YouTube's auto-dubbing now available for more creators in 9 languages
[10:11 Thu,12.December 2024 by blip]
Google has been working on auto-dubbing for YouTube videos since last summer www.slashcam.de/news/single/YouTube-integriert-automatisches-Audio-Dubbing-per-18061.html. Now the AI-based, automatic synchronization has been activated for hundreds of thousands of channels in the YouTube partner program. Primarily creators who produce informative or educational videos were included in this first major roll-out - but the feature will gradually be offered for other types of videos as well.
Automatic synchronization for YouTube videos
This means that from now on, you will see fewer subtitles and will instead come across more German-language audio tracks, for example. Nine languages are currently supported, including English, French, German, Hindi, Indonesian, Italian, Japanese, Portuguese and Spanish. If the original language of a video is English, a dubbed version is created for all of these languages, but only one, namely English. So if a video is published in German, it can also be heard in English.
Automatic synchronization is activated by default if a channel has the feature, but can be deactivated at any time in YouTube Studio on the desktop. Synchronized versions of your own videos can also be checked in advance.
On YouTube, videos with these automatically created audio tracks are marked with “Automatically synchronized” in the video description. Viewers can switch to the original audio track or to another available language at any time in the video settings.
The synchronized audio tracks are generated automatically - AI algorithms recognize and transcribe the original dialogues, translate them and output them again with a synthetic voice. This does not work equally well for all videos; errors can creep in due to pronunciation errors, accents, dialects or background noise in the original video. Problems can also occur when translating proper names, idioms and jargon. The voice used for the dubbing may also be offset from the original voice.
The dubbed tracks sound very good for the fact that they are artificially created, but the emphasis and inflection of the original voice is not properly taken into account. This should change in the future so that the alternative language versions are closer to the original version in terms of the listening experience (“expressive speech”).
;)
;)