YouTube wants to break down language barriers, thus extending the audience of video viewers: in a future that certainly the managers of the platform do not imagine too far away, artificial intelligence will double many clips in as many languages. This was announced yesterday by Google at VidCon, giving credit to the team of Aloud, product of the Area 120 incubator within the Mountain View company that is making available its know-how for the dubbing of YouTube videos based on AI.
It is a scenario of using artificial intelligence with the potential to get everyone to agree, skeptics included: no mass layoffs, it simply brings to the platform an opportunity (useful, no doubt) that was not there before and that probably, due to costs, would never have existed. Aloud’s AI automatically transcribes the audio of the original video, submits it to the creator who can edit the text in case of errors and once received the ok translates and doubles the video.
At the moment the languages supported by the Aloud + YouTube couple are not many (English, Spanish, Portuguese and Indonesian), but Google has committed to investing in the tool to introduce new ones soon. In Mountain View, they evidently believe a lot in Aloud’s technology, so much so that tests have been started with hundreds of creators to refine the mechanism and get to a large implementation of AI dubbing as soon as possible.
However, the dubbing of the AI is already available on YouTube, for the moment on a reduced number of videos.
Google, however, imagines a perfect system. At VidCon he confessed that there are people working to ensure that the AI samples the vocal timbre of the speaker to simulate their voice (a similar perspective for AI had already been talked about on several occasions) and make it more expressive, while in parallel they are trying to instruct artificial intelligence so that the “translated” voice is synchronized as much as possible with the movement of the lips of the speaker on video.
