Machine learning has come a long way in the last few years, and the new real-time translation software for Skype for Windows is tangible proof of this progress. This new feature on the Skype app makes it possible to have live conversations with people all across the world even if you do not speak their language.

Skype Machine Learning

In a blog post released Wednesday, January 13th, the Skype team emphasized that they developed this useful new feature because they were “dedicated to continue the growth of this new chapter in communication which has enabled people to connect globally, for free.”

Excerpt from Skype on new real-time translation feature

The blog post also noted: “Through voice-to-voice translation, all of our Skype for Windows customers can now speak in seven languages: Chinese Mandarin, English, French, German, Italian, Portuguese and Spanish. Through text-to-text translation, our customers can now write in over 50 instant message languages.”

Details on new Skype translation service

From now on, Skype users on Windows systems can start translating with a simple click on the globe icon in the upper right hand corner. If you do not see this globe icon, you probably need to update your Skype app to the latest version.

The full-scale roll out of the translation feature comes about three months Skype first introduced the translation service to Windows users back in October of last year. The Skype team first noted the feature at Re/code’s Code Conference in May of 2014, and a preview version of the service was available in December 2014. Of note, the voice translation system was only able to translate two languages in real-time at that point.

In its Wednesday post, Skype also noted that French-to-English translation has been the most frequently used language pair to date

More on machine learning

Machine learning technologies have been under development for some time, and are now finding their way into various consumer products and related applications. Languages are an ideal application for machine learning, as linguistic systems are tremendously complex and variable, but are near-finite (very slow change) and do function systematically. Given that machine learning algorithms learn what to do in the future based on their experiences in the past, the more conversations the new Skype system translates, the better it will become over time.