Podcasters have another way to reach non-English speakers if they produce a video episode. YouTube is rolling out a new auto dubbing feature that will automatically detect the language of a video and create dubbed versions in other languages. The updates come by leveraging the artificial intelligence of Google DeepMind with Google Translate.
Nine languages are available at launch. If a video is in English, it will be dubbed into Spanish, French, German, Hindi, Indonesian, Italian, Japanese, and Portuguese. If a video is in any of those other languages, it will be dubbed into English. YouTube says that creators will have the option to review episodes before they are published to ensure they are happy with the results.
“It’s important to remember that this technology is still pretty new, and it won’t always be perfect,” the company says in a blog post. “We’re working hard to make it as accurate as possible, but there might be times when the translation isn’t quite right or the dubbed voice doesn’t accurately represent the original speaker.”
Viewers of the episodes will see an “auto-dubbed” label to distinguish when a video has auto-dubbed audio tracks. They will also be able to use the track selector to hear the video in its original language. And it will remember their language of choice for future videos too.
YouTube first announced the auto dubbing tool in September, saying it was working with a small group of creators to beta test the feature. Chief Product Officer Johanna Voolich said they are also piloting another capability that will transfer a creator’s tone and intonation, and the ambiance of surroundings and infuse it into the dubbed audio. She says the goal of what they are calling “expressive speech” is to make the experience sound “even more natural.”
Comments