A new feature of Google Translate, based on AI, will facilitate real-time transcription of speech.
At an artificial intelligence conference in San Francisco, Google announced it was working on a new feature for Google Translate that uses this technology.
With artificial intelligence, Google Translate will be able to transcribe in real time what it hears. For example, at a conference where the speaker addresses in German, Google Translate will write the Romanian version on the phone.
This feature, according to Google, is slightly more complex than regular text translation, so it can only be used when the phone is connected to the Internet. Some required components run on company servers.
Also, at least at the beginning, the new feature will not work with the recorded audio files, but only with the sound captured in real time. A strange limitation that can be easily fooled by playing the file through the device’s speaker.
Google explains how the algorithm constantly evaluates the translation, writing and modifying sentences as it hears new words. In real time and depending on what you hear, Google Translate will correct some words and add punctuation.
It is not known when exactly this feature will be launched, with Google avoiding offering even a rough date.