Microsoft Updates Translator App with Conversational Translation Feature for Groups

The new version of Microsoft Translator now enables translation in real time, supporting up to 100 speakers. This new feature allows face to face communication using native language on personal devices.

By Sead Fadilpasic

December 14, 2016 2:59 pm CET

Microsoft Translator supports 60 languages, within that voice translation for nine languages and type-in translation for the rest of them.

The updated app now translates spoken conversation in real-time with as many as 100 participants. Because of this new feature, people are able to interact with different languages directly face-to-face using their own language.

Participants can join the conversation using their smartphone, tablet, or PC by entering a code or scanning a QR code. Microsoft states there are multiple ways you can use the new app, such as:

Be part of the conversation: Join a conversation regardless of your language. Chat. Share experiences. Make a connection.
Immerse as you travel: Interact with ease when traveling internationally. Meet locals. Make new friends.
Present to a wider audience: Break the languages barrier and make your message heard by everyone in the room.
Make everyone feel welcome: Greet anyone regardless of the language they speak. Provide people with needed information.

A personal universal translator

Microsoft Translator is powered by the same, cutting-edge speech technology used by Skype Translator. The service allows interaction in real time over the internet while speaking different languages.

However, one thing it can’t do is face-to-face communication. That limitation is why Microsoft is pushing the Translator app by integrating the existing technology onto personal devices.

The technology uses deep neural network-based translations. These present a more natural, human-sounding translations than the previous technology known as statistical machine translation.

Microsoft Translator live devices Image credit: Microsoft — Microsoft Translator live devices
Image credit: Microsoft

Both methods use training algorithms with previously professionally translated documents, but there’s a key difference. The statistical method limits translating a word within the context of one or two surrounding words.

On the other hand, neural network translations work based on a pattern-recognition process, the same as found in the brain of a multilingual individual. This leads to a smoother, more natural-sounding translation.

Despite all the efforts, the experience is far from perfect. For instance, some features are not available in all languages. However, the quality of the translation improves with continual use and experience, meaning it will only get better in time.

Tip: Centus offers a powerful localization suite based for developers, designers, translators and other roles that deal with translations on a daily basis. Their machine translation based service can help you manage your tasks seamlessly and in no time.

Last Updated on May 19, 2023 11:19 am CEST

SourceMicrosoft