Microsoft announce public preview stage of Skype Translator
Microsoft’s Skype Translator, developed to bring down the language barrier is now into ‘preview’ stage. Skype Translator is a tool that translates your speech in real time into a language that can be understood by the person on the other side of the call. This is not just a translation of what you type in Skype chat. The translation is for your voice too. Whatever you speak, say in English will be translated on the fly and the other person will hear a voice in a different language. As of now, only English and Spanish language are added to the program. Lately, Microsoft has been making some serious research and bringing out applications/features that change the way we communicate and interact with world and this one definitely sits among the top of the list. If you are interested in trying this, go ahead and sign-up at this sign-up page. Do note that you will need Windows 8.1 on the desktop or device.
“This is just the beginning of a journey that will transform the way we communicate with people around the world. Our long-term goal for speech translation is to translate as many languages as possible on as many platforms as possible and deliver the best Skype Translator experience on each individual platform for our more than 300 million connected users.”, said, Gurdeep Pall, Corporate Vice President, Skype and Lync , Microsoft
How does Skype Translator work? Skype Translator was developed from decades of research in speech recognition, automated translation and machine learning technologies. By adding machine learning capability, Skype will continuously learn to recognize and translate speech in a better way, know to pick up different accents of same language and all the nuances. The machine learning protocols combines speech recognition and automatic machine translation into one optimized task. The optimization is done to such a level that sounds like umm, hmm, arrrr are removed, the text is divided into sentences and punctuation etc is added on the fly.
At Microsoft research, the training data was pulled from web pages, videos with captions, converstaions that are already translated and transcribed. All this data was used to create material for Speech Recognition and Machine Translation engines. The machine learning software then analyzes the data provided and create a statistical model of words in conversations and their context. The same technology is already in use, powering Bing Translator on the web.
Microsoft had a customized bot tht creates a call, sends audio streams to speec engine and get translation and transcription. For more detail on this, download this pdf as it contains in detail how Skype Translator works. You can also view the following YouTube video that explains “Behind the Scenes” aspects.