Google has updated its Translate app on iOS and Android with offline neural machine translation or NMT supporting 59 different languages. Neural machine translation is a kind of machine translation that uses artificial neural networks to build predictive models for likelihood of sequence of words in a sentence.
This will ensure that the quality of translation is very high even when the device is not connected to the internet and all the computations required to do efficient translation is available offline to those devices. Update will be rolled out to a smaller segment of users initially and over the course of next few weeks it will be rolled out to all the users.
Earlier the computation required to do translation was done over the cloud and not on the client device, but due to the advancements in software and hardware devices, mobile phones are capable of running highly advanced machine learning models without the need of being connected to the cloud. Microsoft released a similar update to its Translate application with support for about 12 languages. Companies like Google and Microsoft have found ways to compress these machine learning models to smaller manageable sizes such as 30 to 40 megabytes per language in Google's case.
Here is a full list of languages :
Afrikaans, Albanian, Arabic, Belarusian, Bengali, Bulgarian, Catalan, Chinese, Croatian, Czech, Danish, Dutch, English, Esperanto, Estonian, Filipino, Finnish, French, Galician, Georgian, German, Greek, Gujarati, Haitian, Creole, Hebrew, Hindi, Hungarian, Icelandic, Indonesian, Irish, Italian, Japanese, Kannada, Korean, Latvian, Lithuanian, Macedonian, Malay, Maltese, Marathi, Norwegian, Persian, Polish, Portuguese, Romanian, Russian, Slovak, Slovenian, Spanish, Swahili, Swedish, Tamil, Telugu, Thai, Turkish, Ukrainian, Urdu, Vietnamese and Welsh.