Listen to this article

Imagine speaking many different languages right now, without putting in any lengthy learning. And now, there is just such a technology in AI, from Verbalate™.

It is a simple interface where you select a video clip, select the language to translate to, and depending on how long your video is, it will be automatically done.


The technology is not perfect. I tried the “lip sync” format, and the resultant videos have artifacts around the lower jaw and lips for my test clip. But the “dub” format is quite usable, and I could select the languages I want to imagine I can speak in, and the platform will create a synthesized voice clone speaking from an AI translated (from my original English video) into the language I choose. So, here are some samples.

Original video (English)

Dubbed video (AI voice clone in Chinese)

Dubbed video (AI voice clone in Indonesian)

Dubbed video (AI voice clone in Korean)

I have to be honest, I have no idea how accurate the translations in Indonesian and Korean are, but the Chinese translation is roughly usable. The eerie thing is the voice clone, which did not require me to recite phrases or anything, and the platform’s AI simply took my one test clip in English, and created a synthetic voice that does sound like me, and can then “speak” in any of the languages I choose.

Uncanny, eerie, and does have lots of implications.