How did AI help Val Kilmer speak in Top Gun: Maverick?

Val Kilmer worked with Sonantic after throat-cancer treatment affected his speech, using archival recordings to build a voice model.

With Kilmer’s participation, Sonantic trained a voice model on archival recordings after throat-cancer treatment had affected his speech. The result supported his brief spoken dialogue—an example of AI used as an assistive creative tool rather than an unauthorized replacement.

Why consent makes this example different

Many people worry about AI voices because they imagine one copied without permission. Kilmer's case was the opposite: he took part, supplied recordings from his own career, and approved the project that let him deliver lines again. That cooperation matters. The technology rebuilt a familiar voice, but the choices about when and how to use it stayed with the actor and the filmmakers, which is why the scene plays as a return, not a replacement.

What to do, step by step

1. Start with recordings the person provides or approves, often drawn from films, interviews, and studio sessions.
2. Clean and sort that audio, so the model learns the voice rather than background noise.
3. Train the model on the samples, then generate the new lines in that learned voice.
4. Have the actor or family review the result, and adjust or reject takes that sound wrong.
5. Credit and disclose the tool in production, so audiences know how the performance was achieved.