian languages. (Google)

Google Search will also start showing results in more Indian languages. This will work for local language queries typed in English, and it's also aimed at benefiting bilingual users who would prefer reading articles in both English and an Indian language. The new update will start showing over the next month, and in Hindi, Bangla, Marathi, Tamil and Telugu.

Google is also making it easier for users to change their preferred language on Google Assistant and Discover. Users can do this by going to the app settings, and selecting the language they prefer. Google said a third of Assistant users in India use the app in an Indian language, and over 50% of users view Discover content in Indian languages.

Google Lens Homework (Google)

Google Lens Homework (Google)

Google Lens is getting an interesting feature called “Homework” that can solve math problems. This feature is available in Hindi and English. All users need to do is snap a photo of a math problem using Google Lens, and it will show step-by-step guides and videos to explain the problem.

MuRIL

MuRIL stands for Multilingual Representations for Indian Languages. This new multilingual model can understand and interpret Hindi text written in Roman script better than previous models. According to Google India, the sentence “Achha hua account bandh nahi hua” would have previously been interpreted as something negative but MuRIL can correctly identify this as a positive statement.

MuRIL is free and open source on TensorFlow Hub, and it currently supports 16 Indian languages and English.