Google wants to make life easier for YouTube video creators and will soon introduce AI dubbing. That’s why the company called on specialized service teams Aloudof its 120 division responsible for various types of pilot projects.
An overview video explains how everything works under the hood. The service first creates a copy of the text, which the user can edit and adapt. The AI then translates and reads the text to be easily grafted into the video. It is possible to get an idea of the outcome of this video from The Amoeba Sisters channel: simply select the track, dubbed in Spanish, by clicking on the cogwheel at the bottom right of the player.
If Aloud Early 2022, its integration into YouTube today is not surprising. For some time now, the platform has launched a function that allows you to present several audio tracks over a video. Thus, MrBeast videographer shows no less than 13 dubbings His latest videos, ranging from Japanese to French, passing through Arabic or Turkish. Currently, Aloud services are limited to English, Spanish and Portuguese, but other languages are expected to come later.
YouTube confirmed to the edge that the job is currently in testing with a hundred creators, and that it has the ambition to offer it to many more during 2024. In the long term, the goal is “ Make subtitled audio tracks sound like the creator’s voice, with more articulation and lip-syncing “.
This project is reminiscent of Google’s “Universal Translator”, which was presented at the group’s last major conference. This describes videos using artificial intelligence and convincingly renders a person’s lips in sync. However, this compiler should only be offered to a few partners: Google fears that the technology can be used to create DeepfakeWith the aim of making anyone say anything.