Filmmakers are using AI from companies like Deepdub to create foreign-language versions of their movies using samples of the original actors' voices
Ellen Gamerman / Wall Street Journal : Tweets: @mattocko and @stefaniei See also Mediagazer Tweets: Matt Ocko / @mattocko : As terrifying as this tech will be in the hands of criminals, & as much as we need way better attestation systems for our voices, photos, videos... ...it's still awesome that @valkilmer can act at the peak of his craft w/out worrying about his voice https://www.wsj.com/... Stefanie Ilgenfritz / @stefaniei : In the next dubbed film or TV show you watch, the dialogue might be AI-generated from samples of actors' voices that lets them ‘speak’ in new languages. The Future of Everything looks at the next evolution in sound technology. https://www.wsj.com/... via @EllenGamerman See also Mediagazer
Context & Ripple Effects
This story lands mid-way through a 2021 wave of synthetic-voice startups: MIT Technology Review had just profiled WellSaid Labs and VocaliD building custom AI voices for assistants and games, and Tel Aviv's Hour One was already paying people to license their likenesses for marketing videos. The WSJ piece extends that playbook to film dubbing, with Deepdub generating foreign-language dialogue from samples of the original actors' own voices.
The stakes go beyond localization convenience — the coverage cites Val Kilmer performing at his peak despite losing much of his natural voice — but it also plants a rights question that the corpus shows maturing over five years into open conflict between studios and performers.
First-order effects
- Studios gain a dubbing pipeline that preserves star voices across markets, while traditional dubbing voice actors lose those assignments on titles where the original performer's cloned voice is used instead.
- Actors whose voices are sampled can now deliver performances in languages they don't speak — but only if they've signed away or licensed the samples.
Second-order effects
- Performers respond commercially before legally: The Guardian's reporting on video game actors shows some negotiating extra payments on top of base minimums when their voices might be reused, a template film talent will press for.
- Dubbing vendors like Deepdub compete not just on audio quality but on consent workflows, since a single misuse scandal would poison the studio relationships their business depends on.
Third-order effects
- The endpoint visible in the corpus is voice actors worldwide mobilizing around personality rights as studios push AI dubbing at scale — making consent, compensation, and voice-likeness clauses standard contract terms rather than edge cases.
- If cloned voices proliferate, verification becomes an industry of its own; even early commentary on this piece flagged the need for far better attestation systems for voices, photos, and video alongside the criminal-misuse risk.
The trend: Performance is becoming a licensable data asset: actors' faces, likenesses, and now voices are being converted into licensed synthetic media, with compensation norms and personality-rights law racing to catch up.