Prime Video fixes bad dubbing with AI lip-sync

There is nothing that ruins a dramatic TV moment quite like watching an actor finish speaking while their mouth keeps moving for another three seconds.

Amazon is finally rolling out a high-tech fix. Prime Video just introduced an AI and visual effects-powered lip-sync dubbing feature aimed at completely eliminating the uncanny valley of foreign language dubbing.

Instead of replacing human actors with synthetic voices, Amazon is doing something different behind the scenes.

Fixing the faces, keeping the voices

The technology modifies the original actors' mouth movements using AI and VFX, seamlessly aligning their lips to match human-dubbed dialogue.

The feature made its debut on Maxton Hall, Prime Video's most-watched international original series to date. Seasons 1 and 2 are now streaming globally with English lip-sync dubs.

Amazon also confirmed the capability will be built in when the third and final season premieres on December 9. The upcoming season is based on Mona Kasten's 2018 novel Save Me.

Raf Soltanovich, vice president of technology at Prime Video and Amazon MGM Studios, noted that the process operates under creative oversight to preserve artistic integrity, adding that Amazon plans to expand the tech to more titles down the line.

Why it matters

We think Amazon found a remarkably smart middle ground for generative tech in Hollywood.

Instead of trying to replace human voice talent with fully synthesized AI audio, a flashpoint for labor unions and creative guilds, Amazon is using AI as an advanced VFX cleanup tool. Human voice actors still get the job, and viewers get rid of the jarring audio mismatch.

For viewers, this dramatically lowers the friction of watching international hits. Subtitles distract, and bad dubs repel; seamless mouth synchronization turns a German drama into something as effortless to watch as a domestic release.

However, altering an actor's facial performance retroactively does raise some questions. Even with creative oversight, digitally tweaking an actor's face across different languages blurs the line of authentic screen acting.

Still, as streaming platforms race to monetize non-English catalogs globally, expect this kind of subtle, performance-level AI localization to become the standard playback experience.