<- Google_ad_section_start ->
a breakthrough in automatic translation of conversations in real time, Microsoft reports in the blog. On the Re / code conference yesterday, Microsoft has demonstrated the progress made.The translation function is to be integrated into Skype and it goes to the end first time going with a corresponding Windows 8 Beta app. On other platforms, the new Skype functionality to be retrofitted later. After an introductory interview with Microsoft CEO Satya Nadella, the current state of the real-time translation was demonstrated live. The Pre-release a conversation took place with German-live translation. Microsoft acknowledges that you look for automatic real-time Übersetung still in its early stages, known from Star Trek universal translator is not Galaxy but far more removed. Until the “Skype Translator” is ready for market, it may take another 2-3 years. Presumably this will be a paid add-on service for Skype. A video of the demonstration, there is the Microsoft blog.
mercy! Certainly – somehow being able to communicate is better understood than no word. But you have to remember that for some time now (loosely over 20 years) is being put together machine translation please. Worth reading this right, this “timeline” of Microsoft Research.Since the computing power has grown huge, but the quality of machine translation has remained pathetic – human translators still far from starving. I’m talking mind of text-based translation! The quality of Microsoft’s “translation robot” I experience regularly in support posts that are only translated fully automated by Microsoft for years. These are “technical emotionless” texts that should be relatively easy to translate. What comes out is often surprisingly good, but too often hair-raising bad, slightly misleading. This applies to all automatic translation mechanisms, not only the Microsoft. Translating words usually works effortlessly, but the quality of the grammar is a gamble. On top of that languages - particularly technical language – now changing at a frantic pace and neologisms arise. I’m talking mind you still only of “text-based translation”. In the interview translation is also the fact that spoken words have to be recognized at all for now. Again, there is still a significant failure rate. I will not make Microsoft’s work on the “universal translator” bad – as a technology is exciting and great. Whether in 2-3 years a real usability is achieved, but I doubt it. For “emergency communications” it will probably (like now already) rich, but more unlikely.